# HG changeset patch
# User Alex Williamson <alex.williamson@xxxxxx>
# Date 1209067709 21600
# Node ID dc510776dd598f3f479af749865bec225e32634d
# Parent 239b44eeb2d6d235ddee581b6e89398c80278a2f
# Parent 97da69831384f0819caeeb8b8bdff0f942b2d690
merge with xen-unstable.hg
---
xen/arch/x86/hvm/svm/x86_32/Makefile | 1
xen/arch/x86/hvm/svm/x86_32/exits.S | 131 -------
xen/arch/x86/hvm/svm/x86_64/Makefile | 1
xen/arch/x86/hvm/svm/x86_64/exits.S | 148 --------
xen/arch/x86/hvm/vmx/x86_32/Makefile | 1
xen/arch/x86/hvm/vmx/x86_32/exits.S | 148 --------
xen/arch/x86/hvm/vmx/x86_64/Makefile | 1
xen/arch/x86/hvm/vmx/x86_64/exits.S | 165 ---------
.hgignore | 1
Makefile | 7
docs/misc/vtd.txt | 2
docs/src/user.tex | 7
docs/xen-api/revision-history.tex | 10
docs/xen-api/xenapi-coversheet.tex | 2
docs/xen-api/xenapi-datamodel.tex | 180 ++++++++--
extras/mini-os/Makefile | 29 +
extras/mini-os/app.lds | 11
extras/mini-os/arch/ia64/minios-ia64.lds | 5
extras/mini-os/arch/x86/minios-x86_32.lds | 1
extras/mini-os/arch/x86/minios-x86_64.lds | 1
extras/mini-os/arch/x86/mm.c | 7
extras/mini-os/fbfront.c | 98 +++--
extras/mini-os/hypervisor.c | 15
extras/mini-os/include/fbfront.h | 3
extras/mini-os/include/hypervisor.h | 4
extras/mini-os/include/ia64/arch_mm.h | 2
extras/mini-os/include/ia64/os.h | 11
extras/mini-os/include/lib.h | 1
extras/mini-os/include/mm.h | 2
extras/mini-os/include/x86/os.h | 1
extras/mini-os/kernel.c | 15
extras/mini-os/lib/sys.c | 35 ++
extras/mini-os/main.c | 2
extras/mini-os/mm.c | 12
extras/mini-os/sched.c | 136 ++------
tools/blktap/drivers/blktapctrl.c | 83 ++++
tools/blktap/drivers/tapdisk.h | 2
tools/console/daemon/io.c | 18 -
tools/console/daemon/main.c | 13
tools/examples/blktap | 22 -
tools/firmware/hvmloader/Makefile | 5
tools/firmware/hvmloader/acpi/build.c | 4
tools/firmware/hvmloader/cacheattr.c | 99 +++++
tools/firmware/hvmloader/config.h | 3
tools/firmware/hvmloader/hvmloader.c | 3
tools/firmware/hvmloader/smp.c | 9
tools/ioemu/Makefile | 2
tools/ioemu/hw/cirrus_vga.c | 4
tools/ioemu/hw/pci.c | 16
tools/ioemu/hw/vga.c | 6
tools/ioemu/hw/xen_blktap.c | 45 --
tools/ioemu/hw/xenfb.c | 153 ++++++---
tools/ioemu/tapdisk-ioemu.c | 14
tools/ioemu/target-i386-dm/helper2.c | 2
tools/ioemu/vl.c | 17 -
tools/ioemu/vl.h | 1
tools/libfsimage/ext2fs/fsys_ext2fs.c | 61 +++
tools/libxc/xc_hvm_build.c | 2
tools/python/xen/util/acmpolicy.py | 27 +
tools/python/xen/xend/XendDomain.py | 26 +
tools/python/xen/xend/XendDomainInfo.py | 31 +
tools/python/xen/xend/XendXSPolicyAdmin.py | 1
tools/python/xen/xend/image.py | 46 ++
tools/python/xen/xm/main.py | 11
tools/tests/test_x86_emulator.c | 48 --
tools/tests/x86_emulate.c | 6
tools/xenmon/xenbaked.c | 32 +
xen/Makefile | 12
xen/arch/x86/Makefile | 2
xen/arch/x86/bitops.c | 32 +
xen/arch/x86/cpu/mtrr/main.c | 7
xen/arch/x86/domain.c | 11
xen/arch/x86/domain_build.c | 4
xen/arch/x86/hvm/emulate.c | 133 +++++++-
xen/arch/x86/hvm/hvm.c | 152 ++++++++-
xen/arch/x86/hvm/mtrr.c | 312 ++++--------------
xen/arch/x86/hvm/svm/Makefile | 4
xen/arch/x86/hvm/svm/entry.S | 178 ++++++++++
xen/arch/x86/hvm/svm/intr.c | 6
xen/arch/x86/hvm/svm/svm.c | 40 --
xen/arch/x86/hvm/vmx/Makefile | 4
xen/arch/x86/hvm/vmx/entry.S | 198 +++++++++++
xen/arch/x86/hvm/vmx/intr.c | 6
xen/arch/x86/hvm/vmx/vmx.c | 101 ------
xen/arch/x86/mm.c | 35 --
xen/arch/x86/mm/shadow/common.c | 94 ++---
xen/arch/x86/mm/shadow/multi.c | 13
xen/arch/x86/setup.c | 4
xen/arch/x86/smp.c | 27 -
xen/arch/x86/traps.c | 41 +-
xen/arch/x86/x86_emulate.c | 18 -
xen/arch/x86/x86_emulate/x86_emulate.c | 282 +++++++++++-----
xen/arch/x86/x86_emulate/x86_emulate.h | 54 +--
xen/common/trace.c | 11
xen/common/xencomm.c | 2
xen/drivers/char/console.c | 2
xen/drivers/char/serial.c | 53 ++-
xen/drivers/passthrough/amd/iommu_acpi.c | 214 +++++-------
xen/drivers/passthrough/amd/iommu_detect.c | 62 +++
xen/drivers/passthrough/amd/iommu_init.c | 306 ++++++++++++++++++
xen/drivers/passthrough/amd/iommu_map.c | 21 -
xen/drivers/passthrough/amd/pci_amd_iommu.c | 210 +++++-------
xen/drivers/passthrough/iommu.c | 30 +
xen/drivers/passthrough/vtd/dmar.c | 58 ---
xen/drivers/passthrough/vtd/iommu.c | 432 ++++++++------------------
xen/drivers/passthrough/vtd/utils.c | 2
xen/drivers/passthrough/vtd/x86/vtd.c | 16
xen/include/asm-x86/amd-iommu.h | 6
xen/include/asm-x86/bitops.h | 52 +--
xen/include/asm-x86/hvm/hvm.h | 2
xen/include/asm-x86/hvm/support.h | 2
xen/include/asm-x86/hvm/svm/amd-iommu-defs.h | 37 ++
xen/include/asm-x86/hvm/svm/amd-iommu-proto.h | 19 -
xen/include/asm-x86/hvm/vcpu.h | 11
xen/include/asm-x86/mtrr.h | 8
xen/include/asm-x86/paging.h | 2
xen/include/public/xsm/acm.h | 7
xen/include/xen/iommu.h | 4
xen/include/xen/serial.h | 8
xen/include/xsm/acm/acm_core.h | 1
xen/tools/Makefile | 4
xen/tools/figlet/figlet.c | 24 -
xen/xsm/acm/acm_policy.c | 8
123 files changed, 2967 insertions(+), 2425 deletions(-)
diff -r 239b44eeb2d6 -r dc510776dd59 .hgignore
--- a/.hgignore Thu Apr 24 14:02:16 2008 -0600
+++ b/.hgignore Thu Apr 24 14:08:29 2008 -0600
@@ -243,6 +243,7 @@
^tools/xm-test/lib/XmTestLib/config.py$
^tools/xm-test/lib/XmTestReport/xmtest.py$
^tools/xm-test/tests/.*\.test$
+^xen/\.banner.*$
^xen/BLOG$
^xen/System.map$
^xen/TAGS$
diff -r 239b44eeb2d6 -r dc510776dd59 Makefile
--- a/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -121,6 +121,13 @@ distclean:
# Linux name for GNU distclean
.PHONY: mrproper
mrproper: distclean
+
+# Prepare for source tarball
+.PHONY: src-tarball
+src-tarball: distclean
+ $(MAKE) -C xen .banner
+ rm -rf xen/tools/figlet .[a-z]*
+ $(MAKE) -C xen distclean
.PHONY: help
help:
diff -r 239b44eeb2d6 -r dc510776dd59 docs/misc/vtd.txt
--- a/docs/misc/vtd.txt Thu Apr 24 14:02:16 2008 -0600
+++ b/docs/misc/vtd.txt Thu Apr 24 14:08:29 2008 -0600
@@ -21,7 +21,7 @@ 11) "hide" pci device from dom0 as follo
title Xen-Fedora Core (2.6.18-xen)
root (hd0,0)
- kernel /boot/xen.gz com1=115200,8n1 console=com1 vtd=1
+ kernel /boot/xen.gz com1=115200,8n1 console=com1
module /boot/vmlinuz-2.6.18.8-xen root=LABEL=/ ro console=tty0
console=ttyS0,115200,8n1 pciback.hide=(01:00.0)(03:00.0)
pciback.verbose_request=1 apic=debug
module /boot/initrd-2.6.18-xen.img
diff -r 239b44eeb2d6 -r dc510776dd59 docs/src/user.tex
--- a/docs/src/user.tex Thu Apr 24 14:02:16 2008 -0600
+++ b/docs/src/user.tex Thu Apr 24 14:08:29 2008 -0600
@@ -4088,6 +4088,8 @@ editing \path{grub.conf}.
a list of pages not to be allocated for use because they contain bad
bytes. For example, if your memory tester says that byte 0x12345678
is bad, you would place `badpage=0x12345' on Xen's command line.
+\item [ serial\_tx\_buffer=$<$size$>$ ] Size of serial transmit
+ buffers. Default is 16kB.
\item [ com1=$<$baud$>$,DPS,$<$io\_base$>$,$<$irq$>$
com2=$<$baud$>$,DPS,$<$io\_base$>$,$<$irq$>$ ] \mbox{}\\
Xen supports up to two 16550-compatible serial ports. For example:
@@ -4239,10 +4241,11 @@ In addition to the standard Linux kernel
\begin{tabular}{l}
`xencons=off': disable virtual console \\
`xencons=tty': attach console to /dev/tty1 (tty0 at boot-time) \\
- `xencons=ttyS': attach console to /dev/ttyS0
+ `xencons=ttyS': attach console to /dev/ttyS0 \\
+ `xencons=xvc': attach console to /dev/xvc0
\end{tabular}
\end{center}
-The default is ttyS for dom0 and tty for all other domains.
+The default is ttyS for dom0 and xvc for all other domains.
\end{description}
diff -r 239b44eeb2d6 -r dc510776dd59 docs/xen-api/revision-history.tex
--- a/docs/xen-api/revision-history.tex Thu Apr 24 14:02:16 2008 -0600
+++ b/docs/xen-api/revision-history.tex Thu Apr 24 14:08:29 2008 -0600
@@ -37,5 +37,15 @@
\end{flushleft}
\end{minipage}\\
\hline
+ 1.0.5 & 17th Apr. 08 & S. Berger &
+ \begin{minipage}[t]{7cm}
+ \begin{flushleft}
+ Added undocumented fields and methods for default\_netmask and
+ default\_gateway to the Network class. Removed an unimplemented
+ method from the XSPolicy class and removed the 'optional' from
+ 'oldlabel' parameters.
+ \end{flushleft}
+ \end{minipage}\\
+ \hline
\end{tabular}
\end{center}
diff -r 239b44eeb2d6 -r dc510776dd59 docs/xen-api/xenapi-coversheet.tex
--- a/docs/xen-api/xenapi-coversheet.tex Thu Apr 24 14:02:16 2008 -0600
+++ b/docs/xen-api/xenapi-coversheet.tex Thu Apr 24 14:08:29 2008 -0600
@@ -22,7 +22,7 @@
\newcommand{\releasestatement}{Stable Release}
%% Document revision
-\newcommand{\revstring}{API Revision 1.0.4}
+\newcommand{\revstring}{API Revision 1.0.5}
%% Document authors
\newcommand{\docauthors}{
diff -r 239b44eeb2d6 -r dc510776dd59 docs/xen-api/xenapi-datamodel.tex
--- a/docs/xen-api/xenapi-datamodel.tex Thu Apr 24 14:02:16 2008 -0600
+++ b/docs/xen-api/xenapi-datamodel.tex Thu Apr 24 14:08:29 2008 -0600
@@ -4467,7 +4467,7 @@ security_label, string old_label)\end{ve
{\bf type} & {\bf name} & {\bf description} \\ \hline
{\tt VM ref } & self & reference to the object \\ \hline
{\tt string } & security\_label & security label for the VM \\ \hline
-{\tt string } & old\_label & Optional label value that the security label \\
+{\tt string } & old\_label & Label value that the security label \\
& & must currently have for the change to succeed.\\ \hline
\end{tabular}
@@ -7619,6 +7619,8 @@ Quals & Field & Type & Description \\
$\mathit{RW}$ & {\tt name/description} & string & a notes field containg
human-readable description \\
$\mathit{RO}_\mathit{run}$ & {\tt VIFs} & (VIF ref) Set & list of connected
vifs \\
$\mathit{RO}_\mathit{run}$ & {\tt PIFs} & (PIF ref) Set & list of connected
pifs \\
+$\mathit{RW}$ & {\tt default\_gateway} & string & default gateway \\
+$\mathit{RW}$ & {\tt default\_netmask} & string & default netmask \\
$\mathit{RW}$ & {\tt other\_config} & (string $\rightarrow$ string) Map &
additional configuration \\
\hline
\end{longtable}
@@ -7869,6 +7871,138 @@ Get the PIFs field of the given network.
value of the field
+\vspace{0.3cm}
+\vspace{0.3cm}
+\vspace{0.3cm}
+\subsubsection{RPC name:~get\_default\_gateway}
+
+{\bf Overview:}
+Get the default\_gateway field of the given network.
+
+ \noindent {\bf Signature:}
+\begin{verbatim} string get_default_gateway (session_id s, network ref
self)\end{verbatim}
+
+
+\noindent{\bf Arguments:}
+
+
+\vspace{0.3cm}
+\begin{tabular}{|c|c|p{7cm}|}
+ \hline
+{\bf type} & {\bf name} & {\bf description} \\ \hline
+{\tt network ref } & self & reference to the object \\ \hline
+
+\end{tabular}
+
+\vspace{0.3cm}
+
+ \noindent {\bf Return Type:}
+{\tt
+string
+}
+
+
+value of the field
+\vspace{0.3cm}
+\vspace{0.3cm}
+\vspace{0.3cm}
+\subsubsection{RPC name:~set\_default\_gateway}
+
+{\bf Overview:}
+Set the default\_gateway field of the given network.
+
+ \noindent {\bf Signature:}
+\begin{verbatim} void set_default_gateway (session_id s, network ref self,
string value)\end{verbatim}
+
+
+\noindent{\bf Arguments:}
+
+
+\vspace{0.3cm}
+\begin{tabular}{|c|c|p{7cm}|}
+ \hline
+{\bf type} & {\bf name} & {\bf description} \\ \hline
+{\tt network ref } & self & reference to the object \\ \hline
+
+{\tt string } & value & New value to set \\ \hline
+
+\end{tabular}
+
+\vspace{0.3cm}
+
+ \noindent {\bf Return Type:}
+{\tt
+void
+}
+
+
+
+\vspace{0.3cm}
+\vspace{0.3cm}
+\vspace{0.3cm}
+\subsubsection{RPC name:~get\_default\_netmask}
+
+{\bf Overview:}
+Get the default\_netmask field of the given network.
+
+ \noindent {\bf Signature:}
+\begin{verbatim} string get_default_netmask (session_id s, network ref
self)\end{verbatim}
+
+
+\noindent{\bf Arguments:}
+
+
+\vspace{0.3cm}
+\begin{tabular}{|c|c|p{7cm}|}
+ \hline
+{\bf type} & {\bf name} & {\bf description} \\ \hline
+{\tt network ref } & self & reference to the object \\ \hline
+
+\end{tabular}
+
+\vspace{0.3cm}
+
+ \noindent {\bf Return Type:}
+{\tt
+string
+}
+
+
+value of the field
+\vspace{0.3cm}
+\vspace{0.3cm}
+\vspace{0.3cm}
+\subsubsection{RPC name:~set\_default\_netmask}
+
+{\bf Overview:}
+Set the default\_netmask field of the given network.
+
+ \noindent {\bf Signature:}
+\begin{verbatim} void set_default_netmask (session_id s, network ref self,
string value)\end{verbatim}
+
+
+\noindent{\bf Arguments:}
+
+
+\vspace{0.3cm}
+\begin{tabular}{|c|c|p{7cm}|}
+ \hline
+{\bf type} & {\bf name} & {\bf description} \\ \hline
+{\tt network ref } & self & reference to the object \\ \hline
+
+{\tt string } & value & New value to set \\ \hline
+
+\end{tabular}
+
+\vspace{0.3cm}
+
+ \noindent {\bf Return Type:}
+{\tt
+void
+}
+
+
+
\vspace{0.3cm}
\vspace{0.3cm}
\vspace{0.3cm}
@@ -8999,7 +9133,7 @@ security_label, string old_label)\end{ve
{\tt VIF ref } & self & reference to the object \\ \hline
{\tt string } & security\_label & New value of the security label \\ \hline
-{\tt string } & old\_label & Optional label value that the security label \\
+{\tt string } & old\_label & Label value that the security label \\
& & must currently have for the change to succeed.\\ \hline
\end{tabular}
@@ -11504,7 +11638,7 @@ security_label, string old_label)\end{ve
{\tt VDI ref } & self & reference to the object \\ \hline
{\tt string } & security\_label & New value of the security label \\ \hline
-{\tt string } & old\_label & Optional label value that the security label \\
+{\tt string } & old\_label & Label value that the security label \\
& & must currently have for the change to succeed.\\ \hline
\end{tabular}
@@ -14898,46 +15032,6 @@ The label of the given resource.
\vspace{0.3cm}
\vspace{0.3cm}
\vspace{0.3cm}
-\subsubsection{RPC name:~activate\_xspolicy}
-
-{\bf Overview:}
-Load the referenced policy into the hypervisor.
-
- \noindent {\bf Signature:}
-\begin{verbatim} xs_instantiationflags activate_xspolicy (session_id s, xs_ref
xspolicy,
-xs_instantiationflags flags)\end{verbatim}
-
-
-\noindent{\bf Arguments:}
-
-
-\vspace{0.3cm}
-\begin{tabular}{|c|c|p{7cm}|}
- \hline
-{\bf type} & {\bf name} & {\bf description} \\ \hline
-{\tt xs ref } & self & reference to the object \\ \hline
-{\tt xs\_instantiationflags } & flags & flags to activate on a policy; flags
- can only be set \\ \hline
-
-\end{tabular}
-
-\vspace{0.3cm}
-
-
- \noindent {\bf Return Type:}
-{\tt
-xs\_instantiationflags
-}
-
-
-Currently active instantiation flags.
-\vspace{0.3cm}
-
-\noindent{\bf Possible Error Codes:} {\tt SECURITY\_ERROR}
-
-\vspace{0.3cm}
-\vspace{0.3cm}
-\vspace{0.3cm}
\subsubsection{RPC name:~can\_run}
{\bf Overview:}
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/Makefile
--- a/extras/mini-os/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -19,6 +19,7 @@ include minios.mk
# Define some default flags for linking.
LDLIBS :=
+APP_LDLIBS :=
LDARCHLIB := -L$(TARGET_ARCH_DIR) -l$(ARCH_LIB_NAME)
LDFLAGS_FINAL := -T $(TARGET_ARCH_DIR)/minios-$(XEN_TARGET_ARCH).lds
@@ -33,6 +34,7 @@ SUBDIRS := lib xenbus console
SUBDIRS := lib xenbus console
# The common mini-os objects to build.
+APP_OBJS :=
OBJS := $(patsubst %.c,%.o,$(wildcard *.c))
OBJS += $(patsubst %.c,%.o,$(wildcard lib/*.c))
OBJS += $(patsubst %.c,%.o,$(wildcard xenbus/*.c))
@@ -75,28 +77,28 @@ OBJS := $(filter-out lwip%.o $(LWO), $(O
ifeq ($(caml),y)
CAMLLIB = $(shell ocamlc -where)
-OBJS += $(CAMLDIR)/caml.o
-OBJS += $(CAMLLIB)/libasmrun.a
+APP_OBJS += main-caml.o
+APP_OBJS += $(CAMLDIR)/caml.o
+APP_OBJS += $(CAMLLIB)/libasmrun.a
CFLAGS += -I$(CAMLLIB)
-LDLIBS += -lm
-else
+APP_LDLIBS += -lm
+endif
OBJS := $(filter-out main-caml.o, $(OBJS))
-endif
ifeq ($(qemu),y)
-OBJS += $(QEMUDIR)/i386-dm-stubdom/qemu.a $(QEMUDIR)/i386-dm-stubdom/libqemu.a
+APP_OBJS += $(QEMUDIR)/i386-dm-stubdom/qemu.a
$(QEMUDIR)/i386-dm-stubdom/libqemu.a
CFLAGS += -DCONFIG_QEMU
endif
ifneq ($(CDIR),)
-OBJS += $(CDIR)/main.a
-LDLIBS +=
+APP_OBJS += $(CDIR)/main.a
+APP_LDLIBS +=
endif
ifeq ($(libc),y)
LDLIBS += -L$(XEN_ROOT)/stubdom/libxc -lxenctrl -lxenguest
-LDLIBS += -lpci
-LDLIBS += -lz
+APP_LDLIBS += -lpci
+APP_LDLIBS += -lz
LDLIBS += -lc
endif
@@ -104,8 +106,11 @@ OBJS := $(filter-out daytime.o, $(OBJS))
OBJS := $(filter-out daytime.o, $(OBJS))
endif
-$(TARGET): links $(OBJS) arch_lib
- $(LD) -r $(LDFLAGS) $(HEAD_OBJ) $(OBJS) $(LDARCHLIB) $(LDLIBS) -o $@.o
+app.o: $(APP_OBJS) app.lds
+ $(LD) -r -d $(LDFLAGS) $^ $(APP_LDLIBS) --undefined main -o $@
+
+$(TARGET): links $(OBJS) app.o arch_lib
+ $(LD) -r $(LDFLAGS) $(HEAD_OBJ) app.o $(OBJS) $(LDARCHLIB) $(LDLIBS) -o
$@.o
$(OBJCOPY) -w -G $(GLOBAL_PREFIX)* -G _start $@.o $@.o
$(LD) $(LDFLAGS) $(LDFLAGS_FINAL) $@.o $(EXTRA_OBJS) -o $@
gzip -f -9 -c $@ >$@.gz
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/app.lds
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/extras/mini-os/app.lds Thu Apr 24 14:08:29 2008 -0600
@@ -0,0 +1,11 @@
+SECTIONS
+{
+ .app.bss : {
+ __app_bss_start = . ;
+ *(.bss .bss.*)
+ *(COMMON)
+ *(.lbss .lbss.*)
+ *(LARGE_COMMON)
+ __app_bss_end = . ;
+ }
+}
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/arch/ia64/minios-ia64.lds
--- a/extras/mini-os/arch/ia64/minios-ia64.lds Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/arch/ia64/minios-ia64.lds Thu Apr 24 14:08:29 2008 -0600
@@ -59,7 +59,10 @@ SECTIONS
{ *(.IA_64.unwind) }
.bss : AT(ADDR(.bss) - (((5<<(61))+0x100000000) - (1 << 20)))
- { *(.bss) }
+ {
+ *(.bss)
+ *(.app.bss)
+ }
_end = .;
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/arch/x86/minios-x86_32.lds
--- a/extras/mini-os/arch/x86/minios-x86_32.lds Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/arch/x86/minios-x86_32.lds Thu Apr 24 14:08:29 2008 -0600
@@ -38,6 +38,7 @@ SECTIONS
__bss_start = .; /* BSS */
.bss : {
*(.bss)
+ *(.app.bss)
}
_end = . ;
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/arch/x86/minios-x86_64.lds
--- a/extras/mini-os/arch/x86/minios-x86_64.lds Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/arch/x86/minios-x86_64.lds Thu Apr 24 14:08:29 2008 -0600
@@ -38,6 +38,7 @@ SECTIONS
__bss_start = .; /* BSS */
.bss : {
*(.bss)
+ *(.app.bss)
}
_end = . ;
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/arch/x86/mm.c
--- a/extras/mini-os/arch/x86/mm.c Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/arch/x86/mm.c Thu Apr 24 14:08:29 2008 -0600
@@ -556,7 +556,6 @@ void *map_frames_ex(unsigned long *f, un
static void clear_bootstrap(void)
{
- struct xen_memory_reservation reservation;
xen_pfn_t mfns[] = { virt_to_mfn(&shared_info) };
int n = sizeof(mfns)/sizeof(*mfns);
pte_t nullpte = { };
@@ -567,11 +566,7 @@ static void clear_bootstrap(void)
if (HYPERVISOR_update_va_mapping((unsigned long) &_text, nullpte,
UVMF_INVLPG))
printk("Unable to unmap first page\n");
- set_xen_guest_handle(reservation.extent_start, mfns);
- reservation.nr_extents = n;
- reservation.extent_order = 0;
- reservation.domid = DOMID_SELF;
- if (HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation) != n)
+ if (free_physical_pages(mfns, n) != n)
printk("Unable to free bootstrap pages\n");
}
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/fbfront.c
--- a/extras/mini-os/fbfront.c Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/fbfront.c Thu Apr 24 14:08:29 2008 -0600
@@ -243,12 +243,12 @@ struct fbfront_dev {
char *backend;
int request_update;
- char *data;
int width;
int height;
int depth;
- int line_length;
+ int stride;
int mem_length;
+ int offset;
};
void fbfront_handler(evtchn_port_t port, struct pt_regs *regs, void *data)
@@ -256,7 +256,7 @@ void fbfront_handler(evtchn_port_t port,
wake_up(&fbfront_queue);
}
-struct fbfront_dev *init_fbfront(char *nodename, void *data, int width, int
height, int depth, int line_length, int mem_length)
+struct fbfront_dev *init_fbfront(char *nodename, unsigned long *mfns, int
width, int height, int depth, int stride, int n)
{
xenbus_transaction_t xbt;
char* err;
@@ -289,24 +289,17 @@ struct fbfront_dev *init_fbfront(char *n
dev->width = s->width = width;
dev->height = s->height = height;
dev->depth = s->depth = depth;
- dev->line_length = s->line_length = line_length;
- dev->mem_length = s->mem_length = mem_length;
-
- ASSERT(!((unsigned long)data & ~PAGE_MASK));
- dev->data = data;
+ dev->stride = s->line_length = stride;
+ dev->mem_length = s->mem_length = n * PAGE_SIZE;
+ dev->offset = 0;
const int max_pd = sizeof(s->pd) / sizeof(s->pd[0]);
unsigned long mapped = 0;
- for (i = 0; mapped < mem_length && i < max_pd; i++) {
+ for (i = 0; mapped < n && i < max_pd; i++) {
unsigned long *pd = (unsigned long *) alloc_page();
- for (j = 0; mapped < mem_length && j < PAGE_SIZE / sizeof(unsigned
long); j++) {
- /* Trigger CoW */
- * ((char *)data + mapped) = 0;
- barrier();
- pd[j] = virtual_to_mfn((unsigned long) data + mapped);
- mapped += PAGE_SIZE;
- }
+ for (j = 0; mapped < n && j < PAGE_SIZE / sizeof(unsigned long); j++)
+ pd[j] = mfns[mapped++];
for ( ; j < PAGE_SIZE / sizeof(unsigned long); j++)
pd[j] = 0;
s->pd[i] = virt_to_mfn(pd);
@@ -395,31 +388,11 @@ done:
return dev;
}
-void fbfront_update(struct fbfront_dev *dev, int x, int y, int width, int
height)
+static void fbfront_out_event(struct fbfront_dev *dev, union xenfb_out_event
*event)
{
struct xenfb_page *page = dev->page;
uint32_t prod;
DEFINE_WAIT(w);
-
- if (dev->request_update <= 0)
- return;
-
- if (x < 0) {
- width += x;
- x = 0;
- }
- if (x + width > dev->width)
- width = dev->width - x;
-
- if (y < 0) {
- height += y;
- y = 0;
- }
- if (y + height > dev->height)
- height = dev->height - y;
-
- if (width <= 0 || height <= 0)
- return;
add_waiter(w, fbfront_queue);
while (page->out_prod - page->out_cons == XENFB_OUT_RING_LEN)
@@ -428,14 +401,55 @@ void fbfront_update(struct fbfront_dev *
prod = page->out_prod;
mb(); /* ensure ring space available */
- XENFB_OUT_RING_REF(page, prod).type = XENFB_TYPE_UPDATE;
- XENFB_OUT_RING_REF(page, prod).update.x = x;
- XENFB_OUT_RING_REF(page, prod).update.y = y;
- XENFB_OUT_RING_REF(page, prod).update.width = width;
- XENFB_OUT_RING_REF(page, prod).update.height = height;
+ XENFB_OUT_RING_REF(page, prod) = *event;
wmb(); /* ensure ring contents visible */
page->out_prod = prod + 1;
notify_remote_via_evtchn(dev->evtchn);
+}
+
+void fbfront_update(struct fbfront_dev *dev, int x, int y, int width, int
height)
+{
+ struct xenfb_update update;
+
+ if (dev->request_update <= 0)
+ return;
+
+ if (x < 0) {
+ width += x;
+ x = 0;
+ }
+ if (x + width > dev->width)
+ width = dev->width - x;
+
+ if (y < 0) {
+ height += y;
+ y = 0;
+ }
+ if (y + height > dev->height)
+ height = dev->height - y;
+
+ if (width <= 0 || height <= 0)
+ return;
+
+ update.type = XENFB_TYPE_UPDATE;
+ update.x = x;
+ update.y = y;
+ update.width = width;
+ update.height = height;
+ fbfront_out_event(dev, (union xenfb_out_event *) &update);
+}
+
+void fbfront_resize(struct fbfront_dev *dev, int width, int height, int
stride, int depth, int offset)
+{
+ struct xenfb_resize resize;
+
+ resize.type = XENFB_TYPE_RESIZE;
+ dev->width = resize.width = width;
+ dev->height = resize.height = height;
+ dev->stride = resize.stride = stride;
+ dev->depth = resize.depth = depth;
+ dev->offset = resize.offset = offset;
+ fbfront_out_event(dev, (union xenfb_out_event *) &resize);
}
void shutdown_fbfront(struct fbfront_dev *dev)
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/hypervisor.c
--- a/extras/mini-os/hypervisor.c Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/hypervisor.c Thu Apr 24 14:08:29 2008 -0600
@@ -66,6 +66,21 @@ void do_hypervisor_callback(struct pt_re
in_callback = 0;
}
+void force_evtchn_callback(void)
+{
+ vcpu_info_t *vcpu;
+ vcpu = &HYPERVISOR_shared_info->vcpu_info[smp_processor_id()];
+ int save = vcpu->evtchn_upcall_mask;
+
+ while (vcpu->evtchn_upcall_pending) {
+ vcpu->evtchn_upcall_mask = 1;
+ barrier();
+ do_hypervisor_callback(NULL);
+ barrier();
+ vcpu->evtchn_upcall_mask = save;
+ barrier();
+ };
+}
inline void mask_evtchn(u32 port)
{
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/include/fbfront.h
--- a/extras/mini-os/include/fbfront.h Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/include/fbfront.h Thu Apr 24 14:08:29 2008 -0600
@@ -31,11 +31,12 @@ void shutdown_kbdfront(struct kbdfront_d
void shutdown_kbdfront(struct kbdfront_dev *dev);
-struct fbfront_dev *init_fbfront(char *nodename, void *data, int width, int
height, int depth, int line_length, int mem_length);
+struct fbfront_dev *init_fbfront(char *nodename, unsigned long *mfns, int
width, int height, int depth, int stride, int n);
#ifdef HAVE_LIBC
int fbfront_open(struct fbfront_dev *dev);
#endif
void fbfront_update(struct fbfront_dev *dev, int x, int y, int width, int
height);
+void fbfront_resize(struct fbfront_dev *dev, int width, int height, int
stride, int depth, int offset);
void shutdown_fbfront(struct fbfront_dev *dev);
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/include/hypervisor.h
--- a/extras/mini-os/include/hypervisor.h Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/include/hypervisor.h Thu Apr 24 14:08:29 2008 -0600
@@ -24,6 +24,7 @@
#else
#error "Unsupported architecture"
#endif
+#include <traps.h>
/*
* a placeholder for the start of day information passed up from the hypervisor
@@ -37,7 +38,8 @@ extern union start_info_union start_info
#define start_info (start_info_union.start_info)
/* hypervisor.c */
-//void do_hypervisor_callback(struct pt_regs *regs);
+void force_evtchn_callback(void);
+void do_hypervisor_callback(struct pt_regs *regs);
void mask_evtchn(u32 port);
void unmask_evtchn(u32 port);
void clear_evtchn(u32 port);
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/include/ia64/arch_mm.h
--- a/extras/mini-os/include/ia64/arch_mm.h Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/include/ia64/arch_mm.h Thu Apr 24 14:08:29 2008 -0600
@@ -38,6 +38,6 @@
#define map_frames(f, n) map_frames_ex(f, n, 1, 0, 1, DOMID_SELF, 0, 0)
/* TODO */
#define map_zero(n, a) map_frames_ex(NULL, n, 0, 0, a, DOMID_SELF, 0, 0)
-#define do_map_zero(start, n) ((void)0)
+#define do_map_zero(start, n) ASSERT(n == 0)
#endif /* __ARCH_MM_H__ */
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/include/ia64/os.h
--- a/extras/mini-os/include/ia64/os.h Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/include/ia64/os.h Thu Apr 24 14:08:29 2008 -0600
@@ -189,17 +189,6 @@ __synch_cmpxchg(volatile void *ptr, uint
return ia64_cmpxchg_acq_64(ptr, old, new);
}
-/*
- * Force a proper event-channel callback from Xen after clearing the
- * callback mask. We do this in a very simple manner, by making a call
- * down into Xen. The pending flag will be checked by Xen on return.
- */
-static inline void
-force_evtchn_callback(void)
-{
- (void)HYPERVISOR_xen_version(0, NULL);
-}
-
extern shared_info_t *HYPERVISOR_shared_info;
static inline int
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/include/lib.h
--- a/extras/mini-os/include/lib.h Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/include/lib.h Thu Apr 24 14:08:29 2008 -0600
@@ -187,6 +187,7 @@ int alloc_fd(enum fd_type type);
int alloc_fd(enum fd_type type);
void close_all_files(void);
extern struct thread *main_thread;
+void sparse(unsigned long data, size_t size);
#endif
#endif /* _LIB_H_ */
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/include/mm.h
--- a/extras/mini-os/include/mm.h Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/include/mm.h Thu Apr 24 14:08:29 2008 -0600
@@ -70,4 +70,6 @@ extern unsigned long heap, brk, heap_map
extern unsigned long heap, brk, heap_mapped, heap_end;
#endif
+int free_physical_pages(xen_pfn_t *mfns, int n);
+
#endif /* _MM_H_ */
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/include/x86/os.h
--- a/extras/mini-os/include/x86/os.h Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/include/x86/os.h Thu Apr 24 14:08:29 2008 -0600
@@ -28,7 +28,6 @@ extern void do_exit(void) __attribute__(
#include <xen/xen.h>
-#define force_evtchn_callback() ((void)HYPERVISOR_xen_version(0, 0))
#define __KERNEL_CS FLAT_KERNEL_CS
#define __KERNEL_DS FLAT_KERNEL_DS
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/kernel.c
--- a/extras/mini-os/kernel.c Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/kernel.c Thu Apr 24 14:08:29 2008 -0600
@@ -297,9 +297,20 @@ static void fbfront_thread(void *p)
{
size_t line_length = WIDTH * (DEPTH / 8);
size_t memsize = HEIGHT * line_length;
-
+ unsigned long *mfns;
+ int i, n = (memsize + PAGE_SIZE-1) / PAGE_SIZE;
+
+ memsize = n * PAGE_SIZE;
fb = _xmalloc(memsize, PAGE_SIZE);
- fb_dev = init_fbfront(NULL, fb, WIDTH, HEIGHT, DEPTH, line_length,
memsize);
+ mfns = xmalloc_array(unsigned long, n);
+ for (i = 0; i < n; i++) {
+ /* trigger CoW */
+ ((char *) fb) [i * PAGE_SIZE] = 0;
+ barrier();
+ mfns[i] = virtual_to_mfn((char *) fb + i * PAGE_SIZE);
+ }
+ fb_dev = init_fbfront(NULL, mfns, WIDTH, HEIGHT, DEPTH, line_length, n);
+ xfree(mfns);
if (!fb_dev) {
xfree(fb);
return;
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/lib/sys.c
--- a/extras/mini-os/lib/sys.c Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/lib/sys.c Thu Apr 24 14:08:29 2008 -0600
@@ -1108,6 +1108,41 @@ int munmap(void *start, size_t length)
return 0;
}
+void sparse(unsigned long data, size_t size)
+{
+ unsigned long newdata;
+ xen_pfn_t *mfns;
+ int i, n;
+
+ newdata = (data + PAGE_SIZE - 1) & PAGE_MASK;
+ if (newdata - data > size)
+ return;
+ size -= newdata - data;
+ data = newdata;
+ n = size / PAGE_SIZE;
+ size = n * PAGE_SIZE;
+
+ mfns = malloc(n * sizeof(*mfns));
+ for (i = 0; i < n; i++) {
+#ifdef LIBC_DEBUG
+ int j;
+ for (j=0; j<PAGE_SIZE; j++)
+ if (((char*)data + i * PAGE_SIZE)[j]) {
+ printk("%lx is not zero!\n", data + i * PAGE_SIZE + j);
+ exit(1);
+ }
+#endif
+ mfns[i] = virtual_to_mfn(data + i * PAGE_SIZE);
+ }
+
+ printk("sparsing %ldMB at %lx\n", size >> 20, data);
+
+ munmap((void *) data, size);
+ free_physical_pages(mfns, n);
+ do_map_zero(data, n);
+}
+
+
/* Not supported by FS yet. */
unsupported_function_crash(link);
unsupported_function(int, readlink, -1);
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/main.c
--- a/extras/mini-os/main.c Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/main.c Thu Apr 24 14:08:29 2008 -0600
@@ -39,6 +39,7 @@ void _fini(void)
{
}
+extern char __app_bss_start, __app_bss_end;
static void call_main(void *p)
{
char *args, /**path,*/ *msg, *c;
@@ -56,6 +57,7 @@ static void call_main(void *p)
* crashing. */
//sleep(1);
+ sparse((unsigned long) &__app_bss_start, &__app_bss_end -
&__app_bss_start);
start_networking();
init_fs_frontend();
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/mm.c
--- a/extras/mini-os/mm.c Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/mm.c Thu Apr 24 14:08:29 2008 -0600
@@ -36,6 +36,7 @@
#include <os.h>
#include <hypervisor.h>
+#include <xen/memory.h>
#include <mm.h>
#include <types.h>
#include <lib.h>
@@ -360,6 +361,17 @@ void free_pages(void *pointer, int order
}
+int free_physical_pages(xen_pfn_t *mfns, int n)
+{
+ struct xen_memory_reservation reservation;
+
+ set_xen_guest_handle(reservation.extent_start, mfns);
+ reservation.nr_extents = n;
+ reservation.extent_order = 0;
+ reservation.domid = DOMID_SELF;
+ return HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
+}
+
#ifdef HAVE_LIBC
void *sbrk(ptrdiff_t increment)
{
diff -r 239b44eeb2d6 -r dc510776dd59 extras/mini-os/sched.c
--- a/extras/mini-os/sched.c Thu Apr 24 14:02:16 2008 -0600
+++ b/extras/mini-os/sched.c Thu Apr 24 14:08:29 2008 -0600
@@ -70,62 +70,15 @@ void inline print_runqueue(void)
printk("\n");
}
-/* Find the time when the next timeout expires. If this is more than
- 10 seconds from now, return 10 seconds from now. */
-static s_time_t blocking_time(void)
-{
- struct thread *thread;
- struct list_head *iterator;
- s_time_t min_wakeup_time;
- unsigned long flags;
- local_irq_save(flags);
- /* default-block the domain for 10 seconds: */
- min_wakeup_time = NOW() + SECONDS(10);
-
- /* Thread list needs to be protected */
- list_for_each(iterator, &idle_thread->thread_list)
- {
- thread = list_entry(iterator, struct thread, thread_list);
- if(!is_runnable(thread) && thread->wakeup_time != 0LL)
- {
- if(thread->wakeup_time < min_wakeup_time)
- {
- min_wakeup_time = thread->wakeup_time;
- }
- }
- }
- local_irq_restore(flags);
- return(min_wakeup_time);
-}
-
-/* Wake up all threads with expired timeouts. */
-static void wake_expired(void)
-{
- struct thread *thread;
- struct list_head *iterator;
- s_time_t now = NOW();
- unsigned long flags;
- local_irq_save(flags);
- /* Thread list needs to be protected */
- list_for_each(iterator, &idle_thread->thread_list)
- {
- thread = list_entry(iterator, struct thread, thread_list);
- if(!is_runnable(thread) && thread->wakeup_time != 0LL)
- {
- if(thread->wakeup_time <= now)
- wake(thread);
- }
- }
- local_irq_restore(flags);
-}
-
void schedule(void)
{
struct thread *prev, *next, *thread;
struct list_head *iterator;
unsigned long flags;
+
prev = current;
local_irq_save(flags);
+
if (in_callback) {
printk("Must not call schedule() from a callback\n");
BUG();
@@ -134,6 +87,45 @@ void schedule(void)
printk("Must not call schedule() with IRQs disabled\n");
BUG();
}
+
+ do {
+ /* Examine all threads.
+ Find a runnable thread, but also wake up expired ones and find the
+ time when the next timeout expires, else use 10 seconds. */
+ s_time_t now = NOW();
+ s_time_t min_wakeup_time = now + SECONDS(10);
+ next = NULL;
+ list_for_each(iterator, &idle_thread->thread_list)
+ {
+ thread = list_entry(iterator, struct thread, thread_list);
+ if (!is_runnable(thread) && thread->wakeup_time != 0LL)
+ {
+ if (thread->wakeup_time <= now)
+ wake(thread);
+ else if (thread->wakeup_time < min_wakeup_time)
+ min_wakeup_time = thread->wakeup_time;
+ }
+ if(is_runnable(thread))
+ {
+ next = thread;
+ /* Put this thread on the end of the list */
+ list_del(&thread->thread_list);
+ list_add_tail(&thread->thread_list, &idle_thread->thread_list);
+ break;
+ }
+ }
+ if (next)
+ break;
+ /* block until the next timeout expires, or for 10 secs, whichever
comes first */
+ block_domain(min_wakeup_time);
+ /* handle pending events if any */
+ force_evtchn_callback();
+ } while(1);
+ local_irq_restore(flags);
+ /* Interrupting the switch is equivalent to having the next thread
+ inturrupted at the return instruction. And therefore at safe point. */
+ if(prev != next) switch_threads(prev, next);
+
list_for_each(iterator, &exited_threads)
{
thread = list_entry(iterator, struct thread, thread_list);
@@ -144,24 +136,6 @@ void schedule(void)
xfree(thread);
}
}
- next = idle_thread;
- /* Thread list needs to be protected */
- list_for_each(iterator, &idle_thread->thread_list)
- {
- thread = list_entry(iterator, struct thread, thread_list);
- if(is_runnable(thread))
- {
- next = thread;
- /* Put this thread on the end of the list */
- list_del(&thread->thread_list);
- list_add_tail(&thread->thread_list, &idle_thread->thread_list);
- break;
- }
- }
- local_irq_restore(flags);
- /* Interrupting the switch is equivalent to having the next thread
- inturrupted at the return instruction. And therefore at safe point. */
- if(prev != next) switch_threads(prev, next);
}
struct thread* create_thread(char *name, void (*function)(void *), void *data)
@@ -267,32 +241,10 @@ void wake(struct thread *thread)
void idle_thread_fn(void *unused)
{
- s_time_t until;
threads_started = 1;
- unsigned long flags;
- struct list_head *iterator;
- struct thread *next, *thread;
- for(;;)
- {
- schedule();
- next = NULL;
- local_irq_save(flags);
- list_for_each(iterator, &idle_thread->thread_list)
- {
- thread = list_entry(iterator, struct thread, thread_list);
- if(is_runnable(thread))
- {
- next = thread;
- break;
- }
- }
- if (!next) {
- /* block until the next timeout expires, or for 10 secs, whichever
comes first */
- until = blocking_time();
- block_domain(until);
- }
- local_irq_restore(flags);
- wake_expired();
+ while (1) {
+ block(current);
+ schedule();
}
}
diff -r 239b44eeb2d6 -r dc510776dd59 tools/blktap/drivers/blktapctrl.c
--- a/tools/blktap/drivers/blktapctrl.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/blktap/drivers/blktapctrl.c Thu Apr 24 14:08:29 2008 -0600
@@ -474,9 +474,8 @@ static int read_msg(int fd, int msgtype,
}
-int launch_tapdisk(char *wrctldev, char *rdctldev)
-{
- char *argv[] = { "tapdisk", wrctldev, rdctldev, NULL };
+static int launch_tapdisk_provider(char **argv)
+{
pid_t child;
if ((child = fork()) < 0)
@@ -490,7 +489,9 @@ int launch_tapdisk(char *wrctldev, char
i != STDERR_FILENO)
close(i);
- execvp("tapdisk", argv);
+ execvp(argv[0], argv);
+ DPRINTF("execvp failed: %d (%s)\n", errno, strerror(errno));
+ DPRINTF("PATH = %s\n", getenv("PATH"));
_exit(1);
} else {
pid_t got;
@@ -498,28 +499,78 @@ int launch_tapdisk(char *wrctldev, char
got = waitpid(child, NULL, 0);
} while (got != child);
}
+ return child;
+}
+
+static int launch_tapdisk(char *wrctldev, char *rdctldev)
+{
+ char *argv[] = { "tapdisk", wrctldev, rdctldev, NULL };
+
+ if (launch_tapdisk_provider(argv) < 0)
+ return -1;
+
return 0;
}
-/* Connect to qemu-dm */
-static int connect_qemu(blkif_t *blkif)
+static int launch_tapdisk_ioemu(void)
+{
+ char *argv[] = { "tapdisk-ioemu", NULL };
+ return launch_tapdisk_provider(argv);
+}
+
+/*
+ * Connect to an ioemu based disk provider (qemu-dm or tapdisk-ioemu)
+ *
+ * If the domain has a device model, connect to qemu-dm through the
+ * domain specific pipe. Otherwise use a single tapdisk-ioemu instance
+ * which is represented by domid 0 and provides access for Dom0 and
+ * all DomUs without device model.
+ */
+static int connect_qemu(blkif_t *blkif, int domid)
{
char *rdctldev, *wrctldev;
-
- if (asprintf(&rdctldev, BLKTAP_CTRL_DIR "/qemu-read-%d",
- blkif->domid) < 0)
- return -1;
-
- if (asprintf(&wrctldev, BLKTAP_CTRL_DIR "/qemu-write-%d",
- blkif->domid) < 0) {
+
+ static int tapdisk_ioemu_pid = 0;
+ static int dom0_readfd = 0;
+ static int dom0_writefd = 0;
+
+ if (asprintf(&rdctldev, BLKTAP_CTRL_DIR "/qemu-read-%d", domid) < 0)
+ return -1;
+
+ if (asprintf(&wrctldev, BLKTAP_CTRL_DIR "/qemu-write-%d", domid) < 0) {
free(rdctldev);
return -1;
}
DPRINTF("Using qemu blktap pipe: %s\n", rdctldev);
- blkif->fds[READ] = open_ctrl_socket(wrctldev);
- blkif->fds[WRITE] = open_ctrl_socket(rdctldev);
+ if (domid == 0) {
+ /*
+ * tapdisk-ioemu exits as soon as the last image is
+ * disconnected. Check if it is still running.
+ */
+ if (tapdisk_ioemu_pid == 0 || kill(tapdisk_ioemu_pid, 0)) {
+ /* No device model and tapdisk-ioemu doesn't run yet */
+ DPRINTF("Launching tapdisk-ioemu\n");
+ tapdisk_ioemu_pid = launch_tapdisk_ioemu();
+
+ dom0_readfd = open_ctrl_socket(wrctldev);
+ dom0_writefd = open_ctrl_socket(rdctldev);
+ }
+
+ DPRINTF("Using tapdisk-ioemu connection\n");
+ blkif->fds[READ] = dom0_readfd;
+ blkif->fds[WRITE] = dom0_writefd;
+ } else if (access(rdctldev, R_OK | W_OK) == 0) {
+ /* Use existing pipe to the device model */
+ DPRINTF("Using qemu-dm connection\n");
+ blkif->fds[READ] = open_ctrl_socket(wrctldev);
+ blkif->fds[WRITE] = open_ctrl_socket(rdctldev);
+ } else {
+ /* No device model => try with tapdisk-ioemu */
+ DPRINTF("No device model\n");
+ connect_qemu(blkif, 0);
+ }
free(rdctldev);
free(wrctldev);
@@ -599,7 +650,7 @@ int blktapctrl_new_blkif(blkif_t *blkif)
if (!exist) {
if (type == DISK_TYPE_IOEMU) {
- if (connect_qemu(blkif))
+ if (connect_qemu(blkif, blkif->domid))
goto fail;
} else {
if (connect_tapdisk(blkif, minor))
diff -r 239b44eeb2d6 -r dc510776dd59 tools/blktap/drivers/tapdisk.h
--- a/tools/blktap/drivers/tapdisk.h Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/blktap/drivers/tapdisk.h Thu Apr 24 14:08:29 2008 -0600
@@ -235,7 +235,7 @@ static disk_info_t ioemu_disk = {
DISK_TYPE_IOEMU,
"ioemu disk",
"ioemu",
- 0,
+ 1,
#ifdef TAPDISK
NULL
#endif
diff -r 239b44eeb2d6 -r dc510776dd59 tools/console/daemon/io.c
--- a/tools/console/daemon/io.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/console/daemon/io.c Thu Apr 24 14:08:29 2008 -0600
@@ -63,6 +63,7 @@ extern int log_time_hv;
extern int log_time_hv;
extern int log_time_guest;
extern char *log_dir;
+extern int discard_overflowed_data;
static int log_time_hv_needts = 1;
static int log_time_guest_needts = 1;
@@ -201,7 +202,7 @@ static void buffer_append(struct domain
dom->domid, errno, strerror(errno));
}
- if (buffer->max_capacity &&
+ if (discard_overflowed_data && buffer->max_capacity &&
buffer->size > buffer->max_capacity) {
/* Discard the middle of the data. */
@@ -228,6 +229,11 @@ static void buffer_advance(struct buffer
if (buffer->consumed == buffer->size) {
buffer->consumed = 0;
buffer->size = 0;
+ if (buffer->max_capacity &&
+ buffer->capacity > buffer->max_capacity) {
+ buffer->data = realloc(buffer->data,
buffer->max_capacity);
+ buffer->capacity = buffer->max_capacity;
+ }
}
}
@@ -1005,9 +1011,13 @@ void handle_io(void)
d->next_period < next_timeout)
next_timeout = d->next_period;
} else if (d->xce_handle != -1) {
- int evtchn_fd = xc_evtchn_fd(d->xce_handle);
- FD_SET(evtchn_fd, &readfds);
- max_fd = MAX(evtchn_fd, max_fd);
+ if (discard_overflowed_data ||
+ !d->buffer.max_capacity ||
+ d->buffer.size < d->buffer.max_capacity) {
+ int evtchn_fd =
xc_evtchn_fd(d->xce_handle);
+ FD_SET(evtchn_fd, &readfds);
+ max_fd = MAX(evtchn_fd, max_fd);
+ }
}
if (d->master_fd != -1) {
diff -r 239b44eeb2d6 -r dc510776dd59 tools/console/daemon/main.c
--- a/tools/console/daemon/main.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/console/daemon/main.c Thu Apr 24 14:08:29 2008 -0600
@@ -38,6 +38,7 @@ int log_time_hv = 0;
int log_time_hv = 0;
int log_time_guest = 0;
char *log_dir = NULL;
+int discard_overflowed_data = 1;
static void handle_hup(int sig)
{
@@ -46,7 +47,7 @@ static void handle_hup(int sig)
static void usage(char *name)
{
- printf("Usage: %s [-h] [-V] [-v] [-i] [--log=none|guest|hv|all]
[--log-dir=DIR] [--pid-file=PATH] [-t, --timestamp=none|guest|hv|all]\n", name);
+ printf("Usage: %s [-h] [-V] [-v] [-i] [--log=none|guest|hv|all]
[--log-dir=DIR] [--pid-file=PATH] [-t, --timestamp=none|guest|hv|all] [-o,
--overflow-data=discard|keep]\n", name);
}
static void version(char *name)
@@ -56,7 +57,7 @@ static void version(char *name)
int main(int argc, char **argv)
{
- const char *sopts = "hVvit:";
+ const char *sopts = "hVvit:o:";
struct option lopts[] = {
{ "help", 0, 0, 'h' },
{ "version", 0, 0, 'V' },
@@ -66,6 +67,7 @@ int main(int argc, char **argv)
{ "log-dir", 1, 0, 'r' },
{ "pid-file", 1, 0, 'p' },
{ "timestamp", 1, 0, 't' },
+ { "overflow-data", 1, 0, 'o'},
{ 0 },
};
bool is_interactive = false;
@@ -119,6 +121,13 @@ int main(int argc, char **argv)
log_time_hv = 0;
}
break;
+ case 'o':
+ if (!strcmp(optarg, "keep")) {
+ discard_overflowed_data = 0;
+ } else if (!strcmp(optarg, "discard")) {
+ discard_overflowed_data = 1;
+ }
+ break;
case '?':
fprintf(stderr,
"Try `%s --help' for more information\n",
diff -r 239b44eeb2d6 -r dc510776dd59 tools/examples/blktap
--- a/tools/examples/blktap Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/examples/blktap Thu Apr 24 14:08:29 2008 -0600
@@ -54,10 +54,6 @@ check_blktap_sharing()
echo 'ok'
}
-FRONTEND_ID=$(xenstore_read "$XENBUS_PATH/frontend-id")
-FRONTEND_UUID=$(xenstore_read "/local/domain/$FRONTEND_ID/vm")
-mode=$(xenstore_read "$XENBUS_PATH/mode")
-mode=$(canonicalise_mode "$mode")
t=$(xenstore_read_default "$XENBUS_PATH/type" 'MISSING')
if [ -n "$t" ]
@@ -77,15 +73,21 @@ else
file="$p"
fi
-if [ "$mode" != '!' ]
-then
- result=$(check_blktap_sharing "$file" "$mode")
- [ "$result" = 'ok' ] || ebusy "$file already in use by other domain"
-fi
-
if [ "$command" = 'add' ]
then
[ -e "$file" ] || { fatal $file does not exist; }
+
+ FRONTEND_ID=$(xenstore_read "$XENBUS_PATH/frontend-id")
+ FRONTEND_UUID=$(xenstore_read "/local/domain/$FRONTEND_ID/vm")
+ mode=$(xenstore_read "$XENBUS_PATH/mode")
+ mode=$(canonicalise_mode "$mode")
+
+ if [ "$mode" != '!' ]
+ then
+ result=$(check_blktap_sharing "$file" "$mode")
+ [ "$result" = 'ok' ] || ebusy "$file already in use by other domain"
+ fi
+
success
fi
diff -r 239b44eeb2d6 -r dc510776dd59 tools/firmware/hvmloader/Makefile
--- a/tools/firmware/hvmloader/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/firmware/hvmloader/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -28,8 +28,9 @@ LOADADDR = 0x100000
CFLAGS += $(CFLAGS_include) -I.
-SRCS = hvmloader.c mp_tables.c util.c smbios.c 32bitbios_support.c smp.c
-OBJS = $(patsubst %.c,%.o,$(SRCS))
+SRCS = hvmloader.c mp_tables.c util.c smbios.c
+SRCS += 32bitbios_support.c smp.c cacheattr.c
+OBJS = $(patsubst %.c,%.o,$(SRCS))
.PHONY: all
all: hvmloader
diff -r 239b44eeb2d6 -r dc510776dd59 tools/firmware/hvmloader/acpi/build.c
--- a/tools/firmware/hvmloader/acpi/build.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/firmware/hvmloader/acpi/build.c Thu Apr 24 14:08:29 2008 -0600
@@ -84,8 +84,8 @@ static int construct_bios_info_table(uin
bios_info->hpet_present = hpet_exists(ACPI_HPET_ADDRESS);
- bios_info->pci_min = 0xf0000000;
- bios_info->pci_len = 0x0c000000;
+ bios_info->pci_min = PCI_MEMBASE;
+ bios_info->pci_len = PCI_MEMSIZE;
return align16(sizeof(*bios_info));
}
diff -r 239b44eeb2d6 -r dc510776dd59 tools/firmware/hvmloader/cacheattr.c
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/tools/firmware/hvmloader/cacheattr.c Thu Apr 24 14:08:29 2008 -0600
@@ -0,0 +1,99 @@
+/*
+ * cacheattr.c: MTRR and PAT initialisation.
+ *
+ * Copyright (c) 2008, Citrix Systems, Inc.
+ *
+ * Authors:
+ * Keir Fraser <keir.fraser@xxxxxxxxxx>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
+ * Place - Suite 330, Boston, MA 02111-1307 USA.
+ */
+
+#include "util.h"
+#include "config.h"
+
+#define MSR_MTRRphysBase(reg) (0x200 + 2 * (reg))
+#define MSR_MTRRphysMask(reg) (0x200 + 2 * (reg) + 1)
+#define MSR_MTRRcap 0x00fe
+#define MSR_MTRRfix64K_00000 0x0250
+#define MSR_MTRRfix16K_80000 0x0258
+#define MSR_MTRRfix16K_A0000 0x0259
+#define MSR_MTRRfix4K_C0000 0x0268
+#define MSR_MTRRfix4K_C8000 0x0269
+#define MSR_MTRRfix4K_D0000 0x026a
+#define MSR_MTRRfix4K_D8000 0x026b
+#define MSR_MTRRfix4K_E0000 0x026c
+#define MSR_MTRRfix4K_E8000 0x026d
+#define MSR_MTRRfix4K_F0000 0x026e
+#define MSR_MTRRfix4K_F8000 0x026f
+#define MSR_PAT 0x0277
+#define MSR_MTRRdefType 0x02ff
+
+void cacheattr_init(void)
+{
+ uint32_t eax, ebx, ecx, edx;
+ uint64_t mtrr_cap, mtrr_def, content, addr_mask;
+ unsigned int i, nr_var_ranges, phys_bits = 36;
+
+ /* Does the CPU support architectural MTRRs? */
+ cpuid(0x00000001, &eax, &ebx, &ecx, &edx);
+ if ( !(edx & (1u << 12)) )
+ return;
+
+ /* Find the physical address size for this CPU. */
+ cpuid(0x80000000, &eax, &ebx, &ecx, &edx);
+ if ( eax >= 0x80000008 )
+ {
+ cpuid(0x80000008, &eax, &ebx, &ecx, &edx);
+ phys_bits = (uint8_t)eax;
+ }
+
+ printf("%u-bit phys ... ", phys_bits);
+
+ addr_mask = ((1ull << phys_bits) - 1) & ~((1ull << 12) - 1);
+ mtrr_cap = rdmsr(MSR_MTRRcap);
+ mtrr_def = (1u << 11) | 6; /* E, default type WB */
+
+ /* Fixed-range MTRRs supported? */
+ if ( mtrr_cap & (1u << 8) )
+ {
+ /* 0x00000-0x9ffff: Write Back (WB) */
+ content = 0x0606060606060606ull;
+ wrmsr(MSR_MTRRfix64K_00000, content);
+ wrmsr(MSR_MTRRfix16K_80000, content);
+ /* 0xa0000-0xbffff: Write Combining (WC) */
+ if ( mtrr_cap & (1u << 10) ) /* WC supported? */
+ content = 0x0101010101010101ull;
+ wrmsr(MSR_MTRRfix16K_A0000, content);
+ /* 0xc0000-0xfffff: Write Back (WB) */
+ content = 0x0606060606060606ull;
+ for ( i = 0; i < 8; i++ )
+ wrmsr(MSR_MTRRfix4K_C0000 + i, content);
+ mtrr_def |= 1u << 10; /* FE */
+ printf("fixed MTRRs ... ");
+ }
+
+ /* Variable-range MTRRs supported? */
+ nr_var_ranges = (uint8_t)mtrr_cap;
+ if ( nr_var_ranges != 0 )
+ {
+ /* A single UC range covering PCI space. */
+ wrmsr(MSR_MTRRphysBase(0), PCI_MEMBASE);
+ wrmsr(MSR_MTRRphysMask(0),
+ ((uint64_t)(int32_t)PCI_MEMBASE & addr_mask) | (1u << 11));
+ printf("var MTRRs ... ");
+ }
+
+ wrmsr(MSR_MTRRdefType, mtrr_def);
+}
diff -r 239b44eeb2d6 -r dc510776dd59 tools/firmware/hvmloader/config.h
--- a/tools/firmware/hvmloader/config.h Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/firmware/hvmloader/config.h Thu Apr 24 14:08:29 2008 -0600
@@ -10,6 +10,9 @@
#define PCI_ISA_DEVFN 0x08 /* dev 1, fn 0 */
#define PCI_ISA_IRQ_MASK 0x0c20U /* ISA IRQs 5,10,11 are PCI connected */
+
+#define PCI_MEMBASE 0xf0000000
+#define PCI_MEMSIZE 0x0c000000
#define ROMBIOS_SEG 0xF000
#define ROMBIOS_BEGIN 0x000F0000
diff -r 239b44eeb2d6 -r dc510776dd59 tools/firmware/hvmloader/hvmloader.c
--- a/tools/firmware/hvmloader/hvmloader.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/firmware/hvmloader/hvmloader.c Thu Apr 24 14:08:29 2008 -0600
@@ -96,6 +96,7 @@ asm (
"stack: \n"
" .skip 0x4000 \n"
"stack_top: \n"
+ " .text \n"
);
void smp_initialise(void);
@@ -158,7 +159,7 @@ static void pci_setup(void)
struct resource {
uint32_t base, max;
} *resource;
- struct resource mem_resource = { 0xf0000000, 0xfc000000 };
+ struct resource mem_resource = { PCI_MEMBASE, PCI_MEMBASE + PCI_MEMSIZE };
struct resource io_resource = { 0xc000, 0x10000 };
/* Create a list of device BARs in descending order of size. */
diff -r 239b44eeb2d6 -r dc510776dd59 tools/firmware/hvmloader/smp.c
--- a/tools/firmware/hvmloader/smp.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/firmware/hvmloader/smp.c Thu Apr 24 14:08:29 2008 -0600
@@ -66,12 +66,15 @@ asm (
"stack: \n"
" .skip 0x4000 \n"
"stack_top: \n"
+ " .text \n"
);
+
+extern void cacheattr_init(void);
/*static*/ void ap_start(void)
{
printf(" - CPU%d ... ", ap_cpuid);
-
+ cacheattr_init();
printf("done.\n");
wmb();
ap_callin = 1;
@@ -121,12 +124,10 @@ void smp_initialise(void)
{
unsigned int i, nr_cpus = get_vcpu_nr();
- if ( nr_cpus <= 1 )
- return;
-
memcpy((void *)AP_BOOT_EIP, ap_boot_start, ap_boot_end - ap_boot_start);
printf("Multiprocessor initialisation:\n");
+ ap_start();
for ( i = 1; i < nr_cpus; i++ )
boot_cpu(i);
}
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/Makefile
--- a/tools/ioemu/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -87,7 +87,7 @@ endif
install: all $(if $(BUILD_DOCS),install-doc)
mkdir -p "$(DESTDIR)$(bindir)"
- $(INSTALL) -m 755 -s $(TOOLS) "$(DESTDIR)$(prefix)/sbin"
+ $(INSTALL) -m 755 $(TOOLS) "$(DESTDIR)$(SBINDIR)"
# mkdir -p "$(DESTDIR)$(datadir)"
# for x in bios.bin vgabios.bin vgabios-cirrus.bin ppc_rom.bin \
# video.x openbios-sparc32 linux_boot.bin pxe-ne2k_pci.bin \
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/hw/cirrus_vga.c
--- a/tools/ioemu/hw/cirrus_vga.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/hw/cirrus_vga.c Thu Apr 24 14:08:29 2008 -0600
@@ -2595,6 +2595,10 @@ static void *set_vram_mapping(unsigned l
memset(vram_pointer, 0, nr_extents * TARGET_PAGE_SIZE);
+#ifdef CONFIG_STUBDOM
+ xenfb_pv_display_start(vram_pointer);
+#endif
+
free(extent_start);
return vram_pointer;
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/hw/pci.c
--- a/tools/ioemu/hw/pci.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/hw/pci.c Thu Apr 24 14:08:29 2008 -0600
@@ -79,18 +79,30 @@ int pci_bus_num(PCIBus *s)
void pci_device_save(PCIDevice *s, QEMUFile *f)
{
- qemu_put_be32(f, 1); /* PCI device version */
+ uint8_t irq_state = 0;
+ int i;
+ qemu_put_be32(f, 2); /* PCI device version */
qemu_put_buffer(f, s->config, 256);
+ for (i = 0; i < 4; i++)
+ irq_state |= !!s->irq_state[i] << i;
+ qemu_put_buffer(f, &irq_state, 1);
}
int pci_device_load(PCIDevice *s, QEMUFile *f)
{
uint32_t version_id;
version_id = qemu_get_be32(f);
- if (version_id != 1)
+ if (version_id != 1 && version_id != 2)
return -EINVAL;
qemu_get_buffer(f, s->config, 256);
pci_update_mappings(s);
+ if (version_id == 2) {
+ uint8_t irq_state;
+ int i;
+ qemu_get_buffer(f, &irq_state, 1);
+ for (i = 0; i < 4; i++)
+ pci_set_irq(s, i, !!(irq_state >> i));
+ }
return 0;
}
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/hw/vga.c
--- a/tools/ioemu/hw/vga.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/hw/vga.c Thu Apr 24 14:08:29 2008 -0600
@@ -2067,8 +2067,8 @@ void vga_common_init(VGAState *s, Displa
& ~(TARGET_PAGE_SIZE - 1));
/* Video RAM must be 128-bit aligned for SSE optimizations later */
- s->vram_alloc = qemu_malloc(vga_ram_size + 15);
- s->vram_ptr = (uint8_t *)((long)(s->vram_alloc + 15) & ~15L);
+ /* and page-aligned for PVFB memory sharing */
+ s->vram_ptr = s->vram_alloc = qemu_memalign(TARGET_PAGE_SIZE,
vga_ram_size);
s->vram_offset = vga_ram_offset;
s->vram_size = vga_ram_size;
@@ -2210,7 +2210,7 @@ void *vga_update_vram(VGAState *s, void
}
if (!vga_ram_base) {
- vga_ram_base = qemu_malloc(vga_ram_size + TARGET_PAGE_SIZE + 1);
+ vga_ram_base = qemu_memalign(TARGET_PAGE_SIZE, vga_ram_size +
TARGET_PAGE_SIZE + 1);
if (!vga_ram_base) {
fprintf(stderr, "reallocate error\n");
return NULL;
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/hw/xen_blktap.c
--- a/tools/ioemu/hw/xen_blktap.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/hw/xen_blktap.c Thu Apr 24 14:08:29 2008 -0600
@@ -581,17 +581,13 @@ static void handle_blktap_ctrlmsg(void*
*/
static int open_ctrl_socket(char *devname)
{
- int ret;
int ipc_fd;
if (mkdir(BLKTAP_CTRL_DIR, 0755) == 0)
DPRINTF("Created %s directory\n", BLKTAP_CTRL_DIR);
- ret = mkfifo(devname,S_IRWXU|S_IRWXG|S_IRWXO);
- if ( (ret != 0) && (errno != EEXIST) ) {
- DPRINTF("ERROR: pipe failed (%d)\n", errno);
+ if (access(devname, R_OK | W_OK))
return -1;
- }
ipc_fd = open(devname,O_RDWR|O_NONBLOCK);
@@ -601,42 +597,6 @@ static int open_ctrl_socket(char *devnam
}
return ipc_fd;
-}
-
-/**
- * Unmaps all disks and closes their pipes
- */
-void shutdown_blktap(void)
-{
- fd_list_entry_t *ptr;
- struct td_state *s;
- char *devname;
-
- DPRINTF("Shutdown blktap\n");
-
- /* Unmap all disks */
- ptr = fd_start;
- while (ptr != NULL) {
- s = ptr->s;
- unmap_disk(s);
- close(ptr->tap_fd);
- ptr = ptr->next;
- }
-
- /* Delete control pipes */
- if (asprintf(&devname, BLKTAP_CTRL_DIR "/qemu-read-%d", domid) >= 0) {
- DPRINTF("Delete %s\n", devname);
- if (unlink(devname))
- DPRINTF("Could not delete: %s\n", strerror(errno));
- free(devname);
- }
-
- if (asprintf(&devname, BLKTAP_CTRL_DIR "/qemu-write-%d", domid) >= 0) {
- DPRINTF("Delete %s\n", devname);
- if (unlink(devname))
- DPRINTF("Could not delete: %s\n", strerror(errno));
- free(devname);
- }
}
/**
@@ -679,8 +639,5 @@ int init_blktap(void)
/* Attach a handler to the read pipe (called from qemu main loop) */
qemu_set_fd_handler2(read_fd, NULL, &handle_blktap_ctrlmsg, NULL, NULL);
- /* Register handler to clean up when the domain is destroyed */
- atexit(&shutdown_blktap);
-
return 0;
}
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/hw/xenfb.c
--- a/tools/ioemu/hw/xenfb.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/hw/xenfb.c Thu Apr 24 14:08:29 2008 -0600
@@ -1235,14 +1235,10 @@ static struct semaphore kbd_sem = __SEMA
static struct semaphore kbd_sem = __SEMAPHORE_INITIALIZER(kbd_sem, 0);
static struct kbdfront_dev *kbd_dev;
static char *kbd_path, *fb_path;
+static void *vga_vram, *nonshared_vram;
+static DisplayState *xenfb_ds;
static unsigned char linux2scancode[KEY_MAX + 1];
-
-#define WIDTH 1024
-#define HEIGHT 768
-#define DEPTH 32
-#define LINESIZE (1280 * (DEPTH / 8))
-#define MEMSIZE (LINESIZE * HEIGHT)
int xenfb_connect_vkbd(const char *path)
{
@@ -1256,33 +1252,73 @@ int xenfb_connect_vfb(const char *path)
return 0;
}
-static void xenfb_pv_update(DisplayState *s, int x, int y, int w, int h)
-{
- struct fbfront_dev *fb_dev = s->opaque;
+static void xenfb_pv_update(DisplayState *ds, int x, int y, int w, int h)
+{
+ struct fbfront_dev *fb_dev = ds->opaque;
+ if (!fb_dev)
+ return;
fbfront_update(fb_dev, x, y, w, h);
}
-static void xenfb_pv_resize(DisplayState *s, int w, int h, int linesize)
-{
- struct fbfront_dev *fb_dev = s->opaque;
- fprintf(stderr,"resize to %dx%d required\n", w, h);
- s->width = w;
- s->height = h;
- /* TODO: send resize event if supported */
- memset(s->data, 0, MEMSIZE);
- fbfront_update(fb_dev, 0, 0, WIDTH, HEIGHT);
+static void xenfb_pv_resize(DisplayState *ds, int w, int h, int linesize)
+{
+ struct fbfront_dev *fb_dev = ds->opaque;
+ fprintf(stderr,"resize to %dx%d, %d required\n", w, h, linesize);
+ ds->width = w;
+ ds->height = h;
+ if (!linesize)
+ ds->shared_buf = 0;
+ if (!ds->shared_buf)
+ linesize = w * 4;
+ ds->linesize = linesize;
+ if (!fb_dev)
+ return;
+ if (ds->shared_buf) {
+ ds->data = NULL;
+ } else {
+ ds->data = nonshared_vram;
+ fbfront_resize(fb_dev, w, h, linesize, ds->depth, VGA_RAM_SIZE);
+ }
}
static void xenfb_pv_colourdepth(DisplayState *ds, int depth)
{
- /* TODO: send redepth event if supported */
+ struct fbfront_dev *fb_dev = ds->opaque;
static int lastdepth = -1;
+ if (!depth) {
+ ds->shared_buf = 0;
+ ds->depth = 32;
+ } else {
+ ds->shared_buf = 1;
+ ds->depth = depth;
+ }
if (depth != lastdepth) {
fprintf(stderr,"redepth to %d required\n", depth);
lastdepth = depth;
+ } else return;
+ if (!fb_dev)
+ return;
+ if (ds->shared_buf) {
+ ds->data = NULL;
+ } else {
+ ds->data = nonshared_vram;
+ fbfront_resize(fb_dev, ds->width, ds->height, ds->linesize, ds->depth,
VGA_RAM_SIZE);
}
- /* We can't redepth for now */
- ds->depth = DEPTH;
+}
+
+static void xenfb_pv_setdata(DisplayState *ds, void *pixels)
+{
+ struct fbfront_dev *fb_dev = ds->opaque;
+ int offset = pixels - vga_vram;
+ ds->data = pixels;
+ if (!fb_dev)
+ return;
+ fbfront_resize(fb_dev, ds->width, ds->height, ds->linesize, ds->depth,
offset);
+}
+
+static void xenfb_pv_refresh(DisplayState *ds)
+{
+ vga_hw_update();
}
static void xenfb_kbd_handler(void *opaque)
@@ -1373,13 +1409,6 @@ static void xenfb_kbd_handler(void *opaq
}
}
-static void xenfb_pv_refresh(DisplayState *ds)
-{
- /* always request negociation */
- ds->depth = -1;
- vga_hw_update();
-}
-
static void kbdfront_thread(void *p)
{
int scancode, keycode;
@@ -1399,40 +1428,72 @@ static void kbdfront_thread(void *p)
int xenfb_pv_display_init(DisplayState *ds)
{
- void *data;
+ if (!fb_path || !kbd_path)
+ return -1;
+
+ create_thread("kbdfront", kbdfront_thread, (void*) kbd_path);
+
+ xenfb_ds = ds;
+
+ ds->data = nonshared_vram = qemu_memalign(PAGE_SIZE, VGA_RAM_SIZE);
+ memset(ds->data, 0, VGA_RAM_SIZE);
+ ds->depth = 32;
+ ds->bgr = 0;
+ ds->width = 640;
+ ds->height = 400;
+ ds->linesize = 640 * 4;
+ ds->dpy_update = xenfb_pv_update;
+ ds->dpy_resize = xenfb_pv_resize;
+ ds->dpy_colourdepth = xenfb_pv_colourdepth;
+ ds->dpy_setdata = xenfb_pv_setdata;
+ ds->dpy_refresh = xenfb_pv_refresh;
+ return 0;
+}
+
+int xenfb_pv_display_start(void *data)
+{
+ DisplayState *ds = xenfb_ds;
struct fbfront_dev *fb_dev;
int kbd_fd;
+ int offset = 0;
+ unsigned long *mfns;
+ int n = VGA_RAM_SIZE / PAGE_SIZE;
+ int i;
if (!fb_path || !kbd_path)
- return -1;
-
- create_thread("kbdfront", kbdfront_thread, (void*) kbd_path);
-
- data = qemu_memalign(PAGE_SIZE, VGA_RAM_SIZE);
- fb_dev = init_fbfront(fb_path, data, WIDTH, HEIGHT, DEPTH, LINESIZE,
MEMSIZE);
+ return 0;
+
+ vga_vram = data;
+ mfns = malloc(2 * n * sizeof(*mfns));
+ for (i = 0; i < n; i++)
+ mfns[i] = virtual_to_mfn(vga_vram + i * PAGE_SIZE);
+ for (i = 0; i < n; i++)
+ mfns[n + i] = virtual_to_mfn(nonshared_vram + i * PAGE_SIZE);
+
+ fb_dev = init_fbfront(fb_path, mfns, ds->width, ds->height, ds->depth,
ds->linesize, 2 * n);
+ free(mfns);
if (!fb_dev) {
fprintf(stderr,"can't open frame buffer\n");
exit(1);
}
free(fb_path);
+ if (ds->shared_buf) {
+ offset = (void*) ds->data - vga_vram;
+ } else {
+ offset = VGA_RAM_SIZE;
+ ds->data = nonshared_vram;
+ }
+ if (offset)
+ fbfront_resize(fb_dev, ds->width, ds->height, ds->linesize, ds->depth,
offset);
+
down(&kbd_sem);
free(kbd_path);
kbd_fd = kbdfront_open(kbd_dev);
qemu_set_fd_handler(kbd_fd, xenfb_kbd_handler, NULL, ds);
- ds->data = data;
- ds->linesize = LINESIZE;
- ds->depth = DEPTH;
- ds->bgr = 0;
- ds->width = WIDTH;
- ds->height = HEIGHT;
- ds->dpy_update = xenfb_pv_update;
- ds->dpy_resize = xenfb_pv_resize;
- ds->dpy_colourdepth = xenfb_pv_colourdepth;
- ds->dpy_refresh = xenfb_pv_refresh;
- ds->opaque = fb_dev;
+ xenfb_ds->opaque = fb_dev;
return 0;
}
#endif
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/tapdisk-ioemu.c
--- a/tools/ioemu/tapdisk-ioemu.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/tapdisk-ioemu.c Thu Apr 24 14:08:29 2008 -0600
@@ -4,6 +4,7 @@
#include <string.h>
#include <stdint.h>
#include <signal.h>
+#include <unistd.h>
#include <sys/time.h>
#include <assert.h>
@@ -15,6 +16,8 @@ extern void bdrv_init(void);
extern void *qemu_mallocz(size_t size);
extern void qemu_free(void *ptr);
+
+extern void *fd_start;
int domid = 0;
FILE* logfile;
@@ -95,12 +98,17 @@ int main(void)
int max_fd;
fd_set rfds;
struct timeval tv;
+ void *old_fd_start = NULL;
logfile = stderr;
bdrv_init();
qemu_aio_init();
init_blktap();
+
+ /* Daemonize */
+ if (fork() != 0)
+ exit(0);
/*
* Main loop: Pass events to the corrsponding handlers and check for
@@ -137,6 +145,12 @@ int main(void)
} else
pioh = &ioh->next;
}
+
+ /* Exit when the last image has been closed */
+ if (old_fd_start != NULL && fd_start == NULL)
+ exit(0);
+
+ old_fd_start = fd_start;
}
return 0;
}
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/target-i386-dm/helper2.c
--- a/tools/ioemu/target-i386-dm/helper2.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/target-i386-dm/helper2.c Thu Apr 24 14:08:29 2008 -0600
@@ -482,7 +482,7 @@ void cpu_handle_ioreq(void *opaque)
CPUState *env = opaque;
ioreq_t *req = cpu_get_ioreq();
- handle_buffered_io(env);
+ __handle_buffered_iopage(env);
if (req) {
__handle_ioreq(env, req);
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/vl.c
--- a/tools/ioemu/vl.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/vl.c Thu Apr 24 14:08:29 2008 -0600
@@ -140,9 +140,9 @@
#define MAX_IOPORTS 65536
const char *bios_dir = CONFIG_QEMU_SHAREDIR;
-void **ioport_opaque;
-IOPortReadFunc *(*ioport_read_table)[MAX_IOPORTS];
-IOPortWriteFunc *(*ioport_write_table)[MAX_IOPORTS];
+void *ioport_opaque[MAX_IOPORTS];
+IOPortReadFunc *ioport_read_table[3][MAX_IOPORTS];
+IOPortWriteFunc *ioport_write_table[3][MAX_IOPORTS];
/* Note: bs_table[MAX_DISKS] is a dummy block driver if none available
to store the VM snapshots */
BlockDriverState *bs_table[MAX_DISKS + MAX_SCSI_DISKS + 1], *fd_table[MAX_FD];
@@ -281,9 +281,6 @@ void default_ioport_writel(void *opaque,
void init_ioports(void)
{
- ioport_opaque = calloc(MAX_IOPORTS, sizeof(*ioport_opaque));
- ioport_read_table = calloc(3 * MAX_IOPORTS, sizeof(**ioport_read_table));
- ioport_write_table = calloc(3 * MAX_IOPORTS, sizeof(**ioport_write_table));
}
/* size is the word size in byte */
@@ -6276,12 +6273,6 @@ void qemu_system_powerdown_request(void)
powerdown_requested = 1;
if (cpu_single_env)
cpu_interrupt(cpu_single_env, CPU_INTERRUPT_EXIT);
-}
-
-static void qemu_sighup_handler(int signal)
-{
- fprintf(stderr, "Received SIGHUP, terminating.\n");
- exit(0);
}
void main_loop_wait(int timeout)
@@ -7979,7 +7970,7 @@ int main(int argc, char **argv)
#ifndef CONFIG_STUBDOM
/* Unblock SIGTERM and SIGHUP, which may have been blocked by the caller */
- signal(SIGHUP, qemu_sighup_handler);
+ signal(SIGHUP, SIG_DFL);
sigemptyset(&set);
sigaddset(&set, SIGTERM);
sigaddset(&set, SIGHUP);
diff -r 239b44eeb2d6 -r dc510776dd59 tools/ioemu/vl.h
--- a/tools/ioemu/vl.h Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/ioemu/vl.h Thu Apr 24 14:08:29 2008 -0600
@@ -1545,6 +1545,7 @@ char *xenstore_vm_read(int domid, char *
/* xenfb.c */
int xenfb_pv_display_init(DisplayState *ds);
+int xenfb_pv_display_start(void *vram_start);
int xenfb_connect_vkbd(const char *path);
int xenfb_connect_vfb(const char *path);
diff -r 239b44eeb2d6 -r dc510776dd59 tools/libfsimage/ext2fs/fsys_ext2fs.c
--- a/tools/libfsimage/ext2fs/fsys_ext2fs.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/libfsimage/ext2fs/fsys_ext2fs.c Thu Apr 24 14:08:29 2008 -0600
@@ -77,7 +77,52 @@ struct ext2_super_block
__u32 s_rev_level; /* Revision level */
__u16 s_def_resuid; /* Default uid for reserved blocks */
__u16 s_def_resgid; /* Default gid for reserved blocks */
- __u32 s_reserved[235]; /* Padding to the end of the block */
+ /*
+ * These fields are for EXT2_DYNAMIC_REV superblocks only.
+ *
+ * Note: the difference between the compatible feature set and
+ * the incompatible feature set is that if there is a bit set
+ * in the incompatible feature set that the kernel doesn't
+ * know about, it should refuse to mount the filesystem.
+ *
+ * e2fsck's requirements are more strict; if it doesn't know
+ * about a feature in either the compatible or incompatible
+ * feature set, it must abort and not try to meddle with
+ * things it doesn't understand...
+ */
+ __u32 s_first_ino; /* First non-reserved inode */
+ __u16 s_inode_size; /* size of inode structure */
+ __u16 s_block_group_nr; /* block group # of this superblock */
+ __u32 s_feature_compat; /* compatible feature set */
+ __u32 s_feature_incompat; /* incompatible feature set */
+ __u32 s_feature_ro_compat; /* readonly-compatible feature set */
+ __u8 s_uuid[16]; /* 128-bit uuid for volume */
+ char s_volume_name[16]; /* volume name */
+ char s_last_mounted[64]; /* directory where last mounted */
+ __u32 s_algorithm_usage_bitmap; /* For compression */
+ /*
+ * Performance hints. Directory preallocation should only
+ * happen if the EXT2_FEATURE_COMPAT_DIR_PREALLOC flag is on.
+ */
+ __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
+ __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
+ __u16 s_reserved_gdt_blocks;/* Per group table for online growth */
+ /*
+ * Journaling support valid if EXT2_FEATURE_COMPAT_HAS_JOURNAL set.
+ */
+ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
+ __u32 s_journal_inum; /* inode number of journal file */
+ __u32 s_journal_dev; /* device number of journal file */
+ __u32 s_last_orphan; /* start of list of inodes to delete */
+ __u32 s_hash_seed[4]; /* HTREE hash seed */
+ __u8 s_def_hash_version; /* Default hash version to use */
+ __u8 s_jnl_backup_type; /* Default type of journal backup */
+ __u16 s_reserved_word_pad;
+ __u32 s_default_mount_opts;
+ __u32 s_first_meta_bg; /* First metablock group */
+ __u32 s_mkfs_time; /* When the filesystem was created */
+ __u32 s_jnl_blocks[17]; /* Backup of the journal inode */
+ __u32 s_reserved[172]; /* Padding to the end of the block */
};
struct ext2_group_desc
@@ -216,6 +261,9 @@ struct ext2_dir_entry
#define EXT2_ADDR_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s) / sizeof (__u32))
#define EXT2_ADDR_PER_BLOCK_BITS(s) (log2(EXT2_ADDR_PER_BLOCK(s)))
+#define EXT2_INODE_SIZE(s) (SUPERBLOCK->s_inode_size)
+#define EXT2_INODES_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s)/EXT2_INODE_SIZE(s))
+
/* linux/ext2_fs.h */
#define EXT2_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
/* kind of from ext2/super.c */
@@ -537,7 +585,7 @@ ext2fs_dir (fsi_file_t *ffi, char *dirna
gdp = GROUP_DESC;
ino_blk = gdp[desc].bg_inode_table +
(((current_ino - 1) % (SUPERBLOCK->s_inodes_per_group))
- >> log2 (EXT2_BLOCK_SIZE (SUPERBLOCK) / sizeof (struct ext2_inode)));
+ >> log2 (EXT2_INODES_PER_BLOCK (SUPERBLOCK)));
#ifdef E2DEBUG
printf ("inode table fsblock=%d\n", ino_blk);
#endif /* E2DEBUG */
@@ -549,13 +597,12 @@ ext2fs_dir (fsi_file_t *ffi, char *dirna
/* reset indirect blocks! */
mapblock2 = mapblock1 = -1;
- raw_inode = INODE +
- ((current_ino - 1)
- & (EXT2_BLOCK_SIZE (SUPERBLOCK) / sizeof (struct ext2_inode) - 1));
+ raw_inode = (struct ext2_inode *)((char *)INODE +
+ ((current_ino - 1) & (EXT2_INODES_PER_BLOCK (SUPERBLOCK) - 1)) *
+ EXT2_INODE_SIZE (SUPERBLOCK));
#ifdef E2DEBUG
printf ("ipb=%d, sizeof(inode)=%d\n",
- (EXT2_BLOCK_SIZE (SUPERBLOCK) / sizeof (struct ext2_inode)),
- sizeof (struct ext2_inode));
+ EXT2_INODES_PER_BLOCK (SUPERBLOCK), EXT2_INODE_SIZE (SUPERBLOCK));
printf ("inode=%x, raw_inode=%x\n", INODE, raw_inode);
printf ("offset into inode table block=%d\n", (int) raw_inode - (int)
INODE);
for (i = (unsigned char *) INODE; i <= (unsigned char *) raw_inode;
diff -r 239b44eeb2d6 -r dc510776dd59 tools/libxc/xc_hvm_build.c
--- a/tools/libxc/xc_hvm_build.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/libxc/xc_hvm_build.c Thu Apr 24 14:08:29 2008 -0600
@@ -298,7 +298,7 @@ static int setup_guest(int xc_handle,
_PAGE_ACCESSED | _PAGE_DIRTY | _PAGE_PSE);
munmap(ident_pt, PAGE_SIZE);
xc_set_hvm_param(xc_handle, dom, HVM_PARAM_IDENT_PT,
- special_page_nr + SPECIALPAGE_IDENT_PT);
+ (special_page_nr + SPECIALPAGE_IDENT_PT) << PAGE_SHIFT);
/* Insert JMP <rel32> instruction at address 0x0 to reach entry point. */
entry_eip = elf_uval(&elf, elf.ehdr, e_entry);
diff -r 239b44eeb2d6 -r dc510776dd59 tools/python/xen/util/acmpolicy.py
--- a/tools/python/xen/util/acmpolicy.py Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/python/xen/util/acmpolicy.py Thu Apr 24 14:08:29 2008 -0600
@@ -17,6 +17,7 @@
#============================================================================
import os
+import sha
import stat
import array
import struct
@@ -35,7 +36,7 @@ ACM_POLICIES_DIR = security.policy_dir_p
# Constants needed for generating a binary policy from its XML
# representation
-ACM_POLICY_VERSION = 3 # Latest one
+ACM_POLICY_VERSION = 4 # Latest one
ACM_CHWALL_VERSION = 1
ACM_STE_VERSION = 1
@@ -965,6 +966,10 @@ class ACMPolicy(XSPolicy):
return dom.toxml()
return None
+ def hash(self):
+ """ Calculate a SAH1 hash of the XML policy """
+ return sha.sha(self.toxml())
+
def save(self):
### Save the XML policy into a file ###
rc = -xsconstants.XSERR_FILE_ERROR
@@ -1403,7 +1408,7 @@ class ACMPolicy(XSPolicy):
ste_bin += "\x00"
#Write binary header:
- headerformat="!iiiiiiiiii"
+ headerformat="!iiiiiiiiii20s"
totallen_bin = struct.calcsize(headerformat) + \
len(pr_bin) + len(chw_bin) + len(ste_bin)
polref_offset = struct.calcsize(headerformat)
@@ -1425,7 +1430,8 @@ class ACMPolicy(XSPolicy):
primpoloffset,
secpolcode,
secpoloffset,
- major, minor)
+ major, minor,
+ self.hash().digest())
all_bin = array.array('B')
for s in [ hdr_bin, pr_bin, chw_bin, ste_bin ]:
@@ -1443,6 +1449,21 @@ class ACMPolicy(XSPolicy):
rc = -xsconstants.XSERR_BAD_LABEL
return rc, mapfile, all_bin.tostring()
+ def validate_enforced_policy_hash(self):
+ """ verify that the policy hash embedded in the binary policy
+ that is currently enforce matches the one of the XML policy.
+ """
+ if self.hash().digest() != self.get_enforced_policy_hash():
+ raise Exception('Policy hashes do not match')
+
+ def get_enforced_policy_hash(self):
+ binpol = self.get_enforced_binary()
+ headerformat="!iiiiiiiiii20s"
+ res = struct.unpack(headerformat, binpol[:60])
+ if len(res) >= 11:
+ return res[10]
+ return None
+
def get_enforced_binary(self):
rc, binpol = security.hv_get_policy()
if rc != 0:
diff -r 239b44eeb2d6 -r dc510776dd59 tools/python/xen/xend/XendDomain.py
--- a/tools/python/xen/xend/XendDomain.py Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/python/xen/xend/XendDomain.py Thu Apr 24 14:08:29 2008 -0600
@@ -1622,7 +1622,31 @@ class XendDomain:
vcpu)
except Exception, ex:
raise XendError(str(ex))
-
+
+ def domain_reset(self, domid):
+ """Terminate domain immediately, and then create domain.
+
+ @param domid: Domain ID or Name
+ @type domid: int or string.
+ @rtype: None
+ @raise XendError: Failed to destroy or create
+ @raise XendInvalidDomain: Domain is not valid
+ """
+
+ dominfo = self.domain_lookup_nr(domid)
+ if not dominfo:
+ raise XendInvalidDomain(str(domid))
+ if dominfo and dominfo.getDomid() == DOM0_ID:
+ raise XendError("Cannot reset privileged domain %s" % domid)
+ if dominfo._stateGet() not in (DOM_STATE_RUNNING, DOM_STATE_PAUSED):
+ raise VMBadState("Domain '%s' is not started" % domid,
+ POWER_STATE_NAMES[DOM_STATE_RUNNING],
+ POWER_STATE_NAMES[dominfo._stateGet()])
+ try:
+ dominfo.resetDomain()
+ except Exception, ex:
+ raise XendError(str(ex))
+
def instance():
"""Singleton constructor. Use this instead of the class constructor.
diff -r 239b44eeb2d6 -r dc510776dd59 tools/python/xen/xend/XendDomainInfo.py
--- a/tools/python/xen/xend/XendDomainInfo.py Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/python/xen/xend/XendDomainInfo.py Thu Apr 24 14:08:29 2008 -0600
@@ -1837,6 +1837,9 @@ class XendDomainInfo:
@raise: VmError for invalid devices
"""
+ if self.image:
+ self.image.prepareEnvironment()
+
ordered_refs = self.info.ordered_device_refs()
for dev_uuid in ordered_refs:
devclass, config = self.info['devices'][dev_uuid]
@@ -2323,6 +2326,34 @@ class XendDomainInfo:
self._cleanup_phantom_devs(paths)
+ def resetDomain(self):
+ log.debug("XendDomainInfo.resetDomain(%s)", str(self.domid))
+
+ old_domid = self.domid
+ prev_vm_xend = self._listRecursiveVm('xend')
+ new_dom_info = self.info
+ try:
+ self._unwatchVm()
+ self.destroy()
+
+ new_dom = None
+ try:
+ from xen.xend import XendDomain
+ new_dom_info['domid'] = None
+ new_dom = XendDomain.instance().domain_create_from_dict(
+ new_dom_info)
+ for x in prev_vm_xend[0][1]:
+ new_dom._writeVm('xend/%s' % x[0], x[1])
+ new_dom.waitForDevices()
+ new_dom.unpause()
+ except:
+ if new_dom:
+ new_dom.destroy()
+ raise
+ except:
+ log.exception('Failed to reset domain %s.', str(old_domid))
+
+
def resumeDomain(self):
log.debug("XendDomainInfo.resumeDomain(%s)", str(self.domid))
diff -r 239b44eeb2d6 -r dc510776dd59 tools/python/xen/xend/XendXSPolicyAdmin.py
--- a/tools/python/xen/xend/XendXSPolicyAdmin.py Thu Apr 24 14:02:16
2008 -0600
+++ b/tools/python/xen/xend/XendXSPolicyAdmin.py Thu Apr 24 14:08:29
2008 -0600
@@ -54,6 +54,7 @@ class XSPolicyAdmin:
try:
self.xsobjs[ref] = ACMPolicy(name=act_pol_name, ref=ref)
self.policies[ref] = (act_pol_name, xsconstants.ACM_POLICY_ID)
+ self.xsobjs[ref].validate_enforced_policy_hash()
except Exception, e:
log.error("Could not find XML representation of policy '%s': "
"%s" % (act_pol_name,e))
diff -r 239b44eeb2d6 -r dc510776dd59 tools/python/xen/xend/image.py
--- a/tools/python/xen/xend/image.py Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/python/xen/xend/image.py Thu Apr 24 14:08:29 2008 -0600
@@ -184,6 +184,42 @@ class ImageHandler:
def buildDomain(self):
"""Build the domain. Define in subclass."""
raise NotImplementedError()
+
+ def prepareEnvironment(self):
+ """Prepare the environment for the execution of the domain. This
+ method is called before any devices are set up."""
+
+ domid = self.vm.getDomid()
+
+ # Delete left-over pipes
+ try:
+ os.unlink('/var/run/tap/qemu-read-%d' % domid)
+ os.unlink('/var/run/tap/qemu-write-%d' % domid)
+ except:
+ pass
+
+ # No device model, don't create pipes
+ if self.device_model is None:
+ return
+
+ # If we use a device model, the pipes for communication between
+ # blktapctrl and ioemu must be present before the devices are
+ # created (blktapctrl must access them for new block devices)
+
+ # mkdir throws an exception if the path already exists
+ try:
+ os.mkdir('/var/run/tap', 0755)
+ except:
+ pass
+
+ try:
+ os.mkfifo('/var/run/tap/qemu-read-%d' % domid, 0600)
+ os.mkfifo('/var/run/tap/qemu-write-%d' % domid, 0600)
+ except OSError, e:
+ log.warn('Could not create blktap pipes for domain %d' % domid)
+ log.exception(e)
+ pass
+
# Return a list of cmd line args to the device models based on the
# xm config file
@@ -411,6 +447,12 @@ class ImageHandler:
self.pid = None
state = xstransact.Remove("/local/domain/0/device-model/%i"
% self.vm.getDomid())
+
+ try:
+ os.unlink('/var/run/tap/qemu-read-%d' % self.vm.getDomid())
+ os.unlink('/var/run/tap/qemu-write-%d' % self.vm.getDomid())
+ except:
+ pass
class LinuxImageHandler(ImageHandler):
@@ -643,7 +685,9 @@ class IA64_HVM_ImageHandler(HVMImageHand
# ROM size for guest firmware, io page, xenstore page
# buffer io page, buffer pio page and memmap info page
extra_pages = 1024 + 5
- return mem_kb + extra_pages * page_kb
+ mem_kb += extra_pages * page_kb
+ # Add 8 MiB overhead for QEMU's video RAM.
+ return mem_kb + 8192
def getRequiredInitialReservation(self):
return self.vm.getMemoryTarget()
diff -r 239b44eeb2d6 -r dc510776dd59 tools/python/xen/xm/main.py
--- a/tools/python/xen/xm/main.py Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/python/xen/xm/main.py Thu Apr 24 14:08:29 2008 -0600
@@ -107,6 +107,7 @@ SUBCOMMAND_HELP = {
'Migrate a domain to another machine.'),
'pause' : ('<Domain>', 'Pause execution of a domain.'),
'reboot' : ('<Domain> [-wa]', 'Reboot a domain.'),
+ 'reset' : ('<Domain>', 'Reset a domain.'),
'restore' : ('<CheckpointFile> [-p]',
'Restore a domain from a saved state.'),
'save' : ('[-c] <Domain> <CheckpointFile>',
@@ -274,6 +275,7 @@ common_commands = [
"migrate",
"pause",
"reboot",
+ "reset",
"restore",
"resume",
"save",
@@ -303,6 +305,7 @@ domain_commands = [
"pause",
"reboot",
"rename",
+ "reset",
"restore",
"resume",
"save",
@@ -1247,6 +1250,13 @@ def xm_shutdown(args):
arg_check(args, "shutdown", 1, 4)
from xen.xm import shutdown
shutdown.main(["shutdown"] + args)
+
+def xm_reset(args):
+ arg_check(args, "reset", 1)
+ dom = args[0]
+
+ # TODO: XenAPI
+ server.xend.domain.reset(dom)
def xm_pause(args):
arg_check(args, "pause", 1)
@@ -2474,6 +2484,7 @@ commands = {
"dump-core": xm_dump_core,
"reboot": xm_reboot,
"rename": xm_rename,
+ "reset": xm_reset,
"restore": xm_restore,
"resume": xm_resume,
"save": xm_save,
diff -r 239b44eeb2d6 -r dc510776dd59 tools/tests/test_x86_emulator.c
--- a/tools/tests/test_x86_emulator.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/tests/test_x86_emulator.c Thu Apr 24 14:08:29 2008 -0600
@@ -26,14 +26,8 @@ static int read(
unsigned int bytes,
struct x86_emulate_ctxt *ctxt)
{
- unsigned long addr = offset;
- switch ( bytes )
- {
- case 1: *val = *(uint8_t *)addr; break;
- case 2: *val = *(uint16_t *)addr; break;
- case 4: *val = *(uint32_t *)addr; break;
- case 8: *val = *(unsigned long *)addr; break;
- }
+ *val = 0;
+ memcpy(val, (void *)offset, bytes);
return X86EMUL_OKAY;
}
@@ -44,48 +38,19 @@ static int write(
unsigned int bytes,
struct x86_emulate_ctxt *ctxt)
{
- unsigned long addr = offset;
- switch ( bytes )
- {
- case 1: *(uint8_t *)addr = (uint8_t)val; break;
- case 2: *(uint16_t *)addr = (uint16_t)val; break;
- case 4: *(uint32_t *)addr = (uint32_t)val; break;
- case 8: *(unsigned long *)addr = val; break;
- }
+ memcpy((void *)offset, &val, bytes);
return X86EMUL_OKAY;
}
static int cmpxchg(
unsigned int seg,
unsigned long offset,
- unsigned long old,
- unsigned long new,
+ void *old,
+ void *new,
unsigned int bytes,
struct x86_emulate_ctxt *ctxt)
{
- unsigned long addr = offset;
- switch ( bytes )
- {
- case 1: *(uint8_t *)addr = (uint8_t)new; break;
- case 2: *(uint16_t *)addr = (uint16_t)new; break;
- case 4: *(uint32_t *)addr = (uint32_t)new; break;
- case 8: *(unsigned long *)addr = new; break;
- }
- return X86EMUL_OKAY;
-}
-
-static int cmpxchg8b(
- unsigned int seg,
- unsigned long offset,
- unsigned long old_lo,
- unsigned long old_hi,
- unsigned long new_lo,
- unsigned long new_hi,
- struct x86_emulate_ctxt *ctxt)
-{
- unsigned long addr = offset;
- ((unsigned long *)addr)[0] = new_lo;
- ((unsigned long *)addr)[1] = new_hi;
+ memcpy((void *)offset, new, bytes);
return X86EMUL_OKAY;
}
@@ -94,7 +59,6 @@ static struct x86_emulate_ops emulops =
.insn_fetch = read,
.write = write,
.cmpxchg = cmpxchg,
- .cmpxchg8b = cmpxchg8b
};
int main(int argc, char **argv)
diff -r 239b44eeb2d6 -r dc510776dd59 tools/tests/x86_emulate.c
--- a/tools/tests/x86_emulate.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/tests/x86_emulate.c Thu Apr 24 14:08:29 2008 -0600
@@ -4,10 +4,4 @@
#include <public/xen.h>
#include "x86_emulate/x86_emulate.h"
-
-#define __emulate_fpu_insn(_op) \
-do{ rc = X86EMUL_UNHANDLEABLE; \
- goto done; \
-} while (0)
-
#include "x86_emulate/x86_emulate.c"
diff -r 239b44eeb2d6 -r dc510776dd59 tools/xenmon/xenbaked.c
--- a/tools/xenmon/xenbaked.c Thu Apr 24 14:02:16 2008 -0600
+++ b/tools/xenmon/xenbaked.c Thu Apr 24 14:08:29 2008 -0600
@@ -509,14 +509,36 @@ int monitor_tbufs(void)
{
for ( i = 0; (i < num) && !interrupted; i++ )
{
- while ( meta[i]->cons != meta[i]->prod )
+ unsigned long start_offset, end_offset, cons, prod;
+
+ cons = meta[i]->cons;
+ prod = meta[i]->prod;
+ xen_rmb(); /* read prod, then read item. */
+
+ if ( cons == prod )
+ continue;
+
+ start_offset = cons % data_size;
+ end_offset = prod % data_size;
+
+ if ( start_offset >= end_offset )
{
- xen_rmb(); /* read prod, then read item. */
+ while ( start_offset != data_size )
+ {
+ rec_size = process_record(
+ i, (struct t_rec *)(data[i] + start_offset));
+ start_offset += rec_size;
+ }
+ start_offset = 0;
+ }
+ while ( start_offset != end_offset )
+ {
rec_size = process_record(
- i, (struct t_rec *)(data[i] + meta[i]->cons % data_size));
- xen_mb(); /* read item, then update cons. */
- meta[i]->cons += rec_size;
+ i, (struct t_rec *)(data[i] + start_offset));
+ start_offset += rec_size;
}
+ xen_mb(); /* read item, then update cons. */
+ meta[i]->cons = prod;
}
wait_for_event();
diff -r 239b44eeb2d6 -r dc510776dd59 xen/Makefile
--- a/xen/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -44,6 +44,7 @@ _clean: delete-unfresh-files
$(MAKE) -f $(BASEDIR)/Rules.mk -C arch/$(TARGET_ARCH) clean
rm -f include/asm *.o $(TARGET)* *~ core
rm -f include/asm-*/asm-offsets.h
+ [ -d tools/figlet ] && rm -f .banner*
.PHONY: _distclean
_distclean: clean
@@ -70,8 +71,14 @@ delete-unfresh-files:
rm -f include/xen/compile.h; \
fi
+.banner: Makefile
+ $(MAKE) -C tools
+ @tools/figlet/figlet -d tools/figlet Xen $(XEN_FULLVERSION) 2>$@2 >$@1
+ @cat $@1 $@2 >$@
+ @rm -f $@1 $@2
+
# compile.h contains dynamic build info. Rebuilt on every 'make' invocation.
-include/xen/compile.h: include/xen/compile.h.in
+include/xen/compile.h: include/xen/compile.h.in .banner
@sed -e 's/@@date@@/$(shell LC_ALL=C date)/g' \
-e 's/@@time@@/$(shell LC_ALL=C date +%T)/g' \
-e 's/@@whoami@@/$(USER)/g' \
@@ -83,7 +90,8 @@ include/xen/compile.h: include/xen/compi
-e 's/@@extraversion@@/$(XEN_EXTRAVERSION)/g' \
-e 's!@@changeset@@!$(shell ((hg parents --template "{date|date}
{rev}:{node|short}" >/dev/null && hg parents --template "{date|date}
{rev}:{node|short}") || echo "unavailable") 2>/dev/null)!g' \
< include/xen/compile.h.in > $@.new
- tools/figlet/figlet -d tools/figlet Xen $(XEN_FULLVERSION) >> $@.new
+ @grep \" .banner >> $@.new
+ @grep -v \" .banner
@mv -f $@.new $@
include/asm-$(TARGET_ARCH)/asm-offsets.h: arch/$(TARGET_ARCH)/asm-offsets.s
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/Makefile
--- a/xen/arch/x86/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -52,6 +52,8 @@ obj-y += tboot.o
obj-$(crash_debug) += gdbstub.o
+x86_emulate.o: x86_emulate/x86_emulate.c x86_emulate/x86_emulate.h
+
$(TARGET): $(TARGET)-syms boot/mkelf32
./boot/mkelf32 $(TARGET)-syms $(TARGET) 0x100000 \
`$(NM) -nr $(TARGET)-syms | head -n 1 | sed -e 's/^\([^ ]*\).*/0x\1/'`
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/bitops.c
--- a/xen/arch/x86/bitops.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/bitops.c Thu Apr 24 14:08:29 2008 -0600
@@ -8,17 +8,18 @@ unsigned int __find_first_bit(
unsigned long d0, d1, res;
asm volatile (
- " xor %%eax,%%eax\n\t" /* also ensures ZF==1 if size==0 */
+ "1: xor %%eax,%%eax\n\t" /* also ensures ZF==1 if size==0 */
" repe; scas"__OS"\n\t"
- " je 1f\n\t"
+ " je 2f\n\t"
+ " bsf -"STR(BITS_PER_LONG/8)"(%2),%0\n\t"
+ " jz 1b\n\t"
" lea -"STR(BITS_PER_LONG/8)"(%2),%2\n\t"
- " bsf (%2),%0\n"
- "1: sub %%ebx,%%edi\n\t"
+ "2: sub %%ebx,%%edi\n\t"
" shl $3,%%edi\n\t"
" add %%edi,%%eax"
: "=&a" (res), "=&c" (d0), "=&D" (d1)
- : "1" ((size + BITS_PER_LONG - 1) / BITS_PER_LONG),
- "2" (addr), "b" ((int)(long)addr) : "memory" );
+ : "1" (BITS_TO_LONGS(size)), "2" (addr), "b" ((int)(long)addr)
+ : "memory" );
return res;
}
@@ -34,8 +35,7 @@ unsigned int __find_next_bit(
if ( bit != 0 )
{
/* Look for a bit in the first word. */
- asm ( "bsf %1,%%"__OP"ax"
- : "=a" (set) : "r" (*p >> bit), "0" (BITS_PER_LONG) );
+ set = __scanbit(*p >> bit, BITS_PER_LONG - bit);
if ( set < (BITS_PER_LONG - bit) )
return (offset + set);
offset += BITS_PER_LONG - bit;
@@ -56,18 +56,20 @@ unsigned int __find_first_zero_bit(
unsigned long d0, d1, d2, res;
asm volatile (
+ "1: xor %%eax,%%eax ; not %3\n\t" /* rAX == ~0ul */
" xor %%edx,%%edx\n\t" /* also ensures ZF==1 if size==0 */
" repe; scas"__OS"\n\t"
- " je 1f\n\t"
+ " je 2f\n\t"
+ " xor -"STR(BITS_PER_LONG/8)"(%2),%3\n\t"
+ " jz 1b\n\t"
+ " bsf %3,%0\n\t"
" lea -"STR(BITS_PER_LONG/8)"(%2),%2\n\t"
- " xor (%2),%3\n\t"
- " bsf %3,%0\n"
- "1: sub %%ebx,%%edi\n\t"
+ "2: sub %%ebx,%%edi\n\t"
" shl $3,%%edi\n\t"
" add %%edi,%%edx"
: "=&d" (res), "=&c" (d0), "=&D" (d1), "=&a" (d2)
- : "1" ((size + BITS_PER_LONG - 1) / BITS_PER_LONG),
- "2" (addr), "b" ((int)(long)addr), "3" (-1L) : "memory" );
+ : "1" (BITS_TO_LONGS(size)), "2" (addr), "b" ((int)(long)addr)
+ : "memory" );
return res;
}
@@ -83,7 +85,7 @@ unsigned int __find_next_zero_bit(
if ( bit != 0 )
{
/* Look for zero in the first word. */
- asm ( "bsf %1,%%"__OP"ax" : "=a" (set) : "r" (~(*p >> bit)) );
+ set = __scanbit(~(*p >> bit), BITS_PER_LONG - bit);
if ( set < (BITS_PER_LONG - bit) )
return (offset + set);
offset += BITS_PER_LONG - bit;
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/cpu/mtrr/main.c
--- a/xen/arch/x86/cpu/mtrr/main.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/cpu/mtrr/main.c Thu Apr 24 14:08:29 2008 -0600
@@ -586,8 +586,6 @@ struct mtrr_value {
unsigned long lsize;
};
-extern void global_init_mtrr_pat(void);
-
/**
* mtrr_bp_init - initialize mtrrs on the boot CPU
*
@@ -654,11 +652,8 @@ void __init mtrr_bp_init(void)
if (mtrr_if) {
set_num_var_ranges();
init_table();
- if (use_intel()) {
+ if (use_intel())
get_mtrr_state();
- /* initialize some global data for MTRR/PAT
virutalization */
- global_init_mtrr_pat();
- }
}
}
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/domain.c
--- a/xen/arch/x86/domain.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/domain.c Thu Apr 24 14:08:29 2008 -0600
@@ -521,10 +521,10 @@ int arch_domain_create(struct domain *d,
clear_page(d->shared_info);
share_xen_page_with_guest(
virt_to_page(d->shared_info), d, XENSHARE_writable);
- }
-
- if ( (rc = iommu_domain_init(d)) != 0 )
- goto fail;
+
+ if ( (rc = iommu_domain_init(d)) != 0 )
+ goto fail;
+ }
if ( is_hvm_domain(d) )
{
@@ -562,7 +562,8 @@ void arch_domain_destroy(struct domain *
if ( is_hvm_domain(d) )
hvm_domain_destroy(d);
- iommu_domain_destroy(d);
+ if ( !is_idle_domain(d) )
+ iommu_domain_destroy(d);
paging_final_teardown(d);
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/domain_build.c
--- a/xen/arch/x86/domain_build.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/domain_build.c Thu Apr 24 14:08:29 2008 -0600
@@ -957,8 +957,8 @@ int __init construct_dom0(
rc |= ioports_deny_access(dom0, 0x40, 0x43);
/* PIT Channel 2 / PC Speaker Control. */
rc |= ioports_deny_access(dom0, 0x61, 0x61);
- /* PCI configuration spaces. */
- rc |= ioports_deny_access(dom0, 0xcf8, 0xcff);
+ /* PCI configuration space (NB. 0xcf8 has special treatment). */
+ rc |= ioports_deny_access(dom0, 0xcfc, 0xcff);
/* Command-line I/O ranges. */
process_dom0_ioports_disable();
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/emulate.c
--- a/xen/arch/x86/hvm/emulate.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/emulate.c Thu Apr 24 14:08:29 2008 -0600
@@ -28,6 +28,33 @@ static int hvmemul_do_io(
ioreq_t *p = &vio->vp_ioreq;
int rc;
+ /* Only retrieve the value from singleton (non-REP) reads. */
+ ASSERT((val == NULL) || ((dir == IOREQ_READ) && !value_is_ptr));
+
+ if ( is_mmio && !value_is_ptr )
+ {
+ /* Part of a multi-cycle read or write? */
+ if ( dir == IOREQ_WRITE )
+ {
+ paddr_t pa = curr->arch.hvm_vcpu.mmio_large_write_pa;
+ unsigned int bytes = curr->arch.hvm_vcpu.mmio_large_write_bytes;
+ if ( (addr >= pa) && ((addr + size) <= (pa + bytes)) )
+ return X86EMUL_OKAY;
+ }
+ else
+ {
+ paddr_t pa = curr->arch.hvm_vcpu.mmio_large_read_pa;
+ unsigned int bytes = curr->arch.hvm_vcpu.mmio_large_read_bytes;
+ if ( (addr >= pa) && ((addr + size) <= (pa + bytes)) )
+ {
+ *val = 0;
+ memcpy(val, &curr->arch.hvm_vcpu.mmio_large_read[addr - pa],
+ size);
+ return X86EMUL_OKAY;
+ }
+ }
+ }
+
switch ( curr->arch.hvm_vcpu.io_state )
{
case HVMIO_none:
@@ -36,8 +63,13 @@ static int hvmemul_do_io(
curr->arch.hvm_vcpu.io_state = HVMIO_none;
if ( val == NULL )
return X86EMUL_UNHANDLEABLE;
- *val = curr->arch.hvm_vcpu.io_data;
- return X86EMUL_OKAY;
+ goto finish_access;
+ case HVMIO_dispatched:
+ /* May have to wait for previous cycle of a multi-write to complete. */
+ if ( is_mmio && !value_is_ptr && (dir == IOREQ_WRITE) &&
+ (addr == (curr->arch.hvm_vcpu.mmio_large_write_pa +
+ curr->arch.hvm_vcpu.mmio_large_write_bytes)) )
+ return X86EMUL_RETRY;
default:
return X86EMUL_UNHANDLEABLE;
}
@@ -80,8 +112,6 @@ static int hvmemul_do_io(
*reps = p->count;
p->state = STATE_IORESP_READY;
hvm_io_assist();
- if ( val != NULL )
- *val = curr->arch.hvm_vcpu.io_data;
curr->arch.hvm_vcpu.io_state = HVMIO_none;
break;
case X86EMUL_UNHANDLEABLE:
@@ -92,7 +122,43 @@ static int hvmemul_do_io(
BUG();
}
- return rc;
+ if ( rc != X86EMUL_OKAY )
+ return rc;
+
+ finish_access:
+ if ( val != NULL )
+ *val = curr->arch.hvm_vcpu.io_data;
+
+ if ( is_mmio && !value_is_ptr )
+ {
+ /* Part of a multi-cycle read or write? */
+ if ( dir == IOREQ_WRITE )
+ {
+ paddr_t pa = curr->arch.hvm_vcpu.mmio_large_write_pa;
+ unsigned int bytes = curr->arch.hvm_vcpu.mmio_large_write_bytes;
+ if ( bytes == 0 )
+ pa = curr->arch.hvm_vcpu.mmio_large_write_pa = addr;
+ if ( addr == (pa + bytes) )
+ curr->arch.hvm_vcpu.mmio_large_write_bytes += size;
+ }
+ else
+ {
+ paddr_t pa = curr->arch.hvm_vcpu.mmio_large_read_pa;
+ unsigned int bytes = curr->arch.hvm_vcpu.mmio_large_read_bytes;
+ if ( bytes == 0 )
+ pa = curr->arch.hvm_vcpu.mmio_large_read_pa = addr;
+ if ( (addr == (pa + bytes)) &&
+ ((bytes + size) <
+ sizeof(curr->arch.hvm_vcpu.mmio_large_read)) )
+ {
+ memcpy(&curr->arch.hvm_vcpu.mmio_large_read[addr - pa],
+ val, size);
+ curr->arch.hvm_vcpu.mmio_large_read_bytes += size;
+ }
+ }
+ }
+
+ return X86EMUL_OKAY;
}
static int hvmemul_do_pio(
@@ -371,11 +437,15 @@ static int hvmemul_cmpxchg(
static int hvmemul_cmpxchg(
enum x86_segment seg,
unsigned long offset,
- unsigned long old,
- unsigned long new,
+ void *p_old,
+ void *p_new,
unsigned int bytes,
struct x86_emulate_ctxt *ctxt)
{
+ unsigned long new = 0;
+ if ( bytes > sizeof(new) )
+ return X86EMUL_UNHANDLEABLE;
+ memcpy(&new, p_new, bytes);
/* Fix this in case the guest is really relying on r-m-w atomicity. */
return hvmemul_write(seg, offset, new, bytes, ctxt);
}
@@ -603,7 +673,7 @@ static int hvmemul_read_msr(
_regs.ecx = (uint32_t)reg;
- if ( (rc = hvm_funcs.msr_read_intercept(&_regs)) != 0 )
+ if ( (rc = hvm_msr_read_intercept(&_regs)) != 0 )
return rc;
*val = ((uint64_t)(uint32_t)_regs.edx << 32) || (uint32_t)_regs.eax;
@@ -621,7 +691,7 @@ static int hvmemul_write_msr(
_regs.eax = (uint32_t)val;
_regs.ecx = (uint32_t)reg;
- return hvm_funcs.msr_write_intercept(&_regs);
+ return hvm_msr_write_intercept(&_regs);
}
static int hvmemul_wbinvd(
@@ -674,11 +744,40 @@ static int hvmemul_inject_sw_interrupt(
return X86EMUL_OKAY;
}
-static void hvmemul_load_fpu_ctxt(
- struct x86_emulate_ctxt *ctxt)
-{
- if ( !current->fpu_dirtied )
+static int hvmemul_get_fpu(
+ void (*exception_callback)(void *, struct cpu_user_regs *),
+ void *exception_callback_arg,
+ enum x86_emulate_fpu_type type,
+ struct x86_emulate_ctxt *ctxt)
+{
+ struct vcpu *curr = current;
+
+ switch ( type )
+ {
+ case X86EMUL_FPU_fpu:
+ break;
+ case X86EMUL_FPU_mmx:
+ if ( !cpu_has_mmx )
+ return X86EMUL_UNHANDLEABLE;
+ break;
+ default:
+ return X86EMUL_UNHANDLEABLE;
+ }
+
+ if ( !curr->fpu_dirtied )
hvm_funcs.fpu_dirty_intercept();
+
+ curr->arch.hvm_vcpu.fpu_exception_callback = exception_callback;
+ curr->arch.hvm_vcpu.fpu_exception_callback_arg = exception_callback_arg;
+
+ return X86EMUL_OKAY;
+}
+
+static void hvmemul_put_fpu(
+ struct x86_emulate_ctxt *ctxt)
+{
+ struct vcpu *curr = current;
+ curr->arch.hvm_vcpu.fpu_exception_callback = NULL;
}
static int hvmemul_invlpg(
@@ -720,7 +819,8 @@ static struct x86_emulate_ops hvm_emulat
.cpuid = hvmemul_cpuid,
.inject_hw_exception = hvmemul_inject_hw_exception,
.inject_sw_interrupt = hvmemul_inject_sw_interrupt,
- .load_fpu_ctxt = hvmemul_load_fpu_ctxt,
+ .get_fpu = hvmemul_get_fpu,
+ .put_fpu = hvmemul_put_fpu,
.invlpg = hvmemul_invlpg
};
@@ -763,6 +863,11 @@ int hvm_emulate_one(
hvmemul_ctxt->exn_pending = 0;
rc = x86_emulate(&hvmemul_ctxt->ctxt, &hvm_emulate_ops);
+
+ if ( rc != X86EMUL_RETRY )
+ curr->arch.hvm_vcpu.mmio_large_read_bytes =
+ curr->arch.hvm_vcpu.mmio_large_write_bytes = 0;
+
if ( rc != X86EMUL_OKAY )
return rc;
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/hvm.c
--- a/xen/arch/x86/hvm/hvm.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/hvm.c Thu Apr 24 14:08:29 2008 -0600
@@ -494,14 +494,14 @@ static int hvm_load_cpu_ctxt(struct doma
((ctxt.cr0 & (X86_CR0_PE|X86_CR0_PG)) == X86_CR0_PG) )
{
gdprintk(XENLOG_ERR, "HVM restore: bad CR0 0x%"PRIx64"\n",
- ctxt.msr_efer);
+ ctxt.cr0);
return -EINVAL;
}
if ( ctxt.cr4 & HVM_CR4_GUEST_RESERVED_BITS )
{
gdprintk(XENLOG_ERR, "HVM restore: bad CR4 0x%"PRIx64"\n",
- ctxt.msr_efer);
+ ctxt.cr4);
return -EINVAL;
}
@@ -620,8 +620,6 @@ HVM_REGISTER_SAVE_RESTORE(CPU, hvm_save_
HVM_REGISTER_SAVE_RESTORE(CPU, hvm_save_cpu_ctxt, hvm_load_cpu_ctxt,
1, HVMSR_PER_VCPU);
-extern int reset_vmsr(struct mtrr_state *m, u64 *p);
-
int hvm_vcpu_initialise(struct vcpu *v)
{
int rc;
@@ -647,7 +645,7 @@ int hvm_vcpu_initialise(struct vcpu *v)
spin_lock_init(&v->arch.hvm_vcpu.tm_lock);
INIT_LIST_HEAD(&v->arch.hvm_vcpu.tm_list);
- rc = reset_vmsr(&v->arch.hvm_vcpu.mtrr, &v->arch.hvm_vcpu.pat_cr);
+ rc = hvm_vcpu_cacheattr_init(v);
if ( rc != 0 )
goto fail3;
@@ -681,6 +679,7 @@ int hvm_vcpu_initialise(struct vcpu *v)
void hvm_vcpu_destroy(struct vcpu *v)
{
+ hvm_vcpu_cacheattr_destroy(v);
vlapic_destroy(v);
hvm_funcs.vcpu_destroy(v);
@@ -1604,6 +1603,9 @@ void hvm_cpuid(unsigned int input, unsig
*ebx &= 0x0000FFFFu;
*ebx |= (current->vcpu_id * 2) << 24;
+ /* We always support MTRR MSRs. */
+ *edx |= bitmaskof(X86_FEATURE_MTRR);
+
*ecx &= (bitmaskof(X86_FEATURE_XMM3) |
bitmaskof(X86_FEATURE_SSSE3) |
bitmaskof(X86_FEATURE_CX16) |
@@ -1653,6 +1655,146 @@ void hvm_cpuid(unsigned int input, unsig
#endif
break;
}
+}
+
+int hvm_msr_read_intercept(struct cpu_user_regs *regs)
+{
+ uint32_t ecx = regs->ecx;
+ uint64_t msr_content = 0;
+ struct vcpu *v = current;
+ uint64_t *var_range_base, *fixed_range_base;
+ int index;
+
+ var_range_base = (uint64_t *)v->arch.hvm_vcpu.mtrr.var_ranges;
+ fixed_range_base = (uint64_t *)v->arch.hvm_vcpu.mtrr.fixed_ranges;
+
+ switch ( ecx )
+ {
+ case MSR_IA32_TSC:
+ msr_content = hvm_get_guest_time(v);
+ break;
+
+ case MSR_IA32_APICBASE:
+ msr_content = vcpu_vlapic(v)->hw.apic_base_msr;
+ break;
+
+ case MSR_IA32_MCG_CAP:
+ case MSR_IA32_MCG_STATUS:
+ case MSR_IA32_MC0_STATUS:
+ case MSR_IA32_MC1_STATUS:
+ case MSR_IA32_MC2_STATUS:
+ case MSR_IA32_MC3_STATUS:
+ case MSR_IA32_MC4_STATUS:
+ case MSR_IA32_MC5_STATUS:
+ /* No point in letting the guest see real MCEs */
+ msr_content = 0;
+ break;
+
+ case MSR_IA32_CR_PAT:
+ msr_content = v->arch.hvm_vcpu.pat_cr;
+ break;
+
+ case MSR_MTRRcap:
+ msr_content = v->arch.hvm_vcpu.mtrr.mtrr_cap;
+ break;
+ case MSR_MTRRdefType:
+ msr_content = v->arch.hvm_vcpu.mtrr.def_type
+ | (v->arch.hvm_vcpu.mtrr.enabled << 10);
+ break;
+ case MSR_MTRRfix64K_00000:
+ msr_content = fixed_range_base[0];
+ break;
+ case MSR_MTRRfix16K_80000:
+ case MSR_MTRRfix16K_A0000:
+ index = regs->ecx - MSR_MTRRfix16K_80000;
+ msr_content = fixed_range_base[index + 1];
+ break;
+ case MSR_MTRRfix4K_C0000...MSR_MTRRfix4K_F8000:
+ index = regs->ecx - MSR_MTRRfix4K_C0000;
+ msr_content = fixed_range_base[index + 3];
+ break;
+ case MSR_IA32_MTRR_PHYSBASE0...MSR_IA32_MTRR_PHYSMASK7:
+ index = regs->ecx - MSR_IA32_MTRR_PHYSBASE0;
+ msr_content = var_range_base[index];
+ break;
+
+ default:
+ return hvm_funcs.msr_read_intercept(regs);
+ }
+
+ regs->eax = (uint32_t)msr_content;
+ regs->edx = (uint32_t)(msr_content >> 32);
+ return X86EMUL_OKAY;
+}
+
+int hvm_msr_write_intercept(struct cpu_user_regs *regs)
+{
+ extern bool_t mtrr_var_range_msr_set(
+ struct mtrr_state *v, u32 msr, u64 msr_content);
+ extern bool_t mtrr_fix_range_msr_set(
+ struct mtrr_state *v, int row, u64 msr_content);
+ extern bool_t mtrr_def_type_msr_set(struct mtrr_state *v, u64 msr_content);
+ extern bool_t pat_msr_set(u64 *pat, u64 msr);
+
+ uint32_t ecx = regs->ecx;
+ uint64_t msr_content = (uint32_t)regs->eax | ((uint64_t)regs->edx << 32);
+ struct vcpu *v = current;
+ int index;
+
+ switch ( ecx )
+ {
+ case MSR_IA32_TSC:
+ hvm_set_guest_time(v, msr_content);
+ pt_reset(v);
+ break;
+
+ case MSR_IA32_APICBASE:
+ vlapic_msr_set(vcpu_vlapic(v), msr_content);
+ break;
+
+ case MSR_IA32_CR_PAT:
+ if ( !pat_msr_set(&v->arch.hvm_vcpu.pat_cr, msr_content) )
+ goto gp_fault;
+ break;
+
+ case MSR_MTRRcap:
+ goto gp_fault;
+ case MSR_MTRRdefType:
+ if ( !mtrr_def_type_msr_set(&v->arch.hvm_vcpu.mtrr, msr_content) )
+ goto gp_fault;
+ break;
+ case MSR_MTRRfix64K_00000:
+ if ( !mtrr_fix_range_msr_set(&v->arch.hvm_vcpu.mtrr, 0, msr_content) )
+ goto gp_fault;
+ break;
+ case MSR_MTRRfix16K_80000:
+ case MSR_MTRRfix16K_A0000:
+ index = regs->ecx - MSR_MTRRfix16K_80000 + 1;
+ if ( !mtrr_fix_range_msr_set(&v->arch.hvm_vcpu.mtrr,
+ index, msr_content) )
+ goto gp_fault;
+ break;
+ case MSR_MTRRfix4K_C0000...MSR_MTRRfix4K_F8000:
+ index = regs->ecx - MSR_MTRRfix4K_C0000 + 3;
+ if ( !mtrr_fix_range_msr_set(&v->arch.hvm_vcpu.mtrr,
+ index, msr_content) )
+ goto gp_fault;
+ break;
+ case MSR_IA32_MTRR_PHYSBASE0...MSR_IA32_MTRR_PHYSMASK7:
+ if ( !mtrr_var_range_msr_set(&v->arch.hvm_vcpu.mtrr,
+ regs->ecx, msr_content) )
+ goto gp_fault;
+ break;
+
+ default:
+ return hvm_funcs.msr_write_intercept(regs);
+ }
+
+ return X86EMUL_OKAY;
+
+gp_fault:
+ hvm_inject_exception(TRAP_gp_fault, 0, 0);
+ return X86EMUL_EXCEPTION;
}
enum hvm_intblk hvm_interrupt_blocked(struct vcpu *v, struct hvm_intack intack)
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/mtrr.c
--- a/xen/arch/x86/hvm/mtrr.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/mtrr.c Thu Apr 24 14:08:29 2008 -0600
@@ -27,7 +27,6 @@
#include <asm/hvm/support.h>
#include <asm/hvm/cacheattr.h>
-/* Xen holds the native MTRR MSRs */
extern struct mtrr_state mtrr_state;
static uint64_t phys_base_msr_mask;
@@ -35,19 +34,17 @@ static uint32_t size_or_mask;
static uint32_t size_or_mask;
static uint32_t size_and_mask;
-static void init_pat_entry_tbl(uint64_t pat);
-static void init_mtrr_epat_tbl(void);
-static uint8_t get_mtrr_type(struct mtrr_state *m, paddr_t pa);
-/* get page attribute fields (PAn) from PAT MSR */
+/* Get page attribute fields (PAn) from PAT MSR. */
#define pat_cr_2_paf(pat_cr,n) ((((uint64_t)pat_cr) >> ((n)<<3)) & 0xff)
-/* pat entry to PTE flags (PAT, PCD, PWT bits) */
+
+/* PAT entry to PTE flags (PAT, PCD, PWT bits). */
static uint8_t pat_entry_2_pte_flags[8] = {
0, _PAGE_PWT,
_PAGE_PCD, _PAGE_PCD | _PAGE_PWT,
_PAGE_PAT, _PAGE_PAT | _PAGE_PWT,
_PAGE_PAT | _PAGE_PCD, _PAGE_PAT | _PAGE_PCD | _PAGE_PWT };
-/* effective mm type lookup table, according to MTRR and PAT */
+/* Effective mm type lookup table, according to MTRR and PAT. */
static uint8_t mm_type_tbl[MTRR_NUM_TYPES][PAT_TYPE_NUMS] = {
/********PAT(UC,WC,RS,RS,WT,WP,WB,UC-)*/
/* RS means reserved type(2,3), and type is hardcoded here */
@@ -67,12 +64,13 @@ static uint8_t mm_type_tbl[MTRR_NUM_TYPE
{0, 1, 2, 2, 4, 5, 6, 0}
};
-/* reverse lookup table, to find a pat type according to MTRR and effective
- * memory type. This table is dynamically generated
+/*
+ * Reverse lookup table, to find a pat type according to MTRR and effective
+ * memory type. This table is dynamically generated.
*/
static uint8_t mtrr_epat_tbl[MTRR_NUM_TYPES][MEMORY_NUM_TYPES];
-/* lookup table for PAT entry of a given PAT value in host pat */
+/* Lookup table for PAT entry of a given PAT value in host PAT. */
static uint8_t pat_entry_tbl[PAT_TYPE_NUMS];
static void get_mtrr_range(uint64_t base_msr, uint64_t mask_msr,
@@ -139,220 +137,63 @@ bool_t is_var_mtrr_overlapped(struct mtr
return 0;
}
-/* reserved mtrr for guest OS */
-#define RESERVED_MTRR 2
+#define MTRR_PHYSMASK_VALID_BIT 11
+#define MTRR_PHYSMASK_SHIFT 12
+
+#define MTRR_PHYSBASE_TYPE_MASK 0xff /* lowest 8 bits */
+#define MTRR_PHYSBASE_SHIFT 12
+#define MTRR_VCNT 8
+
#define MTRRphysBase_MSR(reg) (0x200 + 2 * (reg))
#define MTRRphysMask_MSR(reg) (0x200 + 2 * (reg) + 1)
bool_t mtrr_var_range_msr_set(struct mtrr_state *m, uint32_t msr,
uint64_t msr_content);
-bool_t mtrr_def_type_msr_set(struct mtrr_state *m, uint64_t msr_content);
bool_t mtrr_fix_range_msr_set(struct mtrr_state *m, uint32_t row,
uint64_t msr_content);
-static void set_var_mtrr(uint32_t reg, struct mtrr_state *m,
- uint32_t base, uint32_t size,
- uint32_t type)
-{
- struct mtrr_var_range *vr;
-
- vr = &m->var_ranges[reg];
-
- if ( size == 0 )
- {
- /* The invalid bit is kept in the mask, so we simply clear the
- * relevant mask register to disable a range.
- */
- mtrr_var_range_msr_set(m, MTRRphysMask_MSR(reg), 0);
- }
- else
- {
- vr->base_lo = base << PAGE_SHIFT | type;
- vr->base_hi = (base & size_and_mask) >> (32 - PAGE_SHIFT);
- vr->mask_lo = -size << PAGE_SHIFT | 0x800;
- vr->mask_hi = (-size & size_and_mask) >> (32 - PAGE_SHIFT);
-
- mtrr_var_range_msr_set(m, MTRRphysBase_MSR(reg), *(uint64_t *)vr);
- mtrr_var_range_msr_set(m, MTRRphysMask_MSR(reg),
- *((uint64_t *)vr + 1));
- }
-}
-/* From Intel Vol. III Section 10.11.4, the Range Size and Base Alignment has
- * some kind of requirement:
- * 1. The range size must be 2^N byte for N >= 12 (i.e 4KB minimum).
- * 2. The base address must be 2^N aligned, where the N here is equal to
- * the N in previous requirement. So a 8K range must be 8K aligned not 4K
aligned.
- */
-static uint32_t range_to_mtrr(uint32_t reg, struct mtrr_state *m,
- uint32_t range_startk, uint32_t range_sizek,
- uint8_t type)
-{
- if ( !range_sizek || (reg >= ((m->mtrr_cap & 0xff) - RESERVED_MTRR)) )
- {
- gdprintk(XENLOG_WARNING,
- "Failed to init var mtrr msr[%d]"
- "range_size:%x, total available MSR:%d\n",
- reg, range_sizek,
- (uint32_t)((m->mtrr_cap & 0xff) - RESERVED_MTRR));
- return reg;
- }
-
- while ( range_sizek )
- {
- uint32_t max_align, align, sizek;
-
- max_align = (range_startk == 0) ? 32 : ffs(range_startk);
- align = min_t(uint32_t, fls(range_sizek), max_align);
- sizek = 1 << (align - 1);
-
- set_var_mtrr(reg++, m, range_startk, sizek, type);
-
- range_startk += sizek;
- range_sizek -= sizek;
-
- if ( reg >= ((m->mtrr_cap & 0xff) - RESERVED_MTRR) )
- {
- gdprintk(XENLOG_WARNING,
- "Failed to init var mtrr msr[%d],"
- "total available MSR:%d\n",
- reg, (uint32_t)((m->mtrr_cap & 0xff) - RESERVED_MTRR));
- break;
- }
- }
-
- return reg;
-}
-
-static void setup_fixed_mtrrs(struct vcpu *v)
-{
- uint64_t content;
- int32_t i;
- struct mtrr_state *m = &v->arch.hvm_vcpu.mtrr;
-
- /* 1. Map (0~A0000) as WB */
- content = 0x0606060606060606ull;
- mtrr_fix_range_msr_set(m, 0, content);
- mtrr_fix_range_msr_set(m, 1, content);
- /* 2. Map VRAM(A0000~C0000) as WC */
- content = 0x0101010101010101;
- mtrr_fix_range_msr_set(m, 2, content);
- /* 3. Map (C0000~100000) as UC */
- for ( i = 3; i < 11; i++)
- mtrr_fix_range_msr_set(m, i, 0);
-}
-
-static void setup_var_mtrrs(struct vcpu *v)
-{
- p2m_type_t p2m;
- uint64_t e820_mfn;
- int8_t *p = NULL;
- uint8_t nr = 0;
- int32_t i;
- uint32_t reg = 0;
- uint64_t size = 0;
- uint64_t addr = 0;
- struct e820entry *e820_table;
-
- e820_mfn = mfn_x(gfn_to_mfn(v->domain,
- HVM_E820_PAGE >> PAGE_SHIFT, &p2m));
-
- p = (int8_t *)map_domain_page(e820_mfn);
-
- nr = *(uint8_t*)(p + HVM_E820_NR_OFFSET);
- e820_table = (struct e820entry*)(p + HVM_E820_OFFSET);
- /* search E820 table, set MTRR for RAM */
- for ( i = 0; i < nr; i++)
- {
- if ( (e820_table[i].addr >= 0x100000) &&
- (e820_table[i].type == E820_RAM) )
- {
- if ( e820_table[i].addr == 0x100000 )
- {
- size = e820_table[i].size + 0x100000 + PAGE_SIZE * 5;
- addr = 0;
- }
- else
- {
- /* Larger than 4G */
- size = e820_table[i].size;
- addr = e820_table[i].addr;
- }
-
- reg = range_to_mtrr(reg, &v->arch.hvm_vcpu.mtrr,
- addr >> PAGE_SHIFT, size >> PAGE_SHIFT,
- MTRR_TYPE_WRBACK);
- }
- }
-}
-
-void init_mtrr_in_hyper(struct vcpu *v)
-{
- /* TODO:MTRR should be initialized in BIOS or other places.
- * workaround to do it in here
- */
- if ( v->arch.hvm_vcpu.mtrr.is_initialized )
- return;
-
- setup_fixed_mtrrs(v);
- setup_var_mtrrs(v);
- /* enable mtrr */
- mtrr_def_type_msr_set(&v->arch.hvm_vcpu.mtrr, 0xc00);
-
- v->arch.hvm_vcpu.mtrr.is_initialized = 1;
-}
-
-static int32_t reset_mtrr(struct mtrr_state *m)
-{
- m->var_ranges = xmalloc_array(struct mtrr_var_range, MTRR_VCNT);
- if ( m->var_ranges == NULL )
- return -ENOMEM;
- memset(m->var_ranges, 0, MTRR_VCNT * sizeof(struct mtrr_var_range));
- memset(m->fixed_ranges, 0, sizeof(m->fixed_ranges));
- m->enabled = 0;
- m->def_type = 0;/*mtrr is disabled*/
- m->mtrr_cap = (0x5<<8)|MTRR_VCNT;/*wc,fix enabled, and vcnt=8*/
- m->overlapped = 0;
- return 0;
-}
-
-/* init global variables for MTRR and PAT */
-void global_init_mtrr_pat(void)
+
+static int hvm_mtrr_pat_init(void)
{
extern uint64_t host_pat;
- uint32_t phys_addr;
-
- init_mtrr_epat_tbl();
- init_pat_entry_tbl(host_pat);
- /* Get max physical address, set some global variable */
- if ( cpuid_eax(0x80000000) < 0x80000008 )
- phys_addr = 36;
- else
- phys_addr = cpuid_eax(0x80000008);
-
- phys_base_msr_mask = ~((((uint64_t)1) << phys_addr) - 1) | 0xf00UL;
- phys_mask_msr_mask = ~((((uint64_t)1) << phys_addr) - 1) | 0x7ffUL;
-
- size_or_mask = ~((1 << (phys_addr - PAGE_SHIFT)) - 1);
- size_and_mask = ~size_or_mask & 0xfff00000;
-}
-
-static void init_pat_entry_tbl(uint64_t pat)
-{
- int32_t i, j;
+ unsigned int i, j, phys_addr;
+
+ memset(&mtrr_epat_tbl, INVALID_MEM_TYPE, sizeof(mtrr_epat_tbl));
+ for ( i = 0; i < MTRR_NUM_TYPES; i++ )
+ {
+ for ( j = 0; j < PAT_TYPE_NUMS; j++ )
+ {
+ int32_t tmp = mm_type_tbl[i][j];
+ if ( (tmp >= 0) && (tmp < MEMORY_NUM_TYPES) )
+ mtrr_epat_tbl[i][tmp] = j;
+ }
+ }
memset(&pat_entry_tbl, INVALID_MEM_TYPE,
PAT_TYPE_NUMS * sizeof(pat_entry_tbl[0]));
-
for ( i = 0; i < PAT_TYPE_NUMS; i++ )
{
for ( j = 0; j < PAT_TYPE_NUMS; j++ )
{
- if ( pat_cr_2_paf(pat, j) == i )
+ if ( pat_cr_2_paf(host_pat, j) == i )
{
pat_entry_tbl[i] = j;
break;
}
}
}
-}
+
+ phys_addr = 36;
+ if ( cpuid_eax(0x80000000) >= 0x80000008 )
+ phys_addr = (uint8_t)cpuid_eax(0x80000008);
+
+ phys_base_msr_mask = ~((((uint64_t)1) << phys_addr) - 1) | 0xf00UL;
+ phys_mask_msr_mask = ~((((uint64_t)1) << phys_addr) - 1) | 0x7ffUL;
+
+ size_or_mask = ~((1 << (phys_addr - PAGE_SHIFT)) - 1);
+ size_and_mask = ~size_or_mask & 0xfff00000;
+
+ return 0;
+}
+__initcall(hvm_mtrr_pat_init);
uint8_t pat_type_2_pte_flags(uint8_t pat_type)
{
@@ -368,24 +209,35 @@ uint8_t pat_type_2_pte_flags(uint8_t pat
return pat_entry_2_pte_flags[pat_entry_tbl[PAT_TYPE_UNCACHABLE]];
}
-int32_t reset_vmsr(struct mtrr_state *m, uint64_t *pat_ptr)
-{
- int32_t rc;
-
- rc = reset_mtrr(m);
- if ( rc != 0 )
- return rc;
-
- *pat_ptr = ((uint64_t)PAT_TYPE_WRBACK) | /* PAT0: WB */
- ((uint64_t)PAT_TYPE_WRTHROUGH << 8) | /* PAT1: WT */
- ((uint64_t)PAT_TYPE_UC_MINUS << 16) | /* PAT2: UC- */
- ((uint64_t)PAT_TYPE_UNCACHABLE << 24) | /* PAT3: UC */
- ((uint64_t)PAT_TYPE_WRBACK << 32) | /* PAT4: WB */
- ((uint64_t)PAT_TYPE_WRTHROUGH << 40) | /* PAT5: WT */
- ((uint64_t)PAT_TYPE_UC_MINUS << 48) | /* PAT6: UC- */
- ((uint64_t)PAT_TYPE_UNCACHABLE << 56); /* PAT7: UC */
-
- return 0;
+int hvm_vcpu_cacheattr_init(struct vcpu *v)
+{
+ struct mtrr_state *m = &v->arch.hvm_vcpu.mtrr;
+
+ memset(m, 0, sizeof(*m));
+
+ m->var_ranges = xmalloc_array(struct mtrr_var_range, MTRR_VCNT);
+ if ( m->var_ranges == NULL )
+ return -ENOMEM;
+ memset(m->var_ranges, 0, MTRR_VCNT * sizeof(struct mtrr_var_range));
+
+ m->mtrr_cap = (1u << 10) | (1u << 8) | MTRR_VCNT;
+
+ v->arch.hvm_vcpu.pat_cr =
+ ((uint64_t)PAT_TYPE_WRBACK) | /* PAT0: WB */
+ ((uint64_t)PAT_TYPE_WRTHROUGH << 8) | /* PAT1: WT */
+ ((uint64_t)PAT_TYPE_UC_MINUS << 16) | /* PAT2: UC- */
+ ((uint64_t)PAT_TYPE_UNCACHABLE << 24) | /* PAT3: UC */
+ ((uint64_t)PAT_TYPE_WRBACK << 32) | /* PAT4: WB */
+ ((uint64_t)PAT_TYPE_WRTHROUGH << 40) | /* PAT5: WT */
+ ((uint64_t)PAT_TYPE_UC_MINUS << 48) | /* PAT6: UC- */
+ ((uint64_t)PAT_TYPE_UNCACHABLE << 56); /* PAT7: UC */
+
+ return 0;
+}
+
+void hvm_vcpu_cacheattr_destroy(struct vcpu *v)
+{
+ xfree(v->arch.hvm_vcpu.mtrr.var_ranges);
}
/*
@@ -512,23 +364,6 @@ static uint8_t effective_mm_type(struct
return effective;
}
-static void init_mtrr_epat_tbl(void)
-{
- int32_t i, j;
- /* set default value to an invalid type, just for checking conflict */
- memset(&mtrr_epat_tbl, INVALID_MEM_TYPE, sizeof(mtrr_epat_tbl));
-
- for ( i = 0; i < MTRR_NUM_TYPES; i++ )
- {
- for ( j = 0; j < PAT_TYPE_NUMS; j++ )
- {
- int32_t tmp = mm_type_tbl[i][j];
- if ( (tmp >= 0) && (tmp < MEMORY_NUM_TYPES) )
- mtrr_epat_tbl[i][tmp] = j;
- }
- }
-}
-
uint32_t get_pat_flags(struct vcpu *v,
uint32_t gl1e_flags,
paddr_t gpaddr,
@@ -856,7 +691,6 @@ static int hvm_load_mtrr_msr(struct doma
mtrr_def_type_msr_set(mtrr_state, hw_mtrr.msr_mtrr_def_type);
- v->arch.hvm_vcpu.mtrr.is_initialized = 1;
return 0;
}
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/svm/Makefile
--- a/xen/arch/x86/hvm/svm/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/svm/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -1,8 +1,6 @@ subdir-$(x86_32) += x86_32
-subdir-$(x86_32) += x86_32
-subdir-$(x86_64) += x86_64
-
obj-y += asid.o
obj-y += emulate.o
+obj-y += entry.o
obj-y += intr.o
obj-y += svm.o
obj-y += vmcb.o
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/svm/entry.S
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/xen/arch/x86/hvm/svm/entry.S Thu Apr 24 14:08:29 2008 -0600
@@ -0,0 +1,178 @@
+/*
+ * entry.S: SVM architecture-specific entry/exit handling.
+ * Copyright (c) 2005-2007, Advanced Micro Devices, Inc.
+ * Copyright (c) 2004, Intel Corporation.
+ * Copyright (c) 2008, Citrix Systems, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
+ * Place - Suite 330, Boston, MA 02111-1307 USA.
+ */
+
+#include <xen/config.h>
+#include <xen/errno.h>
+#include <xen/softirq.h>
+#include <asm/types.h>
+#include <asm/asm_defns.h>
+#include <asm/apicdef.h>
+#include <asm/page.h>
+#include <public/xen.h>
+
+#define VMRUN .byte 0x0F,0x01,0xD8
+#define STGI .byte 0x0F,0x01,0xDC
+#define CLGI .byte 0x0F,0x01,0xDD
+
+#define get_current(reg) \
+ mov $STACK_SIZE-BYTES_PER_LONG, r(reg); \
+ or r(sp), r(reg); \
+ and $~(BYTES_PER_LONG-1),r(reg); \
+ mov (r(reg)),r(reg);
+
+#if defined(__x86_64__)
+#define r(reg) %r##reg
+#define addr_of(lbl) lbl(%rip)
+#define call_with_regs(fn) \
+ mov %rsp,%rdi; \
+ call fn;
+#else /* defined(__i386__) */
+#define r(reg) %e##reg
+#define addr_of(lbl) lbl
+#define UREGS_rax UREGS_eax
+#define UREGS_rip UREGS_eip
+#define UREGS_rsp UREGS_esp
+#define call_with_regs(fn) \
+ mov %esp,%eax; \
+ push %eax; \
+ call fn; \
+ add $4,%esp;
+#endif
+
+ENTRY(svm_asm_do_resume)
+ get_current(bx)
+ CLGI
+
+ mov VCPU_processor(r(bx)),%eax
+ shl $IRQSTAT_shift,r(ax)
+ lea addr_of(irq_stat),r(dx)
+ testl $~0,(r(dx),r(ax),1)
+ jnz .Lsvm_process_softirqs
+
+ call svm_asid_handle_vmrun
+ call svm_intr_assist
+
+ cmpb $0,addr_of(tb_init_done)
+ jnz .Lsvm_trace
+.Lsvm_trace_done:
+
+ mov VCPU_svm_vmcb(r(bx)),r(cx)
+ mov UREGS_rax(r(sp)),r(ax)
+ mov r(ax),VMCB_rax(r(cx))
+ mov UREGS_rip(r(sp)),r(ax)
+ mov r(ax),VMCB_rip(r(cx))
+ mov UREGS_rsp(r(sp)),r(ax)
+ mov r(ax),VMCB_rsp(r(cx))
+ mov UREGS_eflags(r(sp)),r(ax)
+ mov r(ax),VMCB_rflags(r(cx))
+
+ mov VCPU_svm_vmcb_pa(r(bx)),r(ax)
+
+#if defined(__x86_64__)
+ pop %r15
+ pop %r14
+ pop %r13
+ pop %r12
+ pop %rbp
+ pop %rbx
+ pop %r11
+ pop %r10
+ pop %r9
+ pop %r8
+ add $8,%rsp /* Skip %rax: restored by VMRUN. */
+ pop %rcx
+ pop %rdx
+ pop %rsi
+ pop %rdi
+#else /* defined(__i386__) */
+ pop %ebx
+ pop %ecx
+ pop %edx
+ pop %esi
+ pop %edi
+ pop %ebp
+#endif
+
+ VMRUN
+
+#if defined(__x86_64__)
+ push %rdi
+ push %rsi
+ push %rdx
+ push %rcx
+ push %rax
+ push %r8
+ push %r9
+ push %r10
+ push %r11
+ push %rbx
+ push %rbp
+ push %r12
+ push %r13
+ push %r14
+ push %r15
+#else /* defined(__i386__) */
+ push %ebp
+ push %edi
+ push %esi
+ push %edx
+ push %ecx
+ push %ebx
+#endif
+
+ get_current(bx)
+ movb $0,VCPU_svm_vmcb_in_sync(r(bx))
+ mov VCPU_svm_vmcb(r(bx)),r(cx)
+ mov VMCB_rax(r(cx)),r(ax)
+ mov r(ax),UREGS_rax(r(sp))
+ mov VMCB_rip(r(cx)),r(ax)
+ mov r(ax),UREGS_rip(r(sp))
+ mov VMCB_rsp(r(cx)),r(ax)
+ mov r(ax),UREGS_rsp(r(sp))
+ mov VMCB_rflags(r(cx)),r(ax)
+ mov r(ax),UREGS_eflags(r(sp))
+
+#ifndef NDEBUG
+ mov $0xbeef,%ax
+ mov %ax,UREGS_error_code(r(sp))
+ mov %ax,UREGS_entry_vector(r(sp))
+ mov %ax,UREGS_saved_upcall_mask(r(sp))
+ mov %ax,UREGS_cs(r(sp))
+ mov %ax,UREGS_ds(r(sp))
+ mov %ax,UREGS_es(r(sp))
+ mov %ax,UREGS_fs(r(sp))
+ mov %ax,UREGS_gs(r(sp))
+ mov %ax,UREGS_ss(r(sp))
+#endif
+
+ STGI
+.globl svm_stgi_label
+svm_stgi_label:
+ call_with_regs(svm_vmexit_handler)
+ jmp svm_asm_do_resume
+
+.Lsvm_process_softirqs:
+ STGI
+ call do_softirq
+ jmp svm_asm_do_resume
+
+.Lsvm_trace:
+ call svm_trace_vmentry
+ jmp .Lsvm_trace_done
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/svm/intr.c
--- a/xen/arch/x86/hvm/svm/intr.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/svm/intr.c Thu Apr 24 14:08:29 2008 -0600
@@ -102,15 +102,17 @@ static void svm_dirq_assist(struct vcpu
struct hvm_irq_dpci *hvm_irq_dpci = d->arch.hvm_domain.irq.dpci;
struct dev_intx_gsi_link *digl;
- if ( !amd_iommu_enabled || (v->vcpu_id != 0) || (hvm_irq_dpci == NULL) )
+ if ( !iommu_enabled || (v->vcpu_id != 0) || (hvm_irq_dpci == NULL) )
return;
for ( irq = find_first_bit(hvm_irq_dpci->dirq_mask, NR_IRQS);
irq < NR_IRQS;
irq = find_next_bit(hvm_irq_dpci->dirq_mask, NR_IRQS, irq + 1) )
{
+ if ( !test_and_clear_bit(irq, &hvm_irq_dpci->dirq_mask) )
+ continue;
+
stop_timer(&hvm_irq_dpci->hvm_timer[irq_to_vector(irq)]);
- clear_bit(irq, &hvm_irq_dpci->dirq_mask);
list_for_each_entry ( digl, &hvm_irq_dpci->mirq[irq].digl_list, list )
{
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/svm/svm.c
--- a/xen/arch/x86/hvm/svm/svm.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/svm/svm.c Thu Apr 24 14:08:29 2008 -0600
@@ -911,6 +911,9 @@ static void svm_cpuid_intercept(
__clear_bit(X86_FEATURE_PAE & 31, edx);
__clear_bit(X86_FEATURE_PSE36 & 31, edx);
+ /* We always support MTRR MSRs. */
+ *edx |= bitmaskof(X86_FEATURE_MTRR);
+
/* Filter all other features according to a whitelist. */
*ecx &= (bitmaskof(X86_FEATURE_LAHF_LM) |
bitmaskof(X86_FEATURE_ALTMOVCR) |
@@ -924,7 +927,9 @@ static void svm_cpuid_intercept(
bitmaskof(X86_FEATURE_SYSCALL) |
bitmaskof(X86_FEATURE_MP) |
bitmaskof(X86_FEATURE_MMXEXT) |
- bitmaskof(X86_FEATURE_FFXSR));
+ bitmaskof(X86_FEATURE_FFXSR) |
+ bitmaskof(X86_FEATURE_3DNOW) |
+ bitmaskof(X86_FEATURE_3DNOWEXT));
break;
case 0x80000007:
@@ -981,14 +986,6 @@ static int svm_msr_read_intercept(struct
switch ( ecx )
{
- case MSR_IA32_TSC:
- msr_content = hvm_get_guest_time(v);
- break;
-
- case MSR_IA32_APICBASE:
- msr_content = vcpu_vlapic(v)->hw.apic_base_msr;
- break;
-
case MSR_EFER:
msr_content = v->arch.hvm_vcpu.guest_efer;
break;
@@ -1013,18 +1010,6 @@ static int svm_msr_read_intercept(struct
case MSR_K8_VM_HSAVE_PA:
goto gpf;
-
- case MSR_IA32_MCG_CAP:
- case MSR_IA32_MCG_STATUS:
- case MSR_IA32_MC0_STATUS:
- case MSR_IA32_MC1_STATUS:
- case MSR_IA32_MC2_STATUS:
- case MSR_IA32_MC3_STATUS:
- case MSR_IA32_MC4_STATUS:
- case MSR_IA32_MC5_STATUS:
- /* No point in letting the guest see real MCEs */
- msr_content = 0;
- break;
case MSR_IA32_DEBUGCTLMSR:
msr_content = vmcb->debugctlmsr;
@@ -1083,15 +1068,6 @@ static int svm_msr_write_intercept(struc
switch ( ecx )
{
- case MSR_IA32_TSC:
- hvm_set_guest_time(v, msr_content);
- pt_reset(v);
- break;
-
- case MSR_IA32_APICBASE:
- vlapic_msr_set(vcpu_vlapic(v), msr_content);
- break;
-
case MSR_K8_VM_HSAVE_PA:
goto gpf;
@@ -1152,12 +1128,12 @@ static void svm_do_msr_access(struct cpu
if ( vmcb->exitinfo1 == 0 )
{
- rc = svm_msr_read_intercept(regs);
+ rc = hvm_msr_read_intercept(regs);
inst_len = __get_instruction_length(v, INSTR_RDMSR, NULL);
}
else
{
- rc = svm_msr_write_intercept(regs);
+ rc = hvm_msr_write_intercept(regs);
inst_len = __get_instruction_length(v, INSTR_WRMSR, NULL);
}
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/svm/x86_32/Makefile
--- a/xen/arch/x86/hvm/svm/x86_32/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ /dev/null Thu Jan 01 00:00:00 1970 +0000
@@ -1,1 +0,0 @@
-obj-y += exits.o
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/svm/x86_32/exits.S
--- a/xen/arch/x86/hvm/svm/x86_32/exits.S Thu Apr 24 14:02:16 2008 -0600
+++ /dev/null Thu Jan 01 00:00:00 1970 +0000
@@ -1,131 +0,0 @@
-/*
- * exits.S: SVM architecture-specific exit handling.
- * Copyright (c) 2005-2007, Advanced Micro Devices, Inc.
- * Copyright (c) 2004, Intel Corporation.
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms and conditions of the GNU General Public License,
- * version 2, as published by the Free Software Foundation.
- *
- * This program is distributed in the hope it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
- * Place - Suite 330, Boston, MA 02111-1307 USA.
- */
-
-#include <xen/config.h>
-#include <xen/errno.h>
-#include <xen/softirq.h>
-#include <asm/asm_defns.h>
-#include <asm/apicdef.h>
-#include <asm/page.h>
-#include <public/xen.h>
-
-#define GET_CURRENT(reg) \
- movl $STACK_SIZE-4,reg; \
- orl %esp,reg; \
- andl $~3,reg; \
- movl (reg),reg;
-
-#define VMRUN .byte 0x0F,0x01,0xD8
-#define STGI .byte 0x0F,0x01,0xDC
-#define CLGI .byte 0x0F,0x01,0xDD
-
-ENTRY(svm_asm_do_resume)
- GET_CURRENT(%ebx)
- CLGI
-
- movl VCPU_processor(%ebx),%eax
- shl $IRQSTAT_shift,%eax
- testl $~0,irq_stat(%eax,1)
- jnz svm_process_softirqs
-
- call svm_asid_handle_vmrun
- call svm_intr_assist
-
- /* Check if the trace buffer is initialized.
- * Because the below condition is unlikely, we jump out of line
- * instead of having a mostly taken branch over the unlikely code.
- */
- cmpb $0,tb_init_done
- jnz svm_trace
-svm_trace_done:
-
- movl VCPU_svm_vmcb(%ebx),%ecx
- movl UREGS_eax(%esp),%eax
- movl %eax,VMCB_rax(%ecx)
- movl UREGS_eip(%esp),%eax
- movl %eax,VMCB_rip(%ecx)
- movl UREGS_esp(%esp),%eax
- movl %eax,VMCB_rsp(%ecx)
- movl UREGS_eflags(%esp),%eax
- movl %eax,VMCB_rflags(%ecx)
-
- movl VCPU_svm_vmcb_pa(%ebx),%eax
- popl %ebx
- popl %ecx
- popl %edx
- popl %esi
- popl %edi
- popl %ebp
-
- VMRUN
-
- pushl %ebp
- pushl %edi
- pushl %esi
- pushl %edx
- pushl %ecx
- pushl %ebx
-
- GET_CURRENT(%ebx)
- movb $0,VCPU_svm_vmcb_in_sync(%ebx)
- movl VCPU_svm_vmcb(%ebx),%ecx
- movl VMCB_rax(%ecx),%eax
- movl %eax,UREGS_eax(%esp)
- movl VMCB_rip(%ecx),%eax
- movl %eax,UREGS_eip(%esp)
- movl VMCB_rsp(%ecx),%eax
- movl %eax,UREGS_esp(%esp)
- movl VMCB_rflags(%ecx),%eax
- movl %eax,UREGS_eflags(%esp)
-
-#ifndef NDEBUG
- movw $0xbeef,%ax
- movw %ax,UREGS_error_code(%esp)
- movw %ax,UREGS_entry_vector(%esp)
- movw %ax,UREGS_saved_upcall_mask(%esp)
- movw %ax,UREGS_cs(%esp)
- movw %ax,UREGS_ds(%esp)
- movw %ax,UREGS_es(%esp)
- movw %ax,UREGS_fs(%esp)
- movw %ax,UREGS_gs(%esp)
- movw %ax,UREGS_ss(%esp)
-#endif
-
- STGI
-.globl svm_stgi_label;
-svm_stgi_label:
- movl %esp,%eax
- push %eax
- call svm_vmexit_handler
- addl $4,%esp
- jmp svm_asm_do_resume
-
- ALIGN
-svm_process_softirqs:
- STGI
- call do_softirq
- jmp svm_asm_do_resume
-
-svm_trace:
- /* Call out to C, as this is not speed critical path
- * Note: svm_trace_vmentry will recheck the tb_init_done,
- * but this is on the slow path, so who cares
- */
- call svm_trace_vmentry
- jmp svm_trace_done
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/svm/x86_64/Makefile
--- a/xen/arch/x86/hvm/svm/x86_64/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ /dev/null Thu Jan 01 00:00:00 1970 +0000
@@ -1,1 +0,0 @@
-obj-y += exits.o
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/svm/x86_64/exits.S
--- a/xen/arch/x86/hvm/svm/x86_64/exits.S Thu Apr 24 14:02:16 2008 -0600
+++ /dev/null Thu Jan 01 00:00:00 1970 +0000
@@ -1,148 +0,0 @@
-/*
- * exits.S: AMD-V architecture-specific exit handling.
- * Copyright (c) 2005-2007, Advanced Micro Devices, Inc.
- * Copyright (c) 2004, Intel Corporation.
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms and conditions of the GNU General Public License,
- * version 2, as published by the Free Software Foundation.
- *
- * This program is distributed in the hope it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
- * Place - Suite 330, Boston, MA 02111-1307 USA.
- */
-
-#include <xen/config.h>
-#include <xen/errno.h>
-#include <xen/softirq.h>
-#include <asm/asm_defns.h>
-#include <asm/apicdef.h>
-#include <asm/page.h>
-#include <public/xen.h>
-
-#define GET_CURRENT(reg) \
- movq $STACK_SIZE-8,reg; \
- orq %rsp,reg; \
- andq $~7,reg; \
- movq (reg),reg;
-
-#define VMRUN .byte 0x0F,0x01,0xD8
-#define STGI .byte 0x0F,0x01,0xDC
-#define CLGI .byte 0x0F,0x01,0xDD
-
-ENTRY(svm_asm_do_resume)
- GET_CURRENT(%rbx)
- CLGI
-
- movl VCPU_processor(%rbx),%eax
- shl $IRQSTAT_shift,%rax
- leaq irq_stat(%rip),%rdx
- testl $~0,(%rdx,%rax,1)
- jnz svm_process_softirqs
-
- call svm_asid_handle_vmrun
- call svm_intr_assist
-
- /* Check if the trace buffer is initialized.
- * Because the below condition is unlikely, we jump out of line
- * instead of having a mostly taken branch over the unlikely code.
- */
- cmpb $0,tb_init_done(%rip)
- jnz svm_trace
-svm_trace_done:
-
- movq VCPU_svm_vmcb(%rbx),%rcx
- movq UREGS_rax(%rsp),%rax
- movq %rax,VMCB_rax(%rcx)
- movq UREGS_rip(%rsp),%rax
- movq %rax,VMCB_rip(%rcx)
- movq UREGS_rsp(%rsp),%rax
- movq %rax,VMCB_rsp(%rcx)
- movq UREGS_eflags(%rsp),%rax
- movq %rax,VMCB_rflags(%rcx)
-
- movq VCPU_svm_vmcb_pa(%rbx),%rax
- popq %r15
- popq %r14
- popq %r13
- popq %r12
- popq %rbp
- popq %rbx
- popq %r11
- popq %r10
- popq %r9
- popq %r8
- addq $8,%rsp /* Skip %rax: restored by VMRUN. */
- popq %rcx
- popq %rdx
- popq %rsi
- popq %rdi
-
- VMRUN
-
- pushq %rdi
- pushq %rsi
- pushq %rdx
- pushq %rcx
- pushq %rax
- pushq %r8
- pushq %r9
- pushq %r10
- pushq %r11
- pushq %rbx
- pushq %rbp
- pushq %r12
- pushq %r13
- pushq %r14
- pushq %r15
-
- GET_CURRENT(%rbx)
- movb $0,VCPU_svm_vmcb_in_sync(%rbx)
- movq VCPU_svm_vmcb(%rbx),%rcx
- movq VMCB_rax(%rcx),%rax
- movq %rax,UREGS_rax(%rsp)
- movq VMCB_rip(%rcx),%rax
- movq %rax,UREGS_rip(%rsp)
- movq VMCB_rsp(%rcx),%rax
- movq %rax,UREGS_rsp(%rsp)
- movq VMCB_rflags(%rcx),%rax
- movq %rax,UREGS_eflags(%rsp)
-
-#ifndef NDEBUG
- movw $0xbeef,%ax
- movw %ax,UREGS_error_code(%rsp)
- movw %ax,UREGS_entry_vector(%rsp)
- movw %ax,UREGS_saved_upcall_mask(%rsp)
- movw %ax,UREGS_cs(%rsp)
- movw %ax,UREGS_ds(%rsp)
- movw %ax,UREGS_es(%rsp)
- movw %ax,UREGS_fs(%rsp)
- movw %ax,UREGS_gs(%rsp)
- movw %ax,UREGS_ss(%rsp)
-#endif
-
- STGI
-.globl svm_stgi_label;
-svm_stgi_label:
- movq %rsp,%rdi
- call svm_vmexit_handler
- jmp svm_asm_do_resume
-
- ALIGN
-svm_process_softirqs:
- STGI
- call do_softirq
- jmp svm_asm_do_resume
-
-svm_trace:
- /* Call out to C, as this is not speed critical path
- * Note: svm_trace_vmentry will recheck the tb_init_done,
- * but this is on the slow path, so who cares
- */
- call svm_trace_vmentry
- jmp svm_trace_done
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/vmx/Makefile
--- a/xen/arch/x86/hvm/vmx/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/vmx/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -1,6 +1,4 @@ subdir-$(x86_32) += x86_32
-subdir-$(x86_32) += x86_32
-subdir-$(x86_64) += x86_64
-
+obj-y += entry.o
obj-y += intr.o
obj-y += realmode.o
obj-y += vmcs.o
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/vmx/entry.S
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/xen/arch/x86/hvm/vmx/entry.S Thu Apr 24 14:08:29 2008 -0600
@@ -0,0 +1,198 @@
+/*
+ * entry.S: VMX architecture-specific entry/exit handling.
+ * Copyright (c) 2004, Intel Corporation.
+ * Copyright (c) 2008, Citrix Systems, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
+ * Place - Suite 330, Boston, MA 02111-1307 USA.
+ */
+
+#include <xen/config.h>
+#include <xen/errno.h>
+#include <xen/softirq.h>
+#include <asm/types.h>
+#include <asm/asm_defns.h>
+#include <asm/apicdef.h>
+#include <asm/page.h>
+#include <public/xen.h>
+
+#define VMRESUME .byte 0x0f,0x01,0xc3
+#define VMLAUNCH .byte 0x0f,0x01,0xc2
+#define VMREAD(off) .byte 0x0f,0x78,0x47,((off)-UREGS_rip)
+#define VMWRITE(off) .byte 0x0f,0x79,0x47,((off)-UREGS_rip)
+
+/* VMCS field encodings */
+#define GUEST_RSP 0x681c
+#define GUEST_RIP 0x681e
+#define GUEST_RFLAGS 0x6820
+
+#define get_current(reg) \
+ mov $STACK_SIZE-BYTES_PER_LONG, r(reg); \
+ or r(sp), r(reg); \
+ and $~(BYTES_PER_LONG-1),r(reg); \
+ mov (r(reg)),r(reg);
+
+#if defined(__x86_64__)
+#define r(reg) %r##reg
+#define addr_of(lbl) lbl(%rip)
+#define call_with_regs(fn) \
+ mov %rsp,%rdi; \
+ call fn;
+#else /* defined(__i386__) */
+#define r(reg) %e##reg
+#define addr_of(lbl) lbl
+#define UREGS_rip UREGS_eip
+#define UREGS_rsp UREGS_esp
+#define call_with_regs(fn) \
+ mov %esp,%eax; \
+ push %eax; \
+ call fn; \
+ add $4,%esp;
+#endif
+
+ ALIGN
+.globl vmx_asm_vmexit_handler
+vmx_asm_vmexit_handler:
+#if defined(__x86_64__)
+ push %rdi
+ push %rsi
+ push %rdx
+ push %rcx
+ push %rax
+ push %r8
+ push %r9
+ push %r10
+ push %r11
+ push %rbx
+ push %rbp
+ push %r12
+ push %r13
+ push %r14
+ push %r15
+#else /* defined(__i386__) */
+ push %eax
+ push %ebp
+ push %edi
+ push %esi
+ push %edx
+ push %ecx
+ push %ebx
+#endif
+
+ get_current(bx)
+
+ movb $1,VCPU_vmx_launched(r(bx))
+
+ lea UREGS_rip(r(sp)),r(di)
+ mov $GUEST_RIP,%eax
+ /*VMREAD(UREGS_rip)*/
+ .byte 0x0f,0x78,0x07 /* vmread r(ax),(r(di)) */
+ mov $GUEST_RSP,%eax
+ VMREAD(UREGS_rsp)
+ mov $GUEST_RFLAGS,%eax
+ VMREAD(UREGS_eflags)
+
+ mov %cr2,r(ax)
+ mov r(ax),VCPU_hvm_guest_cr2(r(bx))
+
+#ifndef NDEBUG
+ mov $0xbeef,%ax
+ mov %ax,UREGS_error_code(r(sp))
+ mov %ax,UREGS_entry_vector(r(sp))
+ mov %ax,UREGS_saved_upcall_mask(r(sp))
+ mov %ax,UREGS_cs(r(sp))
+ mov %ax,UREGS_ds(r(sp))
+ mov %ax,UREGS_es(r(sp))
+ mov %ax,UREGS_fs(r(sp))
+ mov %ax,UREGS_gs(r(sp))
+ mov %ax,UREGS_ss(r(sp))
+#endif
+
+ call_with_regs(vmx_vmexit_handler)
+
+.globl vmx_asm_do_vmentry
+vmx_asm_do_vmentry:
+ get_current(bx)
+ cli
+
+ mov VCPU_processor(r(bx)),%eax
+ shl $IRQSTAT_shift,r(ax)
+ lea addr_of(irq_stat),r(dx)
+ cmpl $0,(r(dx),r(ax),1)
+ jnz .Lvmx_process_softirqs
+
+ call vmx_intr_assist
+
+ testb $0xff,VCPU_vmx_emul(r(bx))
+ jnz .Lvmx_goto_realmode
+
+ mov VCPU_hvm_guest_cr2(r(bx)),r(ax)
+ mov r(ax),%cr2
+ call vmx_trace_vmentry
+
+ lea UREGS_rip(r(sp)),r(di)
+ mov $GUEST_RIP,%eax
+ /*VMWRITE(UREGS_rip)*/
+ .byte 0x0f,0x79,0x07 /* vmwrite (r(di)),r(ax) */
+ mov $GUEST_RSP,%eax
+ VMWRITE(UREGS_rsp)
+ mov $GUEST_RFLAGS,%eax
+ VMWRITE(UREGS_eflags)
+
+ cmpb $0,VCPU_vmx_launched(r(bx))
+#if defined(__x86_64__)
+ pop %r15
+ pop %r14
+ pop %r13
+ pop %r12
+ pop %rbp
+ pop %rbx
+ pop %r11
+ pop %r10
+ pop %r9
+ pop %r8
+ pop %rax
+ pop %rcx
+ pop %rdx
+ pop %rsi
+ pop %rdi
+#else /* defined(__i386__) */
+ pop %ebx
+ pop %ecx
+ pop %edx
+ pop %esi
+ pop %edi
+ pop %ebp
+ pop %eax
+#endif
+ je .Lvmx_launch
+
+/*.Lvmx_resume:*/
+ VMRESUME
+ call vm_resume_fail
+ ud2
+
+.Lvmx_launch:
+ VMLAUNCH
+ call vm_launch_fail
+ ud2
+
+.Lvmx_goto_realmode:
+ sti
+ call_with_regs(vmx_realmode)
+ jmp vmx_asm_do_vmentry
+
+.Lvmx_process_softirqs:
+ sti
+ call do_softirq
+ jmp vmx_asm_do_vmentry
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/vmx/intr.c
--- a/xen/arch/x86/hvm/vmx/intr.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/vmx/intr.c Thu Apr 24 14:08:29 2008 -0600
@@ -111,15 +111,17 @@ static void vmx_dirq_assist(struct vcpu
struct hvm_irq_dpci *hvm_irq_dpci = d->arch.hvm_domain.irq.dpci;
struct dev_intx_gsi_link *digl;
- if ( !vtd_enabled || (v->vcpu_id != 0) || (hvm_irq_dpci == NULL) )
+ if ( !iommu_enabled || (v->vcpu_id != 0) || (hvm_irq_dpci == NULL) )
return;
for ( irq = find_first_bit(hvm_irq_dpci->dirq_mask, NR_IRQS);
irq < NR_IRQS;
irq = find_next_bit(hvm_irq_dpci->dirq_mask, NR_IRQS, irq + 1) )
{
+ if ( !test_and_clear_bit(irq, &hvm_irq_dpci->dirq_mask) )
+ continue;
+
stop_timer(&hvm_irq_dpci->hvm_timer[irq_to_vector(irq)]);
- clear_bit(irq, &hvm_irq_dpci->dirq_mask);
list_for_each_entry ( digl, &hvm_irq_dpci->mirq[irq].digl_list, list )
{
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/vmx/vmx.c
--- a/xen/arch/x86/hvm/vmx/vmx.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/hvm/vmx/vmx.c Thu Apr 24 14:08:29 2008 -0600
@@ -1622,17 +1622,11 @@ static int vmx_msr_read_intercept(struct
u64 msr_content = 0;
u32 ecx = regs->ecx, eax, edx;
struct vcpu *v = current;
- int index;
- u64 *var_range_base = (u64*)v->arch.hvm_vcpu.mtrr.var_ranges;
- u64 *fixed_range_base = (u64*)v->arch.hvm_vcpu.mtrr.fixed_ranges;
HVM_DBG_LOG(DBG_LEVEL_1, "ecx=%x", ecx);
switch ( ecx )
{
- case MSR_IA32_TSC:
- msr_content = hvm_get_guest_time(v);
- break;
case MSR_IA32_SYSENTER_CS:
msr_content = (u32)__vmread(GUEST_SYSENTER_CS);
break;
@@ -1641,35 +1635,6 @@ static int vmx_msr_read_intercept(struct
break;
case MSR_IA32_SYSENTER_EIP:
msr_content = __vmread(GUEST_SYSENTER_EIP);
- break;
- case MSR_IA32_APICBASE:
- msr_content = vcpu_vlapic(v)->hw.apic_base_msr;
- break;
- case MSR_IA32_CR_PAT:
- msr_content = v->arch.hvm_vcpu.pat_cr;
- break;
- case MSR_MTRRcap:
- msr_content = v->arch.hvm_vcpu.mtrr.mtrr_cap;
- break;
- case MSR_MTRRdefType:
- msr_content = v->arch.hvm_vcpu.mtrr.def_type
- | (v->arch.hvm_vcpu.mtrr.enabled << 10);
- break;
- case MSR_MTRRfix64K_00000:
- msr_content = fixed_range_base[0];
- break;
- case MSR_MTRRfix16K_80000:
- case MSR_MTRRfix16K_A0000:
- index = regs->ecx - MSR_MTRRfix16K_80000;
- msr_content = fixed_range_base[index + 1];
- break;
- case MSR_MTRRfix4K_C0000...MSR_MTRRfix4K_F8000:
- index = regs->ecx - MSR_MTRRfix4K_C0000;
- msr_content = fixed_range_base[index + 3];
- break;
- case MSR_IA32_MTRR_PHYSBASE0...MSR_IA32_MTRR_PHYSMASK7:
- index = regs->ecx - MSR_IA32_MTRR_PHYSBASE0;
- msr_content = var_range_base[index];
break;
case MSR_IA32_DEBUGCTLMSR:
msr_content = __vmread(GUEST_IA32_DEBUGCTL);
@@ -1679,17 +1644,6 @@ static int vmx_msr_read_intercept(struct
break;
case MSR_IA32_VMX_BASIC...MSR_IA32_VMX_PROCBASED_CTLS2:
goto gp_fault;
- case MSR_IA32_MCG_CAP:
- case MSR_IA32_MCG_STATUS:
- case MSR_IA32_MC0_STATUS:
- case MSR_IA32_MC1_STATUS:
- case MSR_IA32_MC2_STATUS:
- case MSR_IA32_MC3_STATUS:
- case MSR_IA32_MC4_STATUS:
- case MSR_IA32_MC5_STATUS:
- /* No point in letting the guest see real MCEs */
- msr_content = 0;
- break;
case MSR_IA32_MISC_ENABLE:
rdmsrl(MSR_IA32_MISC_ENABLE, msr_content);
/* Debug Trace Store is not supported. */
@@ -1729,8 +1683,8 @@ static int vmx_msr_read_intercept(struct
goto gp_fault;
}
- regs->eax = msr_content & 0xFFFFFFFF;
- regs->edx = msr_content >> 32;
+ regs->eax = (uint32_t)msr_content;
+ regs->edx = (uint32_t)(msr_content >> 32);
done:
hvmtrace_msr_read(v, ecx, msr_content);
@@ -1833,19 +1787,11 @@ void vmx_vlapic_msr_changed(struct vcpu
vmx_vmcs_exit(v);
}
-extern bool_t mtrr_var_range_msr_set(struct mtrr_state *v,
- u32 msr, u64 msr_content);
-extern bool_t mtrr_fix_range_msr_set(struct mtrr_state *v,
- int row, u64 msr_content);
-extern bool_t mtrr_def_type_msr_set(struct mtrr_state *v, u64 msr_content);
-extern bool_t pat_msr_set(u64 *pat, u64 msr);
-
static int vmx_msr_write_intercept(struct cpu_user_regs *regs)
{
u32 ecx = regs->ecx;
u64 msr_content;
struct vcpu *v = current;
- int index;
HVM_DBG_LOG(DBG_LEVEL_1, "ecx=%x, eax=%x, edx=%x",
ecx, (u32)regs->eax, (u32)regs->edx);
@@ -1856,10 +1802,6 @@ static int vmx_msr_write_intercept(struc
switch ( ecx )
{
- case MSR_IA32_TSC:
- hvm_set_guest_time(v, msr_content);
- pt_reset(v);
- break;
case MSR_IA32_SYSENTER_CS:
__vmwrite(GUEST_SYSENTER_CS, msr_content);
break;
@@ -1869,41 +1811,6 @@ static int vmx_msr_write_intercept(struc
case MSR_IA32_SYSENTER_EIP:
__vmwrite(GUEST_SYSENTER_EIP, msr_content);
break;
- case MSR_IA32_APICBASE:
- vlapic_msr_set(vcpu_vlapic(v), msr_content);
- break;
- case MSR_IA32_CR_PAT:
- if ( !pat_msr_set(&v->arch.hvm_vcpu.pat_cr, msr_content) )
- goto gp_fault;
- break;
- case MSR_MTRRdefType:
- if ( !mtrr_def_type_msr_set(&v->arch.hvm_vcpu.mtrr, msr_content) )
- goto gp_fault;
- break;
- case MSR_MTRRfix64K_00000:
- if ( !mtrr_fix_range_msr_set(&v->arch.hvm_vcpu.mtrr, 0, msr_content) )
- goto gp_fault;
- break;
- case MSR_MTRRfix16K_80000:
- case MSR_MTRRfix16K_A0000:
- index = regs->ecx - MSR_MTRRfix16K_80000 + 1;
- if ( !mtrr_fix_range_msr_set(&v->arch.hvm_vcpu.mtrr,
- index, msr_content) )
- goto gp_fault;
- break;
- case MSR_MTRRfix4K_C0000...MSR_MTRRfix4K_F8000:
- index = regs->ecx - MSR_MTRRfix4K_C0000 + 3;
- if ( !mtrr_fix_range_msr_set(&v->arch.hvm_vcpu.mtrr,
- index, msr_content) )
- goto gp_fault;
- break;
- case MSR_IA32_MTRR_PHYSBASE0...MSR_IA32_MTRR_PHYSMASK7:
- if ( !mtrr_var_range_msr_set(&v->arch.hvm_vcpu.mtrr,
- regs->ecx, msr_content) )
- goto gp_fault;
- break;
- case MSR_MTRRcap:
- goto gp_fault;
case MSR_IA32_DEBUGCTLMSR: {
int i, rc = 0;
@@ -2330,12 +2237,12 @@ asmlinkage void vmx_vmexit_handler(struc
break;
case EXIT_REASON_MSR_READ:
inst_len = __get_instruction_length(); /* Safe: RDMSR */
- if ( vmx_msr_read_intercept(regs) == X86EMUL_OKAY )
+ if ( hvm_msr_read_intercept(regs) == X86EMUL_OKAY )
__update_guest_eip(inst_len);
break;
case EXIT_REASON_MSR_WRITE:
inst_len = __get_instruction_length(); /* Safe: WRMSR */
- if ( vmx_msr_write_intercept(regs) == X86EMUL_OKAY )
+ if ( hvm_msr_write_intercept(regs) == X86EMUL_OKAY )
__update_guest_eip(inst_len);
break;
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/vmx/x86_32/Makefile
--- a/xen/arch/x86/hvm/vmx/x86_32/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ /dev/null Thu Jan 01 00:00:00 1970 +0000
@@ -1,1 +0,0 @@
-obj-y += exits.o
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/vmx/x86_32/exits.S
--- a/xen/arch/x86/hvm/vmx/x86_32/exits.S Thu Apr 24 14:02:16 2008 -0600
+++ /dev/null Thu Jan 01 00:00:00 1970 +0000
@@ -1,148 +0,0 @@
-/*
- * exits.S: VMX architecture-specific exit handling.
- * Copyright (c) 2004, Intel Corporation.
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms and conditions of the GNU General Public License,
- * version 2, as published by the Free Software Foundation.
- *
- * This program is distributed in the hope it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
- * Place - Suite 330, Boston, MA 02111-1307 USA.
- */
-#include <xen/config.h>
-#include <xen/errno.h>
-#include <xen/softirq.h>
-#include <asm/asm_defns.h>
-#include <asm/apicdef.h>
-#include <asm/page.h>
-#include <public/xen.h>
-
-#define VMRESUME .byte 0x0f,0x01,0xc3
-#define VMLAUNCH .byte 0x0f,0x01,0xc2
-#define VMREAD(off) .byte 0x0f,0x78,0x44,0x24,off
-#define VMWRITE(off) .byte 0x0f,0x79,0x44,0x24,off
-
-/* VMCS field encodings */
-#define GUEST_RSP 0x681c
-#define GUEST_RIP 0x681e
-#define GUEST_RFLAGS 0x6820
-
-#define GET_CURRENT(reg) \
- movl $STACK_SIZE-4, reg; \
- orl %esp, reg; \
- andl $~3,reg; \
- movl (reg),reg;
-
-#define HVM_SAVE_ALL_NOSEGREGS \
- pushl %eax; \
- pushl %ebp; \
- pushl %edi; \
- pushl %esi; \
- pushl %edx; \
- pushl %ecx; \
- pushl %ebx;
-
-#define HVM_RESTORE_ALL_NOSEGREGS \
- popl %ebx; \
- popl %ecx; \
- popl %edx; \
- popl %esi; \
- popl %edi; \
- popl %ebp; \
- popl %eax
-
- ALIGN
-ENTRY(vmx_asm_vmexit_handler)
- HVM_SAVE_ALL_NOSEGREGS
- GET_CURRENT(%ebx)
-
- movl $GUEST_RIP,%eax
- VMREAD(UREGS_eip)
- movl $GUEST_RSP,%eax
- VMREAD(UREGS_esp)
- movl $GUEST_RFLAGS,%eax
- VMREAD(UREGS_eflags)
-
- movl %cr2,%eax
- movl %eax,VCPU_hvm_guest_cr2(%ebx)
-
-#ifndef NDEBUG
- movw $0xbeef,%ax
- movw %ax,UREGS_error_code(%esp)
- movw %ax,UREGS_entry_vector(%esp)
- movw %ax,UREGS_saved_upcall_mask(%esp)
- movw %ax,UREGS_cs(%esp)
- movw %ax,UREGS_ds(%esp)
- movw %ax,UREGS_es(%esp)
- movw %ax,UREGS_fs(%esp)
- movw %ax,UREGS_gs(%esp)
- movw %ax,UREGS_ss(%esp)
-#endif
-
- movl %esp,%eax
- push %eax
- call vmx_vmexit_handler
- addl $4,%esp
- jmp vmx_asm_do_vmentry
-
- ALIGN
-vmx_process_softirqs:
- sti
- call do_softirq
- jmp vmx_asm_do_vmentry
-
- ALIGN
-ENTRY(vmx_asm_do_vmentry)
- GET_CURRENT(%ebx)
- cli # tests must not race interrupts
-
- movl VCPU_processor(%ebx),%eax
- shl $IRQSTAT_shift,%eax
- cmpl $0,irq_stat(%eax,1)
- jnz vmx_process_softirqs
-
- call vmx_intr_assist
-
- testb $0xff,VCPU_vmx_emul(%ebx)
- jnz vmx_goto_realmode
-
- movl VCPU_hvm_guest_cr2(%ebx),%eax
- movl %eax,%cr2
- call vmx_trace_vmentry
-
- movl $GUEST_RIP,%eax
- VMWRITE(UREGS_eip)
- movl $GUEST_RSP,%eax
- VMWRITE(UREGS_esp)
- movl $GUEST_RFLAGS,%eax
- VMWRITE(UREGS_eflags)
-
- cmpb $0,VCPU_vmx_launched(%ebx)
- je vmx_launch
-
-/*vmx_resume:*/
- HVM_RESTORE_ALL_NOSEGREGS
- VMRESUME
- call vm_resume_fail
- ud2
-
-vmx_launch:
- movb $1,VCPU_vmx_launched(%ebx)
- HVM_RESTORE_ALL_NOSEGREGS
- VMLAUNCH
- call vm_launch_fail
- ud2
-
-vmx_goto_realmode:
- sti
- movl %esp,%eax
- push %eax
- call vmx_realmode
- addl $4,%esp
- jmp vmx_asm_do_vmentry
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/vmx/x86_64/Makefile
--- a/xen/arch/x86/hvm/vmx/x86_64/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ /dev/null Thu Jan 01 00:00:00 1970 +0000
@@ -1,1 +0,0 @@
-obj-y += exits.o
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/hvm/vmx/x86_64/exits.S
--- a/xen/arch/x86/hvm/vmx/x86_64/exits.S Thu Apr 24 14:02:16 2008 -0600
+++ /dev/null Thu Jan 01 00:00:00 1970 +0000
@@ -1,165 +0,0 @@
-/*
- * exits.S: VMX architecture-specific exit handling.
- * Copyright (c) 2004, Intel Corporation.
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms and conditions of the GNU General Public License,
- * version 2, as published by the Free Software Foundation.
- *
- * This program is distributed in the hope it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
- * Place - Suite 330, Boston, MA 02111-1307 USA.
- */
-#include <xen/config.h>
-#include <xen/errno.h>
-#include <xen/softirq.h>
-#include <asm/asm_defns.h>
-#include <asm/apicdef.h>
-#include <asm/page.h>
-#include <public/xen.h>
-
-#define VMRESUME .byte 0x0f,0x01,0xc3
-#define VMLAUNCH .byte 0x0f,0x01,0xc2
-#define VMREAD(off) .byte 0x0f,0x78,0x47,((off)-UREGS_rip)
-#define VMWRITE(off) .byte 0x0f,0x79,0x47,((off)-UREGS_rip)
-
-/* VMCS field encodings */
-#define GUEST_RSP 0x681c
-#define GUEST_RIP 0x681e
-#define GUEST_RFLAGS 0x6820
-
-#define GET_CURRENT(reg) \
- movq $STACK_SIZE-8, reg; \
- orq %rsp, reg; \
- andq $~7,reg; \
- movq (reg),reg;
-
-#define HVM_SAVE_ALL_NOSEGREGS \
- pushq %rdi; \
- pushq %rsi; \
- pushq %rdx; \
- pushq %rcx; \
- pushq %rax; \
- pushq %r8; \
- pushq %r9; \
- pushq %r10; \
- pushq %r11; \
- pushq %rbx; \
- pushq %rbp; \
- pushq %r12; \
- pushq %r13; \
- pushq %r14; \
- pushq %r15;
-
-#define HVM_RESTORE_ALL_NOSEGREGS \
- popq %r15; \
- popq %r14; \
- popq %r13; \
- popq %r12; \
- popq %rbp; \
- popq %rbx; \
- popq %r11; \
- popq %r10; \
- popq %r9; \
- popq %r8; \
- popq %rax; \
- popq %rcx; \
- popq %rdx; \
- popq %rsi; \
- popq %rdi
-
- ALIGN
-ENTRY(vmx_asm_vmexit_handler)
- HVM_SAVE_ALL_NOSEGREGS
- GET_CURRENT(%rbx)
-
- leaq UREGS_rip(%rsp),%rdi
- movl $GUEST_RIP,%eax
- /*VMREAD(UREGS_rip)*/
- .byte 0x0f,0x78,0x07 /* vmread %rax,(%rdi) */
- movl $GUEST_RSP,%eax
- VMREAD(UREGS_rsp)
- movl $GUEST_RFLAGS,%eax
- VMREAD(UREGS_eflags)
-
- movq %cr2,%rax
- movq %rax,VCPU_hvm_guest_cr2(%rbx)
-
-#ifndef NDEBUG
- movw $0xbeef,%ax
- movw %ax,UREGS_error_code(%rsp)
- movw %ax,UREGS_entry_vector(%rsp)
- movw %ax,UREGS_saved_upcall_mask(%rsp)
- movw %ax,UREGS_cs(%rsp)
- movw %ax,UREGS_ds(%rsp)
- movw %ax,UREGS_es(%rsp)
- movw %ax,UREGS_fs(%rsp)
- movw %ax,UREGS_gs(%rsp)
- movw %ax,UREGS_ss(%rsp)
-#endif
-
- movq %rsp,%rdi
- call vmx_vmexit_handler
- jmp vmx_asm_do_vmentry
-
- ALIGN
-vmx_process_softirqs:
- sti
- call do_softirq
- jmp vmx_asm_do_vmentry
-
- ALIGN
-ENTRY(vmx_asm_do_vmentry)
- GET_CURRENT(%rbx)
- cli # tests must not race interrupts
-
- movl VCPU_processor(%rbx),%eax
- shl $IRQSTAT_shift,%rax
- leaq irq_stat(%rip),%rdx
- cmpl $0,(%rdx,%rax,1)
- jnz vmx_process_softirqs
-
- call vmx_intr_assist
-
- testb $0xff,VCPU_vmx_emul(%rbx)
- jnz vmx_goto_realmode
-
- movq VCPU_hvm_guest_cr2(%rbx),%rax
- movq %rax,%cr2
- call vmx_trace_vmentry
-
- leaq UREGS_rip(%rsp),%rdi
- movl $GUEST_RIP,%eax
- /*VMWRITE(UREGS_rip)*/
- .byte 0x0f,0x79,0x07 /* vmwrite (%rdi),%rax */
- movl $GUEST_RSP,%eax
- VMWRITE(UREGS_rsp)
- movl $GUEST_RFLAGS,%eax
- VMWRITE(UREGS_eflags)
-
- cmpb $0,VCPU_vmx_launched(%rbx)
- je vmx_launch
-
-/*vmx_resume:*/
- HVM_RESTORE_ALL_NOSEGREGS
- VMRESUME
- call vm_resume_fail
- ud2
-
-vmx_launch:
- movb $1,VCPU_vmx_launched(%rbx)
- HVM_RESTORE_ALL_NOSEGREGS
- VMLAUNCH
- call vm_launch_fail
- ud2
-
-vmx_goto_realmode:
- sti
- movq %rsp,%rdi
- call vmx_realmode
- jmp vmx_asm_do_vmentry
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/mm.c
--- a/xen/arch/x86/mm.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/mm.c Thu Apr 24 14:08:29 2008 -0600
@@ -3279,15 +3279,6 @@ long arch_memory_op(int op, XEN_GUEST_HA
case XENMAPSPACE_shared_info:
if ( xatp.idx == 0 )
mfn = virt_to_mfn(d->shared_info);
- /* XXX: assumption here, this is called after E820 table is build
- * need the E820 to initialize MTRR.
- */
- if ( is_hvm_domain(d) ) {
- extern void init_mtrr_in_hyper(struct vcpu *);
- struct vcpu *vs;
- for_each_vcpu(d, vs)
- init_mtrr_in_hyper(vs);
- }
break;
case XENMAPSPACE_grant_table:
spin_lock(&d->grant_table->lock);
@@ -3625,29 +3616,18 @@ static int ptwr_emulated_cmpxchg(
static int ptwr_emulated_cmpxchg(
enum x86_segment seg,
unsigned long offset,
- unsigned long old,
- unsigned long new,
+ void *p_old,
+ void *p_new,
unsigned int bytes,
struct x86_emulate_ctxt *ctxt)
{
+ paddr_t old = 0, new = 0;
+ if ( bytes > sizeof(paddr_t) )
+ return X86EMUL_UNHANDLEABLE;
+ memcpy(&old, p_old, bytes);
+ memcpy(&new, p_new, bytes);
return ptwr_emulated_update(
offset, old, new, bytes, 1,
- container_of(ctxt, struct ptwr_emulate_ctxt, ctxt));
-}
-
-static int ptwr_emulated_cmpxchg8b(
- enum x86_segment seg,
- unsigned long offset,
- unsigned long old,
- unsigned long old_hi,
- unsigned long new,
- unsigned long new_hi,
- struct x86_emulate_ctxt *ctxt)
-{
- if ( CONFIG_PAGING_LEVELS == 2 )
- return X86EMUL_UNHANDLEABLE;
- return ptwr_emulated_update(
- offset, ((u64)old_hi << 32) | old, ((u64)new_hi << 32) | new, 8, 1,
container_of(ctxt, struct ptwr_emulate_ctxt, ctxt));
}
@@ -3656,7 +3636,6 @@ static struct x86_emulate_ops ptwr_emula
.insn_fetch = ptwr_emulated_read,
.write = ptwr_emulated_write,
.cmpxchg = ptwr_emulated_cmpxchg,
- .cmpxchg8b = ptwr_emulated_cmpxchg8b
};
/* Write page fault handler: check if guest is trying to modify a PTE. */
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/mm/shadow/common.c
--- a/xen/arch/x86/mm/shadow/common.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/mm/shadow/common.c Thu Apr 24 14:08:29 2008 -0600
@@ -239,15 +239,15 @@ static int
static int
hvm_emulate_cmpxchg(enum x86_segment seg,
unsigned long offset,
- unsigned long old,
- unsigned long new,
+ void *p_old,
+ void *p_new,
unsigned int bytes,
struct x86_emulate_ctxt *ctxt)
{
struct sh_emulate_ctxt *sh_ctxt =
container_of(ctxt, struct sh_emulate_ctxt, ctxt);
struct vcpu *v = current;
- unsigned long addr;
+ unsigned long addr, old[2], new[2];
int rc;
if ( !is_x86_user_segment(seg) )
@@ -258,35 +258,21 @@ hvm_emulate_cmpxchg(enum x86_segment seg
if ( rc )
return rc;
- return v->arch.paging.mode->shadow.x86_emulate_cmpxchg(
- v, addr, old, new, bytes, sh_ctxt);
-}
-
-static int
-hvm_emulate_cmpxchg8b(enum x86_segment seg,
- unsigned long offset,
- unsigned long old_lo,
- unsigned long old_hi,
- unsigned long new_lo,
- unsigned long new_hi,
- struct x86_emulate_ctxt *ctxt)
-{
- struct sh_emulate_ctxt *sh_ctxt =
- container_of(ctxt, struct sh_emulate_ctxt, ctxt);
- struct vcpu *v = current;
- unsigned long addr;
- int rc;
-
- if ( !is_x86_user_segment(seg) )
- return X86EMUL_UNHANDLEABLE;
-
- rc = hvm_translate_linear_addr(
- seg, offset, 8, hvm_access_write, sh_ctxt, &addr);
- if ( rc )
- return rc;
-
- return v->arch.paging.mode->shadow.x86_emulate_cmpxchg8b(
- v, addr, old_lo, old_hi, new_lo, new_hi, sh_ctxt);
+ old[0] = new[0] = 0;
+ memcpy(old, p_old, bytes);
+ memcpy(new, p_new, bytes);
+
+ if ( bytes <= sizeof(long) )
+ return v->arch.paging.mode->shadow.x86_emulate_cmpxchg(
+ v, addr, old[0], new[0], bytes, sh_ctxt);
+
+#ifdef __i386__
+ if ( bytes == 8 )
+ return v->arch.paging.mode->shadow.x86_emulate_cmpxchg8b(
+ v, addr, old[0], old[1], new[0], new[1], sh_ctxt);
+#endif
+
+ return X86EMUL_UNHANDLEABLE;
}
static struct x86_emulate_ops hvm_shadow_emulator_ops = {
@@ -294,7 +280,6 @@ static struct x86_emulate_ops hvm_shadow
.insn_fetch = hvm_emulate_insn_fetch,
.write = hvm_emulate_write,
.cmpxchg = hvm_emulate_cmpxchg,
- .cmpxchg8b = hvm_emulate_cmpxchg8b,
};
static int
@@ -338,36 +323,34 @@ static int
static int
pv_emulate_cmpxchg(enum x86_segment seg,
unsigned long offset,
- unsigned long old,
- unsigned long new,
+ void *p_old,
+ void *p_new,
unsigned int bytes,
struct x86_emulate_ctxt *ctxt)
{
struct sh_emulate_ctxt *sh_ctxt =
container_of(ctxt, struct sh_emulate_ctxt, ctxt);
+ unsigned long old[2], new[2];
struct vcpu *v = current;
+
if ( !is_x86_user_segment(seg) )
return X86EMUL_UNHANDLEABLE;
- return v->arch.paging.mode->shadow.x86_emulate_cmpxchg(
- v, offset, old, new, bytes, sh_ctxt);
-}
-
-static int
-pv_emulate_cmpxchg8b(enum x86_segment seg,
- unsigned long offset,
- unsigned long old_lo,
- unsigned long old_hi,
- unsigned long new_lo,
- unsigned long new_hi,
- struct x86_emulate_ctxt *ctxt)
-{
- struct sh_emulate_ctxt *sh_ctxt =
- container_of(ctxt, struct sh_emulate_ctxt, ctxt);
- struct vcpu *v = current;
- if ( !is_x86_user_segment(seg) )
- return X86EMUL_UNHANDLEABLE;
- return v->arch.paging.mode->shadow.x86_emulate_cmpxchg8b(
- v, offset, old_lo, old_hi, new_lo, new_hi, sh_ctxt);
+
+ old[0] = new[0] = 0;
+ memcpy(old, p_old, bytes);
+ memcpy(new, p_new, bytes);
+
+ if ( bytes <= sizeof(long) )
+ return v->arch.paging.mode->shadow.x86_emulate_cmpxchg(
+ v, offset, old[0], new[0], bytes, sh_ctxt);
+
+#ifdef __i386__
+ if ( bytes == 8 )
+ return v->arch.paging.mode->shadow.x86_emulate_cmpxchg8b(
+ v, offset, old[0], old[1], new[0], new[1], sh_ctxt);
+#endif
+
+ return X86EMUL_UNHANDLEABLE;
}
static struct x86_emulate_ops pv_shadow_emulator_ops = {
@@ -375,7 +358,6 @@ static struct x86_emulate_ops pv_shadow_
.insn_fetch = pv_emulate_read,
.write = pv_emulate_write,
.cmpxchg = pv_emulate_cmpxchg,
- .cmpxchg8b = pv_emulate_cmpxchg8b,
};
struct x86_emulate_ops *shadow_init_emulation(
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/mm/shadow/multi.c
--- a/xen/arch/x86/mm/shadow/multi.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/mm/shadow/multi.c Thu Apr 24 14:08:29 2008 -0600
@@ -2089,7 +2089,7 @@ static shadow_l1e_t * shadow_get_and_cre
else
{
/* Shadowing an actual guest l1 table */
- if ( !mfn_valid(gw->l2mfn) ) return NULL; /* No guest page. */
+ if ( !mfn_valid(gw->l1mfn) ) return NULL; /* No guest page. */
*sl1mfn = get_shadow_status(v, gw->l1mfn, SH_type_l1_shadow);
if ( !mfn_valid(*sl1mfn) )
{
@@ -4365,7 +4365,7 @@ static void emulate_unmap_dest(struct vc
atomic_inc(&v->domain->arch.paging.shadow.gtable_dirty_version);
}
-int
+static int
sh_x86_emulate_write(struct vcpu *v, unsigned long vaddr, void *src,
u32 bytes, struct sh_emulate_ctxt *sh_ctxt)
{
@@ -4389,7 +4389,7 @@ sh_x86_emulate_write(struct vcpu *v, uns
return X86EMUL_OKAY;
}
-int
+static int
sh_x86_emulate_cmpxchg(struct vcpu *v, unsigned long vaddr,
unsigned long old, unsigned long new,
unsigned int bytes, struct sh_emulate_ctxt *sh_ctxt)
@@ -4432,7 +4432,8 @@ sh_x86_emulate_cmpxchg(struct vcpu *v, u
return rv;
}
-int
+#ifdef __i386__
+static int
sh_x86_emulate_cmpxchg8b(struct vcpu *v, unsigned long vaddr,
unsigned long old_lo, unsigned long old_hi,
unsigned long new_lo, unsigned long new_hi,
@@ -4465,7 +4466,7 @@ sh_x86_emulate_cmpxchg8b(struct vcpu *v,
shadow_unlock(v->domain);
return rv;
}
-
+#endif
/**************************************************************************/
/* Audit tools */
@@ -4738,7 +4739,9 @@ struct paging_mode sh_paging_mode = {
.shadow.detach_old_tables = sh_detach_old_tables,
.shadow.x86_emulate_write = sh_x86_emulate_write,
.shadow.x86_emulate_cmpxchg = sh_x86_emulate_cmpxchg,
+#ifdef __i386__
.shadow.x86_emulate_cmpxchg8b = sh_x86_emulate_cmpxchg8b,
+#endif
.shadow.make_monitor_table = sh_make_monitor_table,
.shadow.destroy_monitor_table = sh_destroy_monitor_table,
#if SHADOW_OPTIMIZATIONS & SHOPT_WRITABLE_HEURISTIC
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/setup.c
--- a/xen/arch/x86/setup.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/setup.c Thu Apr 24 14:08:29 2008 -0600
@@ -1019,10 +1019,6 @@ void __init __start_xen(unsigned long mb
_initrd_len = mod[initrdidx].mod_end - mod[initrdidx].mod_start;
}
- iommu_setup();
-
- amd_iommu_detect();
-
/*
* We're going to setup domain0 using the module(s) that we stashed safely
* above our heap. The second module, if present, is an initrd ramdisk.
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/smp.c
--- a/xen/arch/x86/smp.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/smp.c Thu Apr 24 14:08:29 2008 -0600
@@ -75,20 +75,10 @@ static inline int __prepare_ICR2 (unsign
return SET_APIC_DEST_FIELD(mask);
}
-static inline void check_IPI_mask(cpumask_t cpumask)
-{
- /*
- * Sanity, and necessary. An IPI with no target generates a send accept
- * error with Pentium and P6 APICs.
- */
- ASSERT(cpus_subset(cpumask, cpu_online_map));
- ASSERT(!cpus_empty(cpumask));
-}
-
void apic_wait_icr_idle(void)
{
- while ( apic_read( APIC_ICR ) & APIC_ICR_BUSY )
- cpu_relax();
+ while ( apic_read( APIC_ICR ) & APIC_ICR_BUSY )
+ cpu_relax();
}
void send_IPI_mask_flat(cpumask_t cpumask, int vector)
@@ -97,7 +87,8 @@ void send_IPI_mask_flat(cpumask_t cpumas
unsigned long cfg;
unsigned long flags;
- check_IPI_mask(cpumask);
+ /* An IPI with no target generates a send accept error from P5/P6 APICs. */
+ WARN_ON(mask == 0);
local_irq_save(flags);
@@ -130,17 +121,9 @@ void send_IPI_mask_phys(cpumask_t mask,
unsigned long cfg, flags;
unsigned int query_cpu;
- check_IPI_mask(mask);
-
- /*
- * Hack. The clustered APIC addressing mode doesn't allow us to send
- * to an arbitrary mask, so I do a unicasts to each CPU instead. This
- * should be modified to do 1 message per cluster ID - mbligh
- */
-
local_irq_save(flags);
- for_each_cpu_mask( query_cpu, mask )
+ for_each_cpu_mask ( query_cpu, mask )
{
/*
* Wait for idle.
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/traps.c
--- a/xen/arch/x86/traps.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/traps.c Thu Apr 24 14:08:29 2008 -0600
@@ -479,6 +479,7 @@ static inline void do_trap(
static inline void do_trap(
int trapnr, struct cpu_user_regs *regs, int use_error_code)
{
+ struct vcpu *curr = current;
unsigned long fixup;
DEBUGGER_trap_entry(trapnr, regs);
@@ -494,6 +495,14 @@ static inline void do_trap(
dprintk(XENLOG_ERR, "Trap %d: %p -> %p\n",
trapnr, _p(regs->eip), _p(fixup));
regs->eip = fixup;
+ return;
+ }
+
+ if ( ((trapnr == TRAP_copro_error) || (trapnr == TRAP_simd_error)) &&
+ is_hvm_vcpu(curr) && curr->arch.hvm_vcpu.fpu_exception_callback )
+ {
+ curr->arch.hvm_vcpu.fpu_exception_callback(
+ curr->arch.hvm_vcpu.fpu_exception_callback_arg, regs);
return;
}
@@ -1399,6 +1408,13 @@ static int admin_io_okay(
unsigned int port, unsigned int bytes,
struct vcpu *v, struct cpu_user_regs *regs)
{
+ /*
+ * Port 0xcf8 (CONFIG_ADDRESS) is only visible for DWORD accesses.
+ * We never permit direct access to that register.
+ */
+ if ( (port == 0xcf8) && (bytes == 4) )
+ return 0;
+
return ioports_access_permitted(v->domain, port, port + bytes - 1);
}
@@ -1431,10 +1447,10 @@ static uint32_t guest_io_read(
{
sub_data = pv_pit_handler(port, 0, 0);
}
- else if ( (port & 0xfffc) == 0xcf8 )
- {
- size = min(bytes, 4 - (port & 3));
- sub_data = v->domain->arch.pci_cf8 >> ((port & 3) * 8);
+ else if ( (port == 0xcf8) && (bytes == 4) )
+ {
+ size = 4;
+ sub_data = v->domain->arch.pci_cf8;
}
else if ( ((port & 0xfffc) == 0xcfc) && IS_PRIV(v->domain) )
{
@@ -1489,19 +1505,10 @@ static void guest_io_write(
{
pv_pit_handler(port, (uint8_t)data, 1);
}
- else if ( (port & 0xfffc) == 0xcf8 )
- {
- size = min(bytes, 4 - (port & 3));
- if ( size == 4 )
- {
- v->domain->arch.pci_cf8 = data;
- }
- else
- {
- uint32_t mask = ((1u << (size * 8)) - 1) << ((port & 3) * 8);
- v->domain->arch.pci_cf8 &= ~mask;
- v->domain->arch.pci_cf8 |= (data << ((port & 3) * 8)) & mask;
- }
+ else if ( (port == 0xcf8) && (bytes == 4) )
+ {
+ size = 4;
+ v->domain->arch.pci_cf8 = data;
}
else if ( ((port & 0xfffc) == 0xcfc) && IS_PRIV(v->domain) )
{
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/x86_emulate.c
--- a/xen/arch/x86/x86_emulate.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/x86_emulate.c Thu Apr 24 14:08:29 2008 -0600
@@ -11,23 +11,7 @@
#include <asm/x86_emulate.h>
+/* Avoid namespace pollution. */
#undef cmpxchg
-#define __emulate_fpu_insn(_op) \
-do{ int _exn; \
- asm volatile ( \
- "1: " _op "\n" \
- "2: \n" \
- ".section .fixup,\"ax\"\n" \
- "3: mov $1,%0\n" \
- " jmp 2b\n" \
- ".previous\n" \
- ".section __ex_table,\"a\"\n" \
- " "__FIXUP_ALIGN"\n" \
- " "__FIXUP_WORD" 1b,3b\n" \
- ".previous" \
- : "=r" (_exn) : "0" (0) ); \
- generate_exception_if(_exn, EXC_MF, -1); \
-} while (0)
-
#include "x86_emulate/x86_emulate.c"
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/x86_emulate/x86_emulate.c
--- a/xen/arch/x86/x86_emulate/x86_emulate.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/x86_emulate/x86_emulate.c Thu Apr 24 14:08:29 2008 -0600
@@ -195,9 +195,9 @@ static uint8_t twobyte_table[256] = {
/* 0x50 - 0x5F */
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
/* 0x60 - 0x6F */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ImplicitOps|ModRM,
/* 0x70 - 0x7F */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ImplicitOps|ModRM,
/* 0x80 - 0x87 */
ImplicitOps, ImplicitOps, ImplicitOps, ImplicitOps,
ImplicitOps, ImplicitOps, ImplicitOps, ImplicitOps,
@@ -546,6 +546,62 @@ do {
? (uint16_t)_regs.eip : (uint32_t)_regs.eip); \
} while (0)
+struct fpu_insn_ctxt {
+ uint8_t insn_bytes;
+ uint8_t exn_raised;
+};
+
+static void fpu_handle_exception(void *_fic, struct cpu_user_regs *regs)
+{
+ struct fpu_insn_ctxt *fic = _fic;
+ fic->exn_raised = 1;
+ regs->eip += fic->insn_bytes;
+}
+
+#define get_fpu(_type, _fic) \
+do{ (_fic)->exn_raised = 0; \
+ fail_if(ops->get_fpu == NULL); \
+ rc = ops->get_fpu(fpu_handle_exception, _fic, _type, ctxt); \
+ if ( rc ) goto done; \
+} while (0)
+#define put_fpu(_fic) \
+do{ \
+ if ( ops->put_fpu != NULL ) \
+ ops->put_fpu(ctxt); \
+ generate_exception_if((_fic)->exn_raised, EXC_MF, -1); \
+} while (0)
+
+#define emulate_fpu_insn(_op) \
+do{ struct fpu_insn_ctxt fic; \
+ get_fpu(X86EMUL_FPU_fpu, &fic); \
+ asm volatile ( \
+ "movb $2f-1f,%0 \n" \
+ "1: " _op " \n" \
+ "2: \n" \
+ : "=m" (fic.insn_bytes) : : "memory" ); \
+ put_fpu(&fic); \
+} while (0)
+
+#define emulate_fpu_insn_memdst(_op, _arg) \
+do{ struct fpu_insn_ctxt fic; \
+ get_fpu(X86EMUL_FPU_fpu, &fic); \
+ asm volatile ( \
+ "movb $2f-1f,%0 \n" \
+ "1: " _op " %1 \n" \
+ "2: \n" \
+ : "=m" (fic.insn_bytes), "=m" (_arg) \
+ : : "memory" ); \
+ put_fpu(&fic); \
+} while (0)
+
+#define emulate_fpu_insn_stub(_bytes...) \
+do{ uint8_t stub[] = { _bytes, 0xc3 }; \
+ struct fpu_insn_ctxt fic = { .insn_bytes = sizeof(stub)-1 }; \
+ get_fpu(X86EMUL_FPU_fpu, &fic); \
+ (*(void(*)(void))stub)(); \
+ put_fpu(&fic); \
+} while (0)
+
static unsigned long __get_rep_prefix(
struct cpu_user_regs *int_regs,
struct cpu_user_regs *ext_regs,
@@ -851,6 +907,7 @@ protmode_load_seg(
struct { uint32_t a, b; } desc;
unsigned long val;
uint8_t dpl, rpl, cpl;
+ uint32_t new_desc_b;
int rc, fault_type = EXC_TS;
/* NULL selector? */
@@ -933,10 +990,11 @@ protmode_load_seg(
}
/* Ensure Accessed flag is set. */
+ new_desc_b = desc.b | 0x100;
rc = ((desc.b & 0x100) ? X86EMUL_OKAY :
ops->cmpxchg(
- x86_seg_none, desctab.base + (sel & 0xfff8) + 4, desc.b,
- desc.b | 0x100, 4, ctxt));
+ x86_seg_none, desctab.base + (sel & 0xfff8) + 4,
+ &desc.b, &new_desc_b, 4, ctxt));
} while ( rc == X86EMUL_CMPXCHG_FAILED );
if ( rc )
@@ -2036,8 +2094,8 @@ x86_emulate(
/* nothing to do */;
else if ( lock_prefix )
rc = ops->cmpxchg(
- dst.mem.seg, dst.mem.off, dst.orig_val,
- dst.val, dst.bytes, ctxt);
+ dst.mem.seg, dst.mem.off, &dst.orig_val,
+ &dst.val, dst.bytes, ctxt);
else
rc = ops->write(
dst.mem.seg, dst.mem.off, dst.val, dst.bytes, ctxt);
@@ -2399,9 +2457,7 @@ x86_emulate(
}
case 0x9b: /* wait/fwait */
- fail_if(ops->load_fpu_ctxt == NULL);
- ops->load_fpu_ctxt(ctxt);
- __emulate_fpu_insn("fwait");
+ emulate_fpu_insn("fwait");
break;
case 0x9c: /* pushf */
@@ -2721,77 +2777,89 @@ x86_emulate(
}
case 0xd9: /* FPU 0xd9 */
- fail_if(ops->load_fpu_ctxt == NULL);
- ops->load_fpu_ctxt(ctxt);
switch ( modrm )
{
- case 0xc0: __emulate_fpu_insn(".byte 0xd9,0xc0"); break;
- case 0xc1: __emulate_fpu_insn(".byte 0xd9,0xc1"); break;
- case 0xc2: __emulate_fpu_insn(".byte 0xd9,0xc2"); break;
- case 0xc3: __emulate_fpu_insn(".byte 0xd9,0xc3"); break;
- case 0xc4: __emulate_fpu_insn(".byte 0xd9,0xc4"); break;
- case 0xc5: __emulate_fpu_insn(".byte 0xd9,0xc5"); break;
- case 0xc6: __emulate_fpu_insn(".byte 0xd9,0xc6"); break;
- case 0xc7: __emulate_fpu_insn(".byte 0xd9,0xc7"); break;
- case 0xe0: __emulate_fpu_insn(".byte 0xd9,0xe0"); break;
- case 0xe8: __emulate_fpu_insn(".byte 0xd9,0xe8"); break;
- case 0xee: __emulate_fpu_insn(".byte 0xd9,0xee"); break;
+ case 0xc0 ... 0xc7: /* fld %stN */
+ case 0xc8 ... 0xcf: /* fxch %stN */
+ case 0xd0: /* fnop */
+ case 0xe0: /* fchs */
+ case 0xe1: /* fabs */
+ case 0xe4: /* ftst */
+ case 0xe5: /* fxam */
+ case 0xe8: /* fld1 */
+ case 0xe9: /* fldl2t */
+ case 0xea: /* fldl2e */
+ case 0xeb: /* fldpi */
+ case 0xec: /* fldlg2 */
+ case 0xed: /* fldln2 */
+ case 0xee: /* fldz */
+ case 0xf0: /* f2xm1 */
+ case 0xf1: /* fyl2x */
+ case 0xf2: /* fptan */
+ case 0xf3: /* fpatan */
+ case 0xf4: /* fxtract */
+ case 0xf5: /* fprem1 */
+ case 0xf6: /* fdecstp */
+ case 0xf7: /* fincstp */
+ case 0xf8: /* fprem */
+ case 0xf9: /* fyl2xp1 */
+ case 0xfa: /* fsqrt */
+ case 0xfb: /* fsincos */
+ case 0xfc: /* frndint */
+ case 0xfd: /* fscale */
+ case 0xfe: /* fsin */
+ case 0xff: /* fcos */
+ emulate_fpu_insn_stub(0xd9, modrm);
+ break;
default:
fail_if((modrm_reg & 7) != 7);
fail_if(modrm >= 0xc0);
/* fnstcw m2byte */
ea.bytes = 2;
dst = ea;
- asm volatile ( "fnstcw %0" : "=m" (dst.val) );
+ emulate_fpu_insn_memdst("fnstcw", dst.val);
}
break;
case 0xdb: /* FPU 0xdb */
- fail_if(ops->load_fpu_ctxt == NULL);
- ops->load_fpu_ctxt(ctxt);
fail_if(modrm != 0xe3);
/* fninit */
- asm volatile ( "fninit" );
+ emulate_fpu_insn("fninit");
break;
case 0xdd: /* FPU 0xdd */
- fail_if(ops->load_fpu_ctxt == NULL);
- ops->load_fpu_ctxt(ctxt);
fail_if((modrm_reg & 7) != 7);
fail_if(modrm >= 0xc0);
/* fnstsw m2byte */
ea.bytes = 2;
dst = ea;
- asm volatile ( "fnstsw %0" : "=m" (dst.val) );
+ emulate_fpu_insn_memdst("fnstsw", dst.val);
break;
case 0xde: /* FPU 0xde */
- fail_if(ops->load_fpu_ctxt == NULL);
- ops->load_fpu_ctxt(ctxt);
switch ( modrm )
{
- case 0xd9: __emulate_fpu_insn(".byte 0xde,0xd9"); break;
- case 0xf8: __emulate_fpu_insn(".byte 0xde,0xf8"); break;
- case 0xf9: __emulate_fpu_insn(".byte 0xde,0xf9"); break;
- case 0xfa: __emulate_fpu_insn(".byte 0xde,0xfa"); break;
- case 0xfb: __emulate_fpu_insn(".byte 0xde,0xfb"); break;
- case 0xfc: __emulate_fpu_insn(".byte 0xde,0xfc"); break;
- case 0xfd: __emulate_fpu_insn(".byte 0xde,0xfd"); break;
- case 0xfe: __emulate_fpu_insn(".byte 0xde,0xfe"); break;
- case 0xff: __emulate_fpu_insn(".byte 0xde,0xff"); break;
- default: goto cannot_emulate;
+ case 0xc0 ... 0xc7: /* faddp %stN */
+ case 0xc8 ... 0xcf: /* fmulp %stN */
+ case 0xd9: /* fcompp */
+ case 0xe0 ... 0xe7: /* fsubrp %stN */
+ case 0xe8 ... 0xef: /* fsubp %stN */
+ case 0xf0 ... 0xf7: /* fdivrp %stN */
+ case 0xf8 ... 0xff: /* fdivp %stN */
+ emulate_fpu_insn_stub(0xde, modrm);
+ break;
+ default:
+ goto cannot_emulate;
}
break;
case 0xdf: /* FPU 0xdf */
- fail_if(ops->load_fpu_ctxt == NULL);
- ops->load_fpu_ctxt(ctxt);
fail_if(modrm != 0xe0);
/* fnstsw %ax */
dst.bytes = 2;
dst.type = OP_REG;
dst.reg = (unsigned long *)&_regs.eax;
- asm volatile ( "fnstsw %0" : "=m" (dst.val) );
+ emulate_fpu_insn_memdst("fnstsw", dst.val);
break;
case 0xe0 ... 0xe2: /* loop{,z,nz} */ {
@@ -2975,6 +3043,7 @@ x86_emulate(
case 0xa3: bt: /* bt */
emulate_2op_SrcV_nobyte("bt", src, dst, _regs.eflags);
+ dst.type = OP_NONE;
break;
case 0xa4: /* shld imm8,r,r/m */
@@ -3067,7 +3136,11 @@ x86_emulate(
: "=r" (dst.val), "=q" (zf)
: "r" (src.val), "1" (0) );
_regs.eflags &= ~EFLG_ZF;
- _regs.eflags |= zf ? EFLG_ZF : 0;
+ if ( zf )
+ {
+ _regs.eflags |= EFLG_ZF;
+ dst.type = OP_NONE;
+ }
break;
}
@@ -3077,7 +3150,11 @@ x86_emulate(
: "=r" (dst.val), "=q" (zf)
: "r" (src.val), "1" (0) );
_regs.eflags &= ~EFLG_ZF;
- _regs.eflags |= zf ? EFLG_ZF : 0;
+ if ( zf )
+ {
+ _regs.eflags |= EFLG_ZF;
+ dst.type = OP_NONE;
+ }
break;
}
@@ -3310,6 +3387,44 @@ x86_emulate(
break;
}
+ case 0x6f: /* movq mm/m64,mm */ {
+ uint8_t stub[] = { 0x0f, 0x6f, modrm, 0xc3 };
+ struct fpu_insn_ctxt fic = { .insn_bytes = sizeof(stub)-1 };
+ uint64_t val;
+ if ( ea.type == OP_MEM )
+ {
+ unsigned long lval, hval;
+ if ( (rc = ops->read(ea.mem.seg, ea.mem.off+0, &lval, 4, ctxt)) ||
+ (rc = ops->read(ea.mem.seg, ea.mem.off+4, &hval, 4, ctxt)) )
+ goto done;
+ val = ((uint64_t)hval << 32) | (uint32_t)lval;
+ stub[2] = modrm & 0x38; /* movq (%eax),%mmN */
+ }
+ get_fpu(X86EMUL_FPU_mmx, &fic);
+ asm volatile ( "call *%0" : : "r" (stub), "a" (&val) : "memory" );
+ put_fpu(&fic);
+ break;
+ }
+
+ case 0x7f: /* movq mm,mm/m64 */ {
+ uint8_t stub[] = { 0x0f, 0x7f, modrm, 0xc3 };
+ struct fpu_insn_ctxt fic = { .insn_bytes = sizeof(stub)-1 };
+ uint64_t val;
+ if ( ea.type == OP_MEM )
+ stub[2] = modrm & 0x38; /* movq %mmN,(%eax) */
+ get_fpu(X86EMUL_FPU_mmx, &fic);
+ asm volatile ( "call *%0" : : "r" (stub), "a" (&val) : "memory" );
+ put_fpu(&fic);
+ if ( ea.type == OP_MEM )
+ {
+ unsigned long lval = (uint32_t)val, hval = (uint32_t)(val >> 32);
+ if ( (rc = ops->write(ea.mem.seg, ea.mem.off+0, lval, 4, ctxt)) ||
+ (rc = ops->write(ea.mem.seg, ea.mem.off+4, hval, 4, ctxt)) )
+ goto done;
+ }
+ break;
+ }
+
case 0x80 ... 0x8f: /* jcc (near) */ {
int rel = (((op_bytes == 2) && !mode_64bit())
? (int32_t)insn_fetch_type(int16_t)
@@ -3346,60 +3461,49 @@ x86_emulate(
src.val = x86_seg_gs;
goto pop_seg;
- case 0xc7: /* Grp9 (cmpxchg8b) */
-#if defined(__i386__)
- {
- unsigned long old_lo, old_hi;
+ case 0xc7: /* Grp9 (cmpxchg8b/cmpxchg16b) */ {
+ unsigned long old[2], exp[2], new[2];
+ unsigned int i;
+
generate_exception_if((modrm_reg & 7) != 1, EXC_UD, -1);
generate_exception_if(ea.type != OP_MEM, EXC_UD, -1);
- if ( (rc = ops->read(ea.mem.seg, ea.mem.off+0, &old_lo, 4, ctxt)) ||
- (rc = ops->read(ea.mem.seg, ea.mem.off+4, &old_hi, 4, ctxt)) )
- goto done;
- if ( (old_lo != _regs.eax) || (old_hi != _regs.edx) )
- {
- _regs.eax = old_lo;
- _regs.edx = old_hi;
+ op_bytes *= 2;
+
+ /* Get actual old value. */
+ for ( i = 0; i < (op_bytes/sizeof(long)); i++ )
+ if ( (rc = ops->read(ea.mem.seg, ea.mem.off + i*sizeof(long),
+ &old[i], sizeof(long), ctxt)) != 0 )
+ goto done;
+
+ /* Get expected and proposed values. */
+ if ( op_bytes == 8 )
+ {
+ ((uint32_t *)exp)[0] = _regs.eax; ((uint32_t *)exp)[1] = _regs.edx;
+ ((uint32_t *)new)[0] = _regs.ebx; ((uint32_t *)new)[1] = _regs.ecx;
+ }
+ else
+ {
+ exp[0] = _regs.eax; exp[1] = _regs.edx;
+ new[0] = _regs.ebx; new[1] = _regs.ecx;
+ }
+
+ if ( memcmp(old, exp, op_bytes) )
+ {
+ /* Expected != actual: store actual to rDX:rAX and clear ZF. */
+ _regs.eax = (op_bytes == 8) ? ((uint32_t *)old)[0] : old[0];
+ _regs.edx = (op_bytes == 8) ? ((uint32_t *)old)[1] : old[1];
_regs.eflags &= ~EFLG_ZF;
}
- else if ( ops->cmpxchg8b == NULL )
- {
- rc = X86EMUL_UNHANDLEABLE;
- goto done;
- }
else
{
- if ( (rc = ops->cmpxchg8b(ea.mem.seg, ea.mem.off, old_lo, old_hi,
- _regs.ebx, _regs.ecx, ctxt)) != 0 )
+ /* Expected == actual: attempt atomic cmpxchg and set ZF. */
+ if ( (rc = ops->cmpxchg(ea.mem.seg, ea.mem.off, old,
+ new, op_bytes, ctxt)) != 0 )
goto done;
_regs.eflags |= EFLG_ZF;
}
break;
}
-#elif defined(__x86_64__)
- {
- unsigned long old, new;
- generate_exception_if((modrm_reg & 7) != 1, EXC_UD, -1);
- generate_exception_if(ea.type != OP_MEM, EXC_UD, -1);
- if ( (rc = ops->read(ea.mem.seg, ea.mem.off, &old, 8, ctxt)) != 0 )
- goto done;
- if ( ((uint32_t)(old>>0) != (uint32_t)_regs.eax) ||
- ((uint32_t)(old>>32) != (uint32_t)_regs.edx) )
- {
- _regs.eax = (uint32_t)(old>>0);
- _regs.edx = (uint32_t)(old>>32);
- _regs.eflags &= ~EFLG_ZF;
- }
- else
- {
- new = (_regs.ecx<<32)|(uint32_t)_regs.ebx;
- if ( (rc = ops->cmpxchg(ea.mem.seg, ea.mem.off, old,
- new, 8, ctxt)) != 0 )
- goto done;
- _regs.eflags |= EFLG_ZF;
- }
- break;
- }
-#endif
case 0xc8 ... 0xcf: /* bswap */
dst.type = OP_REG;
diff -r 239b44eeb2d6 -r dc510776dd59 xen/arch/x86/x86_emulate/x86_emulate.h
--- a/xen/arch/x86/x86_emulate/x86_emulate.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/arch/x86/x86_emulate/x86_emulate.h Thu Apr 24 14:08:29 2008 -0600
@@ -95,6 +95,12 @@ struct segment_register {
/* (cmpxchg accessor): CMPXCHG failed. Maps to X86EMUL_RETRY in caller. */
#define X86EMUL_CMPXCHG_FAILED 3
+/* FPU sub-types which may be requested via ->get_fpu(). */
+enum x86_emulate_fpu_type {
+ X86EMUL_FPU_fpu, /* Standard FPU coprocessor instruction set */
+ X86EMUL_FPU_mmx /* MMX instruction set (%mm0-%mm7) */
+};
+
/*
* These operations represent the instruction emulator's interface to memory.
*
@@ -104,8 +110,7 @@ struct segment_register {
* some out-of-band mechanism, unknown to the emulator. The memop signals
* failure by returning X86EMUL_EXCEPTION to the emulator, which will
* then immediately bail.
- * 2. Valid access sizes are 1, 2, 4 and 8 bytes. On x86/32 systems only
- * cmpxchg8b_emulated need support 8-byte accesses.
+ * 2. Valid access sizes are 1, 2, 4 and 8 (x86/64 only) bytes.
* 3. The emulator cannot handle 64-bit mode emulation on an x86/32 system.
*/
struct x86_emulate_ops
@@ -153,34 +158,16 @@ struct x86_emulate_ops
/*
* cmpxchg: Emulate an atomic (LOCKed) CMPXCHG operation.
- * @old: [IN ] Value expected to be current at @addr.
- * @new: [IN ] Value to write to @addr.
+ * @p_old: [IN ] Pointer to value expected to be current at @addr.
+ * @p_new: [IN ] Pointer to value to write to @addr.
+ * @bytes: [IN ] Operation size (up to 8 (x86/32) or 16 (x86/64) bytes).
*/
int (*cmpxchg)(
enum x86_segment seg,
unsigned long offset,
- unsigned long old,
- unsigned long new,
- unsigned int bytes,
- struct x86_emulate_ctxt *ctxt);
-
- /*
- * cmpxchg8b: Emulate an atomic (LOCKed) CMPXCHG8B operation.
- * @old: [IN ] Value expected to be current at @addr.
- * @new: [IN ] Value to write to @addr.
- * NOTES:
- * 1. This function is only ever called when emulating a real CMPXCHG8B.
- * 2. This function is *never* called on x86/64 systems.
- * 2. Not defining this function (i.e., specifying NULL) is equivalent
- * to defining a function that always returns X86EMUL_UNHANDLEABLE.
- */
- int (*cmpxchg8b)(
- enum x86_segment seg,
- unsigned long offset,
- unsigned long old_lo,
- unsigned long old_hi,
- unsigned long new_lo,
- unsigned long new_hi,
+ void *p_old,
+ void *p_new,
+ unsigned int bytes,
struct x86_emulate_ctxt *ctxt);
/*
@@ -342,8 +329,19 @@ struct x86_emulate_ops
uint8_t insn_len,
struct x86_emulate_ctxt *ctxt);
- /* load_fpu_ctxt: Load emulated environment's FPU state onto processor. */
- void (*load_fpu_ctxt)(
+ /*
+ * get_fpu: Load emulated environment's FPU state onto processor.
+ * @exn_callback: On any FPU or SIMD exception, pass control to
+ * (*exception_callback)(exception_callback_arg, regs).
+ */
+ int (*get_fpu)(
+ void (*exception_callback)(void *, struct cpu_user_regs *),
+ void *exception_callback_arg,
+ enum x86_emulate_fpu_type type,
+ struct x86_emulate_ctxt *ctxt);
+
+ /* put_fpu: Relinquish the FPU. Unhook from FPU/SIMD exception handlers. */
+ void (*put_fpu)(
struct x86_emulate_ctxt *ctxt);
/* invlpg: Invalidate paging structures which map addressed byte. */
diff -r 239b44eeb2d6 -r dc510776dd59 xen/common/trace.c
--- a/xen/common/trace.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/common/trace.c Thu Apr 24 14:08:29 2008 -0600
@@ -374,6 +374,15 @@ static inline int insert_lost_records(st
(unsigned char *)&ed);
}
+/*
+ * Notification is performed in qtasklet to avoid deadlocks with contexts
+ * which __trace_var() may be called from (e.g., scheduler critical regions).
+ */
+static void trace_notify_dom0(unsigned long unused)
+{
+ send_guest_global_virq(dom0, VIRQ_TBUF);
+}
+static DECLARE_TASKLET(trace_notify_dom0_tasklet, trace_notify_dom0, 0);
/**
* trace - Enters a trace tuple into the trace buffer for the current CPU.
@@ -506,7 +515,7 @@ void __trace_var(u32 event, int cycles,
/* Notify trace buffer consumer that we've crossed the high water mark. */
if ( started_below_highwater &&
(calc_unconsumed_bytes(buf) >= t_buf_highwater) )
- send_guest_global_virq(dom0, VIRQ_TBUF);
+ tasklet_schedule(&trace_notify_dom0_tasklet);
}
/*
diff -r 239b44eeb2d6 -r dc510776dd59 xen/common/xencomm.c
--- a/xen/common/xencomm.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/common/xencomm.c Thu Apr 24 14:08:29 2008 -0600
@@ -323,7 +323,7 @@ xencomm_copy_chunk_to(
(unsigned long)xencomm_vaddr(paddr, page));
memcpy(xencomm_vaddr(paddr, page), (void *)from, len);
- xencomm_mark_dirty(xencomm_vaddr(paddr, page), len);
+ xencomm_mark_dirty((unsigned long)xencomm_vaddr(paddr, page), len);
put_page(page);
return 0;
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/char/console.c
--- a/xen/drivers/char/console.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/char/console.c Thu Apr 24 14:08:29 2008 -0600
@@ -322,7 +322,7 @@ static long guest_console_write(XEN_GUES
while ( count > 0 )
{
- while ( serial_tx_space(sercon_handle) < (SERIAL_TXBUFSZ / 2) )
+ while ( serial_tx_space(sercon_handle) < (serial_txbufsz / 2) )
{
if ( hypercall_preempt_check() )
break;
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/char/serial.c
--- a/xen/drivers/char/serial.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/char/serial.c Thu Apr 24 14:08:29 2008 -0600
@@ -15,6 +15,19 @@
#include <xen/mm.h>
#include <xen/serial.h>
+/* Never drop characters, even if the async transmit buffer fills. */
+/* #define SERIAL_NEVER_DROP_CHARS 1 */
+
+unsigned int serial_txbufsz = 16384;
+static void __init parse_serial_tx_buffer(const char *s)
+{
+ serial_txbufsz = max((unsigned int)parse_size_and_unit(s, NULL), 512u);
+}
+custom_param("serial_tx_buffer", parse_serial_tx_buffer);
+
+#define mask_serial_rxbuf_idx(_i) ((_i)&(serial_rxbufsz-1))
+#define mask_serial_txbuf_idx(_i) ((_i)&(serial_txbufsz-1))
+
static struct serial_port com[2] = {
{ .rx_lock = SPIN_LOCK_UNLOCKED, .tx_lock = SPIN_LOCK_UNLOCKED },
{ .rx_lock = SPIN_LOCK_UNLOCKED, .tx_lock = SPIN_LOCK_UNLOCKED }
@@ -36,8 +49,8 @@ void serial_rx_interrupt(struct serial_p
fn = port->rx_hi;
else if ( !(c & 0x80) && (port->rx_lo != NULL) )
fn = port->rx_lo;
- else if ( (port->rxbufp - port->rxbufc) != SERIAL_RXBUFSZ )
- port->rxbuf[MASK_SERIAL_RXBUF_IDX(port->rxbufp++)] = c;
+ else if ( (port->rxbufp - port->rxbufc) != serial_rxbufsz )
+ port->rxbuf[mask_serial_rxbuf_idx(port->rxbufp++)] = c;
}
spin_unlock_irqrestore(&port->rx_lock, flags);
@@ -72,7 +85,7 @@ void serial_tx_interrupt(struct serial_p
if ( port->txbufc == port->txbufp )
break;
port->driver->putc(
- port, port->txbuf[MASK_SERIAL_TXBUF_IDX(port->txbufc++)]);
+ port, port->txbuf[mask_serial_txbuf_idx(port->txbufc++)]);
}
}
@@ -81,22 +94,24 @@ void serial_tx_interrupt(struct serial_p
static void __serial_putc(struct serial_port *port, char c)
{
- int i;
-
if ( (port->txbuf != NULL) && !port->sync )
{
/* Interrupt-driven (asynchronous) transmitter. */
- if ( (port->txbufp - port->txbufc) == SERIAL_TXBUFSZ )
- {
- /* Buffer is full: we spin, but could alternatively drop chars. */
+#ifdef SERIAL_NEVER_DROP_CHARS
+ if ( (port->txbufp - port->txbufc) == serial_txbufsz )
+ {
+ /* Buffer is full: we spin waiting for space to appear. */
+ int i;
while ( !port->driver->tx_empty(port) )
cpu_relax();
for ( i = 0; i < port->tx_fifo_size; i++ )
port->driver->putc(
- port, port->txbuf[MASK_SERIAL_TXBUF_IDX(port->txbufc++)]);
- port->txbuf[MASK_SERIAL_TXBUF_IDX(port->txbufp++)] = c;
- }
- else if ( ((port->txbufp - port->txbufc) == 0) &&
+ port, port->txbuf[mask_serial_txbuf_idx(port->txbufc++)]);
+ port->txbuf[mask_serial_txbuf_idx(port->txbufp++)] = c;
+ return;
+ }
+#endif
+ if ( ((port->txbufp - port->txbufc) == 0) &&
port->driver->tx_empty(port) )
{
/* Buffer and UART FIFO are both empty. */
@@ -105,7 +120,7 @@ static void __serial_putc(struct serial_
else
{
/* Normal case: buffer the character. */
- port->txbuf[MASK_SERIAL_TXBUF_IDX(port->txbufp++)] = c;
+ port->txbuf[mask_serial_txbuf_idx(port->txbufp++)] = c;
}
}
else if ( port->driver->tx_empty )
@@ -200,7 +215,7 @@ char serial_getc(int handle)
if ( port->rxbufp != port->rxbufc )
{
- c = port->rxbuf[MASK_SERIAL_RXBUF_IDX(port->rxbufc++)];
+ c = port->rxbuf[mask_serial_rxbuf_idx(port->rxbufc++)];
spin_unlock_irqrestore(&port->rx_lock, flags);
break;
}
@@ -336,7 +351,7 @@ void serial_start_sync(int handle)
while ( !port->driver->tx_empty(port) )
cpu_relax();
port->driver->putc(
- port, port->txbuf[MASK_SERIAL_TXBUF_IDX(port->txbufc++)]);
+ port, port->txbuf[mask_serial_txbuf_idx(port->txbufc++)]);
}
}
@@ -364,9 +379,9 @@ int serial_tx_space(int handle)
{
struct serial_port *port;
if ( handle == -1 )
- return SERIAL_TXBUFSZ;
- port = &com[handle & SERHND_IDX];
- return SERIAL_TXBUFSZ - (port->txbufp - port->txbufc);
+ return serial_txbufsz;
+ port = &com[handle & SERHND_IDX];
+ return serial_txbufsz - (port->txbufp - port->txbufc);
}
void __devinit serial_init_preirq(void)
@@ -431,7 +446,7 @@ void serial_async_transmit(struct serial
BUG_ON(!port->driver->tx_empty);
if ( port->txbuf == NULL )
port->txbuf = alloc_xenheap_pages(
- get_order_from_bytes(SERIAL_TXBUFSZ));
+ get_order_from_bytes(serial_txbufsz));
}
/*
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/amd/iommu_acpi.c
--- a/xen/drivers/passthrough/amd/iommu_acpi.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/passthrough/amd/iommu_acpi.c Thu Apr 24 14:08:29 2008 -0600
@@ -139,7 +139,7 @@ static int __init register_exclusion_ran
iommu = find_iommu_for_device(bus, devfn);
if ( !iommu )
{
- dprintk(XENLOG_ERR, "IVMD Error: No IOMMU for Dev_Id 0x%x!\n", bdf);
+ amd_iov_error("IVMD Error: No IOMMU for Dev_Id 0x%x!\n", bdf);
return -ENODEV;
}
req = ivrs_mappings[bdf].dte_requestor_id;
@@ -221,7 +221,7 @@ static int __init parse_ivmd_device_sele
bdf = ivmd_block->header.dev_id;
if ( bdf >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVMD Error: Invalid Dev_Id 0x%x\n", bdf);
+ amd_iov_error("IVMD Error: Invalid Dev_Id 0x%x\n", bdf);
return -ENODEV;
}
@@ -238,21 +238,18 @@ static int __init parse_ivmd_device_rang
first_bdf = ivmd_block->header.dev_id;
if ( first_bdf >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVMD Error: "
- "Invalid Range_First Dev_Id 0x%x\n", first_bdf);
+ amd_iov_error(
+ "IVMD Error: Invalid Range_First Dev_Id 0x%x\n", first_bdf);
return -ENODEV;
}
last_bdf = ivmd_block->last_dev_id;
if ( (last_bdf >= ivrs_bdf_entries) || (last_bdf <= first_bdf) )
{
- dprintk(XENLOG_ERR, "IVMD Error: "
- "Invalid Range_Last Dev_Id 0x%x\n", last_bdf);
- return -ENODEV;
- }
-
- dprintk(XENLOG_ERR, " Dev_Id Range: 0x%x -> 0x%x\n",
- first_bdf, last_bdf);
+ amd_iov_error(
+ "IVMD Error: Invalid Range_Last Dev_Id 0x%x\n", last_bdf);
+ return -ENODEV;
+ }
for ( bdf = first_bdf, error = 0; (bdf <= last_bdf) && !error; bdf++ )
error = register_exclusion_range_for_device(
@@ -272,8 +269,7 @@ static int __init parse_ivmd_device_iomm
ivmd_block->cap_offset);
if ( !iommu )
{
- dprintk(XENLOG_ERR,
- "IVMD Error: No IOMMU for Dev_Id 0x%x Cap 0x%x\n",
+ amd_iov_error("IVMD Error: No IOMMU for Dev_Id 0x%x Cap 0x%x\n",
ivmd_block->header.dev_id, ivmd_block->cap_offset);
return -ENODEV;
}
@@ -290,7 +286,7 @@ static int __init parse_ivmd_block(struc
if ( ivmd_block->header.length <
sizeof(struct acpi_ivmd_block_header) )
{
- dprintk(XENLOG_ERR, "IVMD Error: Invalid Block Length!\n");
+ amd_iov_error("IVMD Error: Invalid Block Length!\n");
return -ENODEV;
}
@@ -299,10 +295,9 @@ static int __init parse_ivmd_block(struc
base = start_addr & PAGE_MASK;
limit = (start_addr + mem_length - 1) & PAGE_MASK;
- dprintk(XENLOG_INFO, "IVMD Block: Type 0x%x\n",
- ivmd_block->header.type);
- dprintk(XENLOG_INFO, " Start_Addr_Phys 0x%lx\n", start_addr);
- dprintk(XENLOG_INFO, " Mem_Length 0x%lx\n", mem_length);
+ amd_iov_info("IVMD Block: Type 0x%x\n",ivmd_block->header.type);
+ amd_iov_info(" Start_Addr_Phys 0x%lx\n", start_addr);
+ amd_iov_info(" Mem_Length 0x%lx\n", mem_length);
if ( get_field_from_byte(ivmd_block->header.flags,
AMD_IOMMU_ACPI_EXCLUSION_RANGE_MASK,
@@ -321,7 +316,7 @@ static int __init parse_ivmd_block(struc
}
else
{
- dprintk(KERN_ERR, "IVMD Error: Invalid Flag Field!\n");
+ amd_iov_error("IVMD Error: Invalid Flag Field!\n");
return -ENODEV;
}
@@ -344,7 +339,7 @@ static int __init parse_ivmd_block(struc
base, limit, iw, ir);
default:
- dprintk(XENLOG_ERR, "IVMD Error: Invalid Block Type!\n");
+ amd_iov_error("IVMD Error: Invalid Block Type!\n");
return -ENODEV;
}
}
@@ -354,7 +349,7 @@ static u16 __init parse_ivhd_device_padd
{
if ( header_length < (block_length + pad_length) )
{
- dprintk(XENLOG_ERR, "IVHD Error: Invalid Device_Entry Length!\n");
+ amd_iov_error("IVHD Error: Invalid Device_Entry Length!\n");
return 0;
}
@@ -369,8 +364,7 @@ static u16 __init parse_ivhd_device_sele
bdf = ivhd_device->header.dev_id;
if ( bdf >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Device_Entry Dev_Id 0x%x\n", bdf);
+ amd_iov_error("IVHD Error: Invalid Device_Entry Dev_Id 0x%x\n", bdf);
return 0;
}
@@ -393,14 +387,14 @@ static u16 __init parse_ivhd_device_rang
dev_length = sizeof(struct acpi_ivhd_device_range);
if ( header_length < (block_length + dev_length) )
{
- dprintk(XENLOG_ERR, "IVHD Error: Invalid Device_Entry Length!\n");
+ amd_iov_error("IVHD Error: Invalid Device_Entry Length!\n");
return 0;
}
if ( ivhd_device->range.trailer.type !=
AMD_IOMMU_ACPI_IVHD_DEV_RANGE_END )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
+ amd_iov_error("IVHD Error: "
"Invalid Range: End_Type 0x%x\n",
ivhd_device->range.trailer.type);
return 0;
@@ -409,21 +403,20 @@ static u16 __init parse_ivhd_device_rang
first_bdf = ivhd_device->header.dev_id;
if ( first_bdf >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Range: First Dev_Id 0x%x\n", first_bdf);
+ amd_iov_error(
+ "IVHD Error: Invalid Range: First Dev_Id 0x%x\n", first_bdf);
return 0;
}
last_bdf = ivhd_device->range.trailer.dev_id;
if ( (last_bdf >= ivrs_bdf_entries) || (last_bdf <= first_bdf) )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Range: Last Dev_Id 0x%x\n", last_bdf);
- return 0;
- }
-
- dprintk(XENLOG_INFO, " Dev_Id Range: 0x%x -> 0x%x\n",
- first_bdf, last_bdf);
+ amd_iov_error(
+ "IVHD Error: Invalid Range: Last Dev_Id 0x%x\n", last_bdf);
+ return 0;
+ }
+
+ amd_iov_info(" Dev_Id Range: 0x%x -> 0x%x\n", first_bdf, last_bdf);
/* override flags for range of devices */
sys_mgt = get_field_from_byte(ivhd_device->header.flags,
@@ -444,28 +437,25 @@ static u16 __init parse_ivhd_device_alia
dev_length = sizeof(struct acpi_ivhd_device_alias);
if ( header_length < (block_length + dev_length) )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Device_Entry Length!\n");
+ amd_iov_error("IVHD Error: Invalid Device_Entry Length!\n");
return 0;
}
bdf = ivhd_device->header.dev_id;
if ( bdf >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Device_Entry Dev_Id 0x%x\n", bdf);
+ amd_iov_error("IVHD Error: Invalid Device_Entry Dev_Id 0x%x\n", bdf);
return 0;
}
alias_id = ivhd_device->alias.dev_id;
if ( alias_id >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Alias Dev_Id 0x%x\n", alias_id);
- return 0;
- }
-
- dprintk(XENLOG_INFO, " Dev_Id Alias: 0x%x\n", alias_id);
+ amd_iov_error("IVHD Error: Invalid Alias Dev_Id 0x%x\n", alias_id);
+ return 0;
+ }
+
+ amd_iov_info(" Dev_Id Alias: 0x%x\n", alias_id);
/* override requestor_id and flags for device */
ivrs_mappings[bdf].dte_requestor_id = alias_id;
@@ -490,15 +480,14 @@ static u16 __init parse_ivhd_device_alia
dev_length = sizeof(struct acpi_ivhd_device_alias_range);
if ( header_length < (block_length + dev_length) )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Device_Entry Length!\n");
+ amd_iov_error("IVHD Error: Invalid Device_Entry Length!\n");
return 0;
}
if ( ivhd_device->alias_range.trailer.type !=
AMD_IOMMU_ACPI_IVHD_DEV_RANGE_END )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
+ amd_iov_error("IVHD Error: "
"Invalid Range: End_Type 0x%x\n",
ivhd_device->alias_range.trailer.type);
return 0;
@@ -507,30 +496,28 @@ static u16 __init parse_ivhd_device_alia
first_bdf = ivhd_device->header.dev_id;
if ( first_bdf >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR,"IVHD Error: "
- "Invalid Range: First Dev_Id 0x%x\n", first_bdf);
+ amd_iov_error(
+ "IVHD Error: Invalid Range: First Dev_Id 0x%x\n", first_bdf);
return 0;
}
last_bdf = ivhd_device->alias_range.trailer.dev_id;
if ( last_bdf >= ivrs_bdf_entries || last_bdf <= first_bdf )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Range: Last Dev_Id 0x%x\n", last_bdf);
+ amd_iov_error(
+ "IVHD Error: Invalid Range: Last Dev_Id 0x%x\n", last_bdf);
return 0;
}
alias_id = ivhd_device->alias_range.alias.dev_id;
if ( alias_id >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Alias Dev_Id 0x%x\n", alias_id);
- return 0;
- }
-
- dprintk(XENLOG_INFO, " Dev_Id Range: 0x%x -> 0x%x\n",
- first_bdf, last_bdf);
- dprintk(XENLOG_INFO, " Dev_Id Alias: 0x%x\n", alias_id);
+ amd_iov_error("IVHD Error: Invalid Alias Dev_Id 0x%x\n", alias_id);
+ return 0;
+ }
+
+ amd_iov_info(" Dev_Id Range: 0x%x -> 0x%x\n", first_bdf, last_bdf);
+ amd_iov_info(" Dev_Id Alias: 0x%x\n", alias_id);
/* override requestor_id and flags for range of devices */
sys_mgt = get_field_from_byte(ivhd_device->header.flags,
@@ -555,16 +542,14 @@ static u16 __init parse_ivhd_device_exte
dev_length = sizeof(struct acpi_ivhd_device_extended);
if ( header_length < (block_length + dev_length) )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Device_Entry Length!\n");
+ amd_iov_error("IVHD Error: Invalid Device_Entry Length!\n");
return 0;
}
bdf = ivhd_device->header.dev_id;
if ( bdf >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Device_Entry Dev_Id 0x%x\n", bdf);
+ amd_iov_error("IVHD Error: Invalid Device_Entry Dev_Id 0x%x\n", bdf);
return 0;
}
@@ -587,15 +572,14 @@ static u16 __init parse_ivhd_device_exte
dev_length = sizeof(struct acpi_ivhd_device_extended_range);
if ( header_length < (block_length + dev_length) )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Device_Entry Length!\n");
+ amd_iov_error("IVHD Error: Invalid Device_Entry Length!\n");
return 0;
}
if ( ivhd_device->extended_range.trailer.type !=
AMD_IOMMU_ACPI_IVHD_DEV_RANGE_END )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
+ amd_iov_error("IVHD Error: "
"Invalid Range: End_Type 0x%x\n",
ivhd_device->extended_range.trailer.type);
return 0;
@@ -604,20 +588,20 @@ static u16 __init parse_ivhd_device_exte
first_bdf = ivhd_device->header.dev_id;
if ( first_bdf >= ivrs_bdf_entries )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Range: First Dev_Id 0x%x\n", first_bdf);
+ amd_iov_error(
+ "IVHD Error: Invalid Range: First Dev_Id 0x%x\n", first_bdf);
return 0;
}
last_bdf = ivhd_device->extended_range.trailer.dev_id;
if ( (last_bdf >= ivrs_bdf_entries) || (last_bdf <= first_bdf) )
{
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Range: Last Dev_Id 0x%x\n", last_bdf);
- return 0;
- }
-
- dprintk(XENLOG_INFO, " Dev_Id Range: 0x%x -> 0x%x\n",
+ amd_iov_error(
+ "IVHD Error: Invalid Range: Last Dev_Id 0x%x\n", last_bdf);
+ return 0;
+ }
+
+ amd_iov_info(" Dev_Id Range: 0x%x -> 0x%x\n",
first_bdf, last_bdf);
/* override flags for range of devices */
@@ -639,7 +623,7 @@ static int __init parse_ivhd_block(struc
if ( ivhd_block->header.length <
sizeof(struct acpi_ivhd_block_header) )
{
- dprintk(XENLOG_ERR, "IVHD Error: Invalid Block Length!\n");
+ amd_iov_error("IVHD Error: Invalid Block Length!\n");
return -ENODEV;
}
@@ -647,21 +631,16 @@ static int __init parse_ivhd_block(struc
ivhd_block->cap_offset);
if ( !iommu )
{
- dprintk(XENLOG_ERR,
- "IVHD Error: No IOMMU for Dev_Id 0x%x Cap 0x%x\n",
+ amd_iov_error("IVHD Error: No IOMMU for Dev_Id 0x%x Cap 0x%x\n",
ivhd_block->header.dev_id, ivhd_block->cap_offset);
return -ENODEV;
}
- dprintk(XENLOG_INFO, "IVHD Block:\n");
- dprintk(XENLOG_INFO, " Cap_Offset 0x%x\n",
- ivhd_block->cap_offset);
- dprintk(XENLOG_INFO, " MMIO_BAR_Phys 0x%lx\n",
- (unsigned long)ivhd_block->mmio_base);
- dprintk(XENLOG_INFO, " PCI_Segment 0x%x\n",
- ivhd_block->pci_segment);
- dprintk(XENLOG_INFO, " IOMMU_Info 0x%x\n",
- ivhd_block->iommu_info);
+ amd_iov_info("IVHD Block:\n");
+ amd_iov_info(" Cap_Offset 0x%x\n", ivhd_block->cap_offset);
+ amd_iov_info(" MMIO_BAR_Phys 0x%"PRIx64"\n",ivhd_block->mmio_base);
+ amd_iov_info( " PCI_Segment 0x%x\n", ivhd_block->pci_segment);
+ amd_iov_info( " IOMMU_Info 0x%x\n", ivhd_block->iommu_info);
/* override IOMMU support flags */
iommu->coherent = get_field_from_byte(ivhd_block->header.flags,
@@ -692,13 +671,10 @@ static int __init parse_ivhd_block(struc
ivhd_device = (union acpi_ivhd_device *)
((u8 *)ivhd_block + block_length);
- dprintk(XENLOG_INFO, "IVHD Device Entry:\n");
- dprintk(XENLOG_INFO, " Type 0x%x\n",
- ivhd_device->header.type);
- dprintk(XENLOG_INFO, " Dev_Id 0x%x\n",
- ivhd_device->header.dev_id);
- dprintk(XENLOG_INFO, " Flags 0x%x\n",
- ivhd_device->header.flags);
+ amd_iov_info( "IVHD Device Entry:\n");
+ amd_iov_info( " Type 0x%x\n", ivhd_device->header.type);
+ amd_iov_info( " Dev_Id 0x%x\n", ivhd_device->header.dev_id);
+ amd_iov_info( " Flags 0x%x\n", ivhd_device->header.flags);
switch ( ivhd_device->header.type )
{
@@ -741,8 +717,7 @@ static int __init parse_ivhd_block(struc
ivhd_block->header.length, block_length);
break;
default:
- dprintk(XENLOG_ERR, "IVHD Error: "
- "Invalid Device Type!\n");
+ amd_iov_error("IVHD Error: Invalid Device Type!\n");
dev_length = 0;
break;
}
@@ -774,46 +749,49 @@ static int __init parse_ivrs_block(struc
return parse_ivmd_block(ivmd_block);
default:
- dprintk(XENLOG_ERR, "IVRS Error: Invalid Block Type!\n");
+ amd_iov_error("IVRS Error: Invalid Block Type!\n");
return -ENODEV;
}
return 0;
}
-void __init dump_acpi_table_header(struct acpi_table_header *table)
-{
+static void __init dump_acpi_table_header(struct acpi_table_header *table)
+{
+#ifdef AMD_IOV_DEBUG
int i;
- printk(XENLOG_INFO "AMD IOMMU: ACPI Table:\n");
- printk(XENLOG_INFO " Signature ");
+ amd_iov_info("ACPI Table:\n");
+ amd_iov_info(" Signature ");
for ( i = 0; i < ACPI_NAME_SIZE; i++ )
printk("%c", table->signature[i]);
printk("\n");
- printk(" Length 0x%x\n", table->length);
- printk(" Revision 0x%x\n", table->revision);
- printk(" CheckSum 0x%x\n", table->checksum);
-
- printk(" OEM_Id ");
+ amd_iov_info(" Length 0x%x\n", table->length);
+ amd_iov_info(" Revision 0x%x\n", table->revision);
+ amd_iov_info(" CheckSum 0x%x\n", table->checksum);
+
+ amd_iov_info(" OEM_Id ");
for ( i = 0; i < ACPI_OEM_ID_SIZE; i++ )
printk("%c", table->oem_id[i]);
printk("\n");
- printk(" OEM_Table_Id ");
+ amd_iov_info(" OEM_Table_Id ");
for ( i = 0; i < ACPI_OEM_TABLE_ID_SIZE; i++ )
printk("%c", table->oem_table_id[i]);
printk("\n");
- printk(" OEM_Revision 0x%x\n", table->oem_revision);
-
- printk(" Creator_Id ");
+ amd_iov_info(" OEM_Revision 0x%x\n", table->oem_revision);
+
+ amd_iov_info(" Creator_Id ");
for ( i = 0; i < ACPI_NAME_SIZE; i++ )
printk("%c", table->asl_compiler_id[i]);
printk("\n");
- printk(" Creator_Revision 0x%x\n",
+ amd_iov_info(" Creator_Revision 0x%x\n",
table->asl_compiler_revision);
+#endif
+
}
int __init parse_ivrs_table(unsigned long phys_addr, unsigned long size)
@@ -827,9 +805,7 @@ int __init parse_ivrs_table(unsigned lon
BUG_ON(!table);
-#if 0
dump_acpi_table_header(table);
-#endif
/* validate checksum: sum of entire table == 0 */
checksum = 0;
@@ -838,7 +814,7 @@ int __init parse_ivrs_table(unsigned lon
checksum += raw_table[i];
if ( checksum )
{
- dprintk(XENLOG_ERR, "IVRS Error: "
+ amd_iov_error("IVRS Error: "
"Invalid Checksum 0x%x\n", checksum);
return -ENODEV;
}
@@ -850,15 +826,15 @@ int __init parse_ivrs_table(unsigned lon
ivrs_block = (struct acpi_ivrs_block_header *)
((u8 *)table + length);
- dprintk(XENLOG_INFO, "IVRS Block:\n");
- dprintk(XENLOG_INFO, " Type 0x%x\n", ivrs_block->type);
- dprintk(XENLOG_INFO, " Flags 0x%x\n", ivrs_block->flags);
- dprintk(XENLOG_INFO, " Length 0x%x\n", ivrs_block->length);
- dprintk(XENLOG_INFO, " Dev_Id 0x%x\n", ivrs_block->dev_id);
+ amd_iov_info("IVRS Block:\n");
+ amd_iov_info(" Type 0x%x\n", ivrs_block->type);
+ amd_iov_info(" Flags 0x%x\n", ivrs_block->flags);
+ amd_iov_info(" Length 0x%x\n", ivrs_block->length);
+ amd_iov_info(" Dev_Id 0x%x\n", ivrs_block->dev_id);
if ( table->length < (length + ivrs_block->length) )
{
- dprintk(XENLOG_ERR, "IVRS Error: "
+ amd_iov_error("IVRS Error: "
"Table Length Exceeded: 0x%x -> 0x%lx\n",
table->length,
(length + ivrs_block->length));
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/amd/iommu_detect.c
--- a/xen/drivers/passthrough/amd/iommu_detect.c Thu Apr 24 14:02:16
2008 -0600
+++ b/xen/drivers/passthrough/amd/iommu_detect.c Thu Apr 24 14:08:29
2008 -0600
@@ -85,6 +85,45 @@ int __init get_iommu_last_downstream_bus
return 0;
}
+static int __init get_iommu_msi_capabilities(u8 bus, u8 dev, u8 func,
+ struct amd_iommu *iommu)
+{
+ int cap_ptr, cap_id;
+ u32 cap_header;
+ u16 control;
+ int count = 0;
+
+ cap_ptr = pci_conf_read8(bus, dev, func,
+ PCI_CAPABILITY_LIST);
+
+ while ( cap_ptr >= PCI_MIN_CAP_OFFSET &&
+ count < PCI_MAX_CAP_BLOCKS )
+ {
+ cap_ptr &= PCI_CAP_PTR_MASK;
+ cap_header = pci_conf_read32(bus, dev, func, cap_ptr);
+ cap_id = get_field_from_reg_u32(cap_header,
+ PCI_CAP_ID_MASK, PCI_CAP_ID_SHIFT);
+
+ if ( cap_id == PCI_CAP_ID_MSI )
+ {
+ iommu->msi_cap = cap_ptr;
+ break;
+ }
+ cap_ptr = get_field_from_reg_u32(cap_header,
+ PCI_CAP_NEXT_PTR_MASK, PCI_CAP_NEXT_PTR_SHIFT);
+ count++;
+ }
+
+ if ( !iommu->msi_cap )
+ return -ENODEV;
+
+ amd_iov_info("Found MSI capability block \n");
+ control = pci_conf_read16(bus, dev, func,
+ iommu->msi_cap + PCI_MSI_FLAGS);
+ iommu->maskbit = control & PCI_MSI_FLAGS_MASKBIT;
+ return 0;
+}
+
int __init get_iommu_capabilities(u8 bus, u8 dev, u8 func, u8 cap_ptr,
struct amd_iommu *iommu)
{
@@ -99,8 +138,7 @@ int __init get_iommu_capabilities(u8 bus
if ( ((mmio_bar & 0x1) == 0) || (iommu->mmio_base_phys == 0) )
{
- dprintk(XENLOG_ERR ,
- "AMD IOMMU: Invalid MMIO_BAR = 0x%"PRIx64"\n", mmio_bar);
+ amd_iov_error("Invalid MMIO_BAR = 0x%"PRIx64"\n", mmio_bar);
return -ENODEV;
}
@@ -133,6 +171,8 @@ int __init get_iommu_capabilities(u8 bus
iommu->msi_number = get_field_from_reg_u32(
misc_info, PCI_CAP_MSI_NUMBER_MASK, PCI_CAP_MSI_NUMBER_SHIFT);
+ get_iommu_msi_capabilities(bus, dev, func, iommu);
+
return 0;
}
@@ -176,24 +216,24 @@ static int __init scan_functions_for_iom
int bus, int dev, iommu_detect_callback_ptr_t iommu_detect_callback)
{
int func, hdr_type;
- int count, error = 0;
-
- func = 0;
- count = 1;
- while ( VALID_PCI_VENDOR_ID(pci_conf_read16(bus, dev, func,
- PCI_VENDOR_ID)) &&
- !error && (func < count) )
+ int count = 1, error = 0;
+
+ for ( func = 0;
+ (func < count) && !error &&
+ VALID_PCI_VENDOR_ID(pci_conf_read16(bus, dev, func,
+ PCI_VENDOR_ID));
+ func++ )
+
{
hdr_type = pci_conf_read8(bus, dev, func, PCI_HEADER_TYPE);
- if ( func == 0 && IS_PCI_MULTI_FUNCTION(hdr_type) )
+ if ( (func == 0) && IS_PCI_MULTI_FUNCTION(hdr_type) )
count = PCI_MAX_FUNC_COUNT;
if ( IS_PCI_TYPE0_HEADER(hdr_type) ||
IS_PCI_TYPE1_HEADER(hdr_type) )
error = scan_caps_for_iommu(bus, dev, func,
iommu_detect_callback);
- func++;
}
return error;
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/amd/iommu_init.c
--- a/xen/drivers/passthrough/amd/iommu_init.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/passthrough/amd/iommu_init.c Thu Apr 24 14:08:29 2008 -0600
@@ -27,6 +27,7 @@
#include "../pci_regs.h"
extern int nr_amd_iommus;
+static struct amd_iommu *vector_to_iommu[NR_VECTORS];
int __init map_iommu_mmio_region(struct amd_iommu *iommu)
{
@@ -34,8 +35,7 @@ int __init map_iommu_mmio_region(struct
if ( nr_amd_iommus > MAX_AMD_IOMMUS )
{
- gdprintk(XENLOG_ERR,
- "IOMMU: nr_amd_iommus %d > MAX_IOMMUS\n", nr_amd_iommus);
+ amd_iov_error("nr_amd_iommus %d > MAX_IOMMUS\n", nr_amd_iommus);
return -ENOMEM;
}
@@ -109,6 +109,33 @@ void __init register_iommu_cmd_buffer_in
writel(entry, iommu->mmio_base+IOMMU_CMD_BUFFER_BASE_HIGH_OFFSET);
}
+void __init register_iommu_event_log_in_mmio_space(struct amd_iommu *iommu)
+{
+ u64 addr_64, addr_lo, addr_hi;
+ u32 power_of2_entries;
+ u32 entry;
+
+ addr_64 = (u64)virt_to_maddr(iommu->event_log.buffer);
+ addr_lo = addr_64 & DMA_32BIT_MASK;
+ addr_hi = addr_64 >> 32;
+
+ set_field_in_reg_u32((u32)addr_lo >> PAGE_SHIFT, 0,
+ IOMMU_EVENT_LOG_BASE_LOW_MASK,
+ IOMMU_EVENT_LOG_BASE_LOW_SHIFT, &entry);
+ writel(entry, iommu->mmio_base + IOMMU_EVENT_LOG_BASE_LOW_OFFSET);
+
+ power_of2_entries = get_order_from_bytes(iommu->event_log.alloc_size) +
+ IOMMU_EVENT_LOG_POWER_OF2_ENTRIES_PER_PAGE;
+
+ set_field_in_reg_u32((u32)addr_hi, 0,
+ IOMMU_EVENT_LOG_BASE_HIGH_MASK,
+ IOMMU_EVENT_LOG_BASE_HIGH_SHIFT, &entry);
+ set_field_in_reg_u32(power_of2_entries, entry,
+ IOMMU_EVENT_LOG_LENGTH_MASK,
+ IOMMU_EVENT_LOG_LENGTH_SHIFT, &entry);
+ writel(entry, iommu->mmio_base+IOMMU_EVENT_LOG_BASE_HIGH_OFFSET);
+}
+
static void __init set_iommu_translation_control(struct amd_iommu *iommu,
int enable)
{
@@ -179,10 +206,281 @@ static void __init register_iommu_exclus
writel(entry, iommu->mmio_base+IOMMU_EXCLUSION_BASE_LOW_OFFSET);
}
+static void __init set_iommu_event_log_control(struct amd_iommu *iommu,
+ int enable)
+{
+ u32 entry;
+
+ entry = readl(iommu->mmio_base+IOMMU_CONTROL_MMIO_OFFSET);
+ set_field_in_reg_u32(enable ? IOMMU_CONTROL_ENABLED :
+ IOMMU_CONTROL_DISABLED, entry,
+ IOMMU_CONTROL_EVENT_LOG_ENABLE_MASK,
+ IOMMU_CONTROL_EVENT_LOG_ENABLE_SHIFT, &entry);
+ writel(entry, iommu->mmio_base+IOMMU_CONTROL_MMIO_OFFSET);
+
+ set_field_in_reg_u32(enable ? IOMMU_CONTROL_ENABLED :
+ IOMMU_CONTROL_DISABLED, entry,
+ IOMMU_CONTROL_EVENT_LOG_INT_MASK,
+ IOMMU_CONTROL_EVENT_LOG_INT_SHIFT, &entry);
+ writel(entry, iommu->mmio_base+IOMMU_CONTROL_MMIO_OFFSET);
+
+ set_field_in_reg_u32(enable ? IOMMU_CONTROL_ENABLED :
+ IOMMU_CONTROL_DISABLED, entry,
+ IOMMU_CONTROL_COMP_WAIT_INT_MASK,
+ IOMMU_CONTROL_COMP_WAIT_INT_SHIFT, &entry);
+ writel(entry, iommu->mmio_base+IOMMU_CONTROL_MMIO_OFFSET);
+}
+
+static int amd_iommu_read_event_log(struct amd_iommu *iommu, u32 event[])
+{
+ u32 tail, head, *event_log;
+ int i;
+
+ BUG_ON( !iommu || !event );
+
+ /* make sure there's an entry in the log */
+ tail = get_field_from_reg_u32(
+ readl(iommu->mmio_base + IOMMU_EVENT_LOG_TAIL_OFFSET),
+ IOMMU_EVENT_LOG_TAIL_MASK,
+ IOMMU_EVENT_LOG_TAIL_SHIFT);
+ if ( tail != iommu->event_log_head )
+ {
+ /* read event log entry */
+ event_log = (u32 *)(iommu->event_log.buffer +
+ (iommu->event_log_head *
+ IOMMU_EVENT_LOG_ENTRY_SIZE));
+ for ( i = 0; i < IOMMU_EVENT_LOG_U32_PER_ENTRY; i++ )
+ event[i] = event_log[i];
+ if ( ++iommu->event_log_head == iommu->event_log.entries )
+ iommu->event_log_head = 0;
+
+ /* update head pointer */
+ set_field_in_reg_u32(iommu->event_log_head, 0,
+ IOMMU_EVENT_LOG_HEAD_MASK,
+ IOMMU_EVENT_LOG_HEAD_SHIFT, &head);
+ writel(head, iommu->mmio_base + IOMMU_EVENT_LOG_HEAD_OFFSET);
+ return 0;
+ }
+
+ return -EFAULT;
+}
+
+static void amd_iommu_msi_data_init(struct amd_iommu *iommu, int vector)
+{
+ u32 msi_data;
+ u8 bus = (iommu->bdf >> 8) & 0xff;
+ u8 dev = PCI_SLOT(iommu->bdf & 0xff);
+ u8 func = PCI_FUNC(iommu->bdf & 0xff);
+
+ msi_data = MSI_DATA_TRIGGER_EDGE |
+ MSI_DATA_LEVEL_ASSERT |
+ MSI_DATA_DELIVERY_FIXED |
+ MSI_DATA_VECTOR(vector);
+
+ pci_conf_write32(bus, dev, func,
+ iommu->msi_cap + PCI_MSI_DATA_64, msi_data);
+}
+
+static void amd_iommu_msi_addr_init(struct amd_iommu *iommu, int phy_cpu)
+{
+
+ int bus = (iommu->bdf >> 8) & 0xff;
+ int dev = PCI_SLOT(iommu->bdf & 0xff);
+ int func = PCI_FUNC(iommu->bdf & 0xff);
+
+ u32 address_hi = 0;
+ u32 address_lo = MSI_ADDR_HEADER |
+ MSI_ADDR_DESTMODE_PHYS |
+ MSI_ADDR_REDIRECTION_CPU |
+ MSI_ADDR_DESTID_CPU(phy_cpu);
+
+ pci_conf_write32(bus, dev, func,
+ iommu->msi_cap + PCI_MSI_ADDRESS_LO, address_lo);
+ pci_conf_write32(bus, dev, func,
+ iommu->msi_cap + PCI_MSI_ADDRESS_HI, address_hi);
+}
+
+static void amd_iommu_msi_enable(struct amd_iommu *iommu, int flag)
+{
+ u16 control;
+ int bus = (iommu->bdf >> 8) & 0xff;
+ int dev = PCI_SLOT(iommu->bdf & 0xff);
+ int func = PCI_FUNC(iommu->bdf & 0xff);
+
+ control = pci_conf_read16(bus, dev, func,
+ iommu->msi_cap + PCI_MSI_FLAGS);
+ control &= ~(1);
+ if ( flag )
+ control |= flag;
+ pci_conf_write16(bus, dev, func,
+ iommu->msi_cap + PCI_MSI_FLAGS, control);
+}
+
+static void iommu_msi_unmask(unsigned int vector)
+{
+ unsigned long flags;
+ struct amd_iommu *iommu = vector_to_iommu[vector];
+
+ /* FIXME: do not support mask bits at the moment */
+ if ( iommu->maskbit )
+ return;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ amd_iommu_msi_enable(iommu, IOMMU_CONTROL_ENABLED);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static void iommu_msi_mask(unsigned int vector)
+{
+ unsigned long flags;
+ struct amd_iommu *iommu = vector_to_iommu[vector];
+
+ /* FIXME: do not support mask bits at the moment */
+ if ( iommu->maskbit )
+ return;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ amd_iommu_msi_enable(iommu, IOMMU_CONTROL_DISABLED);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+static unsigned int iommu_msi_startup(unsigned int vector)
+{
+ iommu_msi_unmask(vector);
+ return 0;
+}
+
+static void iommu_msi_end(unsigned int vector)
+{
+ iommu_msi_unmask(vector);
+ ack_APIC_irq();
+}
+
+static void iommu_msi_set_affinity(unsigned int vector, cpumask_t dest)
+{
+ struct amd_iommu *iommu = vector_to_iommu[vector];
+ amd_iommu_msi_addr_init(iommu, cpu_physical_id(first_cpu(dest)));
+}
+
+static struct hw_interrupt_type iommu_msi_type = {
+ .typename = "AMD_IOV_MSI",
+ .startup = iommu_msi_startup,
+ .shutdown = iommu_msi_mask,
+ .enable = iommu_msi_unmask,
+ .disable = iommu_msi_mask,
+ .ack = iommu_msi_mask,
+ .end = iommu_msi_end,
+ .set_affinity = iommu_msi_set_affinity,
+};
+
+static void parse_event_log_entry(u32 entry[])
+{
+ u16 domain_id, device_id;
+ u32 code;
+ u64 *addr;
+ char * event_str[] = {"ILLEGAL_DEV_TABLE_ENTRY",
+ "IO_PAGE_FALT",
+ "DEV_TABLE_HW_ERROR",
+ "PAGE_TABLE_HW_ERROR",
+ "ILLEGAL_COMMAND_ERROR",
+ "COMMAND_HW_ERROR",
+ "IOTLB_INV_TIMEOUT",
+ "INVALID_DEV_REQUEST"};
+
+ code = get_field_from_reg_u32(entry[1],
+ IOMMU_EVENT_CODE_MASK,
+ IOMMU_EVENT_CODE_SHIFT);
+
+ if ( (code > IOMMU_EVENT_INVALID_DEV_REQUEST)
+ || (code < IOMMU_EVENT_ILLEGAL_DEV_TABLE_ENTRY) )
+ {
+ amd_iov_error("Invalid event log entry!\n");
+ return;
+ }
+
+ if ( code == IOMMU_EVENT_IO_PAGE_FALT )
+ {
+ device_id = get_field_from_reg_u32(entry[0],
+ IOMMU_EVENT_DEVICE_ID_MASK,
+ IOMMU_EVENT_DEVICE_ID_SHIFT);
+ domain_id = get_field_from_reg_u32(entry[1],
+ IOMMU_EVENT_DOMAIN_ID_MASK,
+ IOMMU_EVENT_DOMAIN_ID_SHIFT);
+ addr= (u64*) (entry + 2);
+ printk(XENLOG_ERR "AMD_IOV: "
+ "%s: domain:%d, device id:0x%x, fault address:0x%"PRIx64"\n",
+ event_str[code-1], domain_id, device_id, *addr);
+ }
+}
+
+static void amd_iommu_page_fault(int vector, void *dev_id,
+ struct cpu_user_regs *regs)
+{
+ u32 event[4];
+ unsigned long flags;
+ int ret = 0;
+ struct amd_iommu *iommu = dev_id;
+
+ spin_lock_irqsave(&iommu->lock, flags);
+ ret = amd_iommu_read_event_log(iommu, event);
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ if ( ret != 0 )
+ return;
+ parse_event_log_entry(event);
+}
+
+static int set_iommu_interrupt_handler(struct amd_iommu *iommu)
+{
+ int vector, ret;
+ unsigned long flags;
+
+ vector = assign_irq_vector(AUTO_ASSIGN);
+ vector_to_iommu[vector] = iommu;
+
+ /* make irq == vector */
+ irq_vector[vector] = vector;
+ vector_irq[vector] = vector;
+
+ if ( !vector )
+ {
+ amd_iov_error("no vectors\n");
+ return 0;
+ }
+
+ irq_desc[vector].handler = &iommu_msi_type;
+ ret = request_irq(vector, amd_iommu_page_fault, 0, "dmar", iommu);
+ if ( ret )
+ {
+ amd_iov_error("can't request irq\n");
+ return 0;
+ }
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
+ amd_iommu_msi_data_init (iommu, vector);
+ amd_iommu_msi_addr_init(iommu, cpu_physical_id(first_cpu(cpu_online_map)));
+ amd_iommu_msi_enable(iommu, IOMMU_CONTROL_ENABLED);
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ return vector;
+}
+
void __init enable_iommu(struct amd_iommu *iommu)
{
+ unsigned long flags;
+
+ set_iommu_interrupt_handler(iommu);
+
+ spin_lock_irqsave(&iommu->lock, flags);
+
register_iommu_exclusion_range(iommu);
set_iommu_command_buffer_control(iommu, IOMMU_CONTROL_ENABLED);
+ set_iommu_event_log_control(iommu, IOMMU_CONTROL_ENABLED);
set_iommu_translation_control(iommu, IOMMU_CONTROL_ENABLED);
- printk("AMD IOMMU %d: Enabled\n", nr_amd_iommus);
-}
+
+ spin_unlock_irqrestore(&iommu->lock, flags);
+
+ printk("AMD_IOV: IOMMU %d Enabled.\n", nr_amd_iommus);
+}
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/amd/iommu_map.c
--- a/xen/drivers/passthrough/amd/iommu_map.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/passthrough/amd/iommu_map.c Thu Apr 24 14:08:29 2008 -0600
@@ -154,8 +154,7 @@ void flush_command_buffer(struct amd_iom
}
else
{
- dprintk(XENLOG_WARNING, "AMD IOMMU: Warning:"
- " ComWaitInt bit did not assert!\n");
+ amd_iov_warning("Warning: ComWaitInt bit did not assert!\n");
}
}
}
@@ -402,10 +401,9 @@ int amd_iommu_map_page(struct domain *d,
pte = get_pte_from_page_tables(hd->root_table, hd->paging_mode, gfn);
if ( pte == NULL )
{
- dprintk(XENLOG_ERR,
- "AMD IOMMU: Invalid IO pagetable entry gfn = %lx\n", gfn);
+ amd_iov_error("Invalid IO pagetable entry gfn = %lx\n", gfn);
spin_unlock_irqrestore(&hd->mapping_lock, flags);
- return -EIO;
+ return -EFAULT;
}
set_page_table_entry_present((u32 *)pte, maddr, iw, ir);
@@ -439,10 +437,9 @@ int amd_iommu_unmap_page(struct domain *
pte = get_pte_from_page_tables(hd->root_table, hd->paging_mode, gfn);
if ( pte == NULL )
{
- dprintk(XENLOG_ERR,
- "AMD IOMMU: Invalid IO pagetable entry gfn = %lx\n", gfn);
+ amd_iov_error("Invalid IO pagetable entry gfn = %lx\n", gfn);
spin_unlock_irqrestore(&hd->mapping_lock, flags);
- return -EIO;
+ return -EFAULT;
}
/* mark PTE as 'page not present' */
@@ -479,9 +476,8 @@ int amd_iommu_reserve_domain_unity_map(
hd->root_table, hd->paging_mode, phys_addr >> PAGE_SHIFT);
if ( pte == NULL )
{
- dprintk(XENLOG_ERR,
- "AMD IOMMU: Invalid IO pagetable entry "
- "phys_addr = %lx\n", phys_addr);
+ amd_iov_error(
+ "Invalid IO pagetable entry phys_addr = %lx\n", phys_addr);
spin_unlock_irqrestore(&hd->mapping_lock, flags);
return -EFAULT;
}
@@ -528,8 +524,7 @@ int amd_iommu_sync_p2m(struct domain *d)
pte = get_pte_from_page_tables(hd->root_table, hd->paging_mode, gfn);
if ( pte == NULL )
{
- dprintk(XENLOG_ERR,
- "AMD IOMMU: Invalid IO pagetable entry gfn = %lx\n", gfn);
+ amd_iov_error("Invalid IO pagetable entry gfn = %lx\n", gfn);
spin_unlock_irqrestore(&hd->mapping_lock, flags);
return -EFAULT;
}
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/amd/pci_amd_iommu.c
--- a/xen/drivers/passthrough/amd/pci_amd_iommu.c Thu Apr 24 14:02:16
2008 -0600
+++ b/xen/drivers/passthrough/amd/pci_amd_iommu.c Thu Apr 24 14:08:29
2008 -0600
@@ -29,16 +29,11 @@ struct list_head amd_iommu_head;
struct list_head amd_iommu_head;
long amd_iommu_poll_comp_wait = COMPLETION_WAIT_DEFAULT_POLLING_COUNT;
static long amd_iommu_cmd_buffer_entries = IOMMU_CMD_BUFFER_DEFAULT_ENTRIES;
-int nr_amd_iommus = 0;
-
-unsigned short ivrs_bdf_entries = 0;
-struct ivrs_mappings *ivrs_mappings = NULL;
-
-/* will set if amd-iommu HW is found */
-int amd_iommu_enabled = 0;
-
-static int enable_amd_iommu = 0;
-boolean_param("enable_amd_iommu", enable_amd_iommu);
+static long amd_iommu_event_log_entries = IOMMU_EVENT_LOG_DEFAULT_ENTRIES;
+int nr_amd_iommus;
+
+unsigned short ivrs_bdf_entries;
+struct ivrs_mappings *ivrs_mappings;
static void deallocate_domain_page_tables(struct hvm_iommu *hd)
{
@@ -73,25 +68,8 @@ static void __init deallocate_iommu_reso
static void __init deallocate_iommu_resources(struct amd_iommu *iommu)
{
deallocate_iommu_table_struct(&iommu->dev_table);
- deallocate_iommu_table_struct(&iommu->cmd_buffer);;
-}
-
-static void __init detect_cleanup(void)
-{
- struct amd_iommu *iommu, *next;
-
- list_for_each_entry_safe ( iommu, next, &amd_iommu_head, list )
- {
- list_del(&iommu->list);
- deallocate_iommu_resources(iommu);
- xfree(iommu);
- }
-
- if ( ivrs_mappings )
- {
- xfree(ivrs_mappings);
- ivrs_mappings = NULL;
- }
+ deallocate_iommu_table_struct(&iommu->cmd_buffer);
+ deallocate_iommu_table_struct(&iommu->event_log);
}
static int __init allocate_iommu_table_struct(struct table_struct *table,
@@ -102,7 +80,7 @@ static int __init allocate_iommu_table_s
if ( !table->buffer )
{
- dprintk(XENLOG_ERR, "AMD IOMMU: Error allocating %s\n", name);
+ amd_iov_error("Error allocating %s\n", name);
return -ENOMEM;
}
@@ -139,6 +117,20 @@ static int __init allocate_iommu_resourc
"Command Buffer") != 0 )
goto error_out;
+ /* allocate 'event log' in power of 2 increments of 4K */
+ iommu->event_log_head = 0;
+ iommu->event_log.alloc_size =
+ PAGE_SIZE << get_order_from_bytes(
+ PAGE_ALIGN(amd_iommu_event_log_entries *
+ IOMMU_EVENT_LOG_ENTRY_SIZE));
+
+ iommu->event_log.entries =
+ iommu->event_log.alloc_size / IOMMU_EVENT_LOG_ENTRY_SIZE;
+
+ if ( allocate_iommu_table_struct(&iommu->event_log,
+ "Event Log") != 0 )
+ goto error_out;
+
return 0;
error_out:
@@ -153,7 +145,7 @@ int iommu_detect_callback(u8 bus, u8 dev
iommu = (struct amd_iommu *) xmalloc(struct amd_iommu);
if ( !iommu )
{
- dprintk(XENLOG_ERR, "AMD IOMMU: Error allocating amd_iommu\n");
+ amd_iov_error("Error allocating amd_iommu\n");
return -ENOMEM;
}
memset(iommu, 0, sizeof(struct amd_iommu));
@@ -203,6 +195,7 @@ static int __init amd_iommu_init(void)
goto error_out;
register_iommu_dev_table_in_mmio_space(iommu);
register_iommu_cmd_buffer_in_mmio_space(iommu);
+ register_iommu_event_log_in_mmio_space(iommu);
spin_unlock_irqrestore(&iommu->lock, flags);
}
@@ -220,18 +213,14 @@ static int __init amd_iommu_init(void)
}
if ( acpi_table_parse(ACPI_IVRS, parse_ivrs_table) != 0 )
- dprintk(XENLOG_INFO, "AMD IOMMU: Did not find IVRS table!\n");
+ amd_iov_error("Did not find IVRS table!\n");
for_each_amd_iommu ( iommu )
{
- spin_lock_irqsave(&iommu->lock, flags);
/* enable IOMMU translation services */
enable_iommu(iommu);
nr_amd_iommus++;
- spin_unlock_irqrestore(&iommu->lock, flags);
- }
-
- amd_iommu_enabled = 1;
+ }
return 0;
@@ -262,7 +251,7 @@ struct amd_iommu *find_iommu_for_device(
return NULL;
}
-void amd_iommu_setup_domain_device(
+static void amd_iommu_setup_domain_device(
struct domain *domain, struct amd_iommu *iommu, int bdf)
{
void *dte;
@@ -288,12 +277,12 @@ void amd_iommu_setup_domain_device(
sys_mgt = ivrs_mappings[req_id].dte_sys_mgt_enable;
dev_ex = ivrs_mappings[req_id].dte_allow_exclusion;
amd_iommu_set_dev_table_entry((u32 *)dte, root_ptr,
- req_id, sys_mgt, dev_ex,
+ hd->domain_id, sys_mgt, dev_ex,
hd->paging_mode);
invalidate_dev_table_entry(iommu, req_id);
flush_command_buffer(iommu);
- dprintk(XENLOG_INFO, "AMD IOMMU: Set DTE req_id:%x, "
+ amd_iov_info("Enable DTE:0x%x, "
"root_ptr:%"PRIx64", domain_id:%d, paging_mode:%d\n",
req_id, root_ptr, hd->domain_id, hd->paging_mode);
@@ -301,9 +290,9 @@ void amd_iommu_setup_domain_device(
}
}
-void __init amd_iommu_setup_dom0_devices(void)
-{
- struct hvm_iommu *hd = domain_hvm_iommu(dom0);
+static void amd_iommu_setup_dom0_devices(struct domain *d)
+{
+ struct hvm_iommu *hd = domain_hvm_iommu(d);
struct amd_iommu *iommu;
struct pci_dev *pdev;
int bus, dev, func;
@@ -333,80 +322,72 @@ void __init amd_iommu_setup_dom0_devices
find_iommu_for_device(bus, pdev->devfn) : NULL;
if ( iommu )
- amd_iommu_setup_domain_device(dom0, iommu, bdf);
+ amd_iommu_setup_domain_device(d, iommu, bdf);
}
}
}
}
-int amd_iommu_detect(void)
-{
- unsigned long i;
+int amd_iov_detect(void)
+{
int last_bus;
- struct amd_iommu *iommu;
-
- if ( !enable_amd_iommu )
- {
- printk("AMD IOMMU: Disabled\n");
- return 0;
- }
+ struct amd_iommu *iommu, *next;
INIT_LIST_HEAD(&amd_iommu_head);
if ( scan_for_iommu(iommu_detect_callback) != 0 )
{
- dprintk(XENLOG_ERR, "AMD IOMMU: Error detection\n");
+ amd_iov_error("Error detection\n");
goto error_out;
}
if ( !iommu_found() )
{
- printk("AMD IOMMU: Not found!\n");
- return 0;
- }
- else
- {
- /* allocate 'ivrs mappings' table */
- /* note: the table has entries to accomodate all IOMMUs */
- last_bus = 0;
- for_each_amd_iommu ( iommu )
- if ( iommu->last_downstream_bus > last_bus )
- last_bus = iommu->last_downstream_bus;
-
- ivrs_bdf_entries = (last_bus + 1) *
- IOMMU_DEV_TABLE_ENTRIES_PER_BUS;
- ivrs_mappings = xmalloc_array( struct ivrs_mappings, ivrs_bdf_entries);
-
- if ( !ivrs_mappings )
- {
- dprintk(XENLOG_ERR, "AMD IOMMU:"
- " Error allocating IVRS DevMappings table\n");
- goto error_out;
- }
- memset(ivrs_mappings, 0,
- ivrs_bdf_entries * sizeof(struct ivrs_mappings));
- }
+ printk("AMD_IOV: IOMMU not found!\n");
+ goto error_out;
+ }
+
+ /* allocate 'ivrs mappings' table */
+ /* note: the table has entries to accomodate all IOMMUs */
+ last_bus = 0;
+ for_each_amd_iommu ( iommu )
+ if ( iommu->last_downstream_bus > last_bus )
+ last_bus = iommu->last_downstream_bus;
+
+ ivrs_bdf_entries = (last_bus + 1) *
+ IOMMU_DEV_TABLE_ENTRIES_PER_BUS;
+ ivrs_mappings = xmalloc_array( struct ivrs_mappings, ivrs_bdf_entries);
+ if ( ivrs_mappings == NULL )
+ {
+ amd_iov_error("Error allocating IVRS DevMappings table\n");
+ goto error_out;
+ }
+ memset(ivrs_mappings, 0,
+ ivrs_bdf_entries * sizeof(struct ivrs_mappings));
if ( amd_iommu_init() != 0 )
{
- dprintk(XENLOG_ERR, "AMD IOMMU: Error initialization\n");
- goto error_out;
- }
-
- if ( iommu_domain_init(dom0) != 0 )
- goto error_out;
-
- /* setup 1:1 page table for dom0 */
- for ( i = 0; i < max_page; i++ )
- amd_iommu_map_page(dom0, i, i);
-
- amd_iommu_setup_dom0_devices();
+ amd_iov_error("Error initialization\n");
+ goto error_out;
+ }
+
return 0;
error_out:
- detect_cleanup();
+ list_for_each_entry_safe ( iommu, next, &amd_iommu_head, list )
+ {
+ list_del(&iommu->list);
+ deallocate_iommu_resources(iommu);
+ xfree(iommu);
+ }
+
+ if ( ivrs_mappings )
+ {
+ xfree(ivrs_mappings);
+ ivrs_mappings = NULL;
+ }
+
return -ENODEV;
-
}
static int allocate_domain_resources(struct hvm_iommu *hd)
@@ -447,12 +428,10 @@ static int get_paging_mode(unsigned long
return -ENOMEM;
}
- dprintk(XENLOG_INFO, "AMD IOMMU: paging mode = %d\n", level);
-
return level;
}
-int amd_iommu_domain_init(struct domain *domain)
+static int amd_iommu_domain_init(struct domain *domain)
{
struct hvm_iommu *hd = domain_hvm_iommu(domain);
@@ -463,10 +442,18 @@ int amd_iommu_domain_init(struct domain
return -ENOMEM;
}
- if ( is_hvm_domain(domain) )
- hd->paging_mode = IOMMU_PAGE_TABLE_LEVEL_4;
- else
- hd->paging_mode = get_paging_mode(max_page);
+ hd->paging_mode = is_hvm_domain(domain)?
+ IOMMU_PAGE_TABLE_LEVEL_4 : get_paging_mode(max_page);
+
+ if ( domain->domain_id == 0 )
+ {
+ unsigned long i;
+ /* setup 1:1 page table for dom0 */
+ for ( i = 0; i < max_page; i++ )
+ amd_iommu_map_page(domain, i, i);
+
+ amd_iommu_setup_dom0_devices(domain);
+ }
hd->domain_id = domain->domain_id;
@@ -490,7 +477,7 @@ static void amd_iommu_disable_domain_dev
memset (dte, 0, IOMMU_DEV_TABLE_ENTRY_SIZE);
invalidate_dev_table_entry(iommu, req_id);
flush_command_buffer(iommu);
- dprintk(XENLOG_INFO , "AMD IOMMU: disable DTE 0x%x,"
+ amd_iov_info("Disable DTE:0x%x,"
" domain_id:%d, paging_mode:%d\n",
req_id, domain_hvm_iommu(domain)->domain_id,
domain_hvm_iommu(domain)->paging_mode);
@@ -525,7 +512,7 @@ static int reassign_device( struct domai
if ( !iommu )
{
- gdprintk(XENLOG_ERR , "AMD IOMMU: fail to find iommu."
+ amd_iov_error("Fail to find iommu."
" %x:%x.%x cannot be assigned to domain %d\n",
bus, PCI_SLOT(devfn), PCI_FUNC(devfn), target->domain_id);
return -ENODEV;
@@ -540,8 +527,7 @@ static int reassign_device( struct domai
spin_unlock_irqrestore(&source_hd->iommu_list_lock, flags);
amd_iommu_setup_domain_device(target, iommu, bdf);
- gdprintk(XENLOG_INFO ,
- "AMD IOMMU: reassign %x:%x.%x domain %d -> domain %d\n",
+ amd_iov_info("reassign %x:%x.%x domain %d -> domain %d\n",
bus, PCI_SLOT(devfn), PCI_FUNC(devfn),
source->domain_id, target->domain_id);
@@ -550,7 +536,7 @@ static int reassign_device( struct domai
return 0;
}
-int amd_iommu_assign_device(struct domain *d, u8 bus, u8 devfn)
+static int amd_iommu_assign_device(struct domain *d, u8 bus, u8 devfn)
{
int bdf = (bus << 8) | devfn;
int req_id = ivrs_mappings[bdf].dte_requestor_id;
@@ -580,8 +566,7 @@ static void release_domain_devices(struc
{
pdev = list_entry(hd->pdev_list.next, typeof(*pdev), list);
pdev_flr(pdev->bus, pdev->devfn);
- gdprintk(XENLOG_INFO ,
- "AMD IOMMU: release devices %x:%x.%x\n",
+ amd_iov_info("release domain %d devices %x:%x.%x\n", d->domain_id,
pdev->bus, PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));
reassign_device(d, dom0, pdev->bus, pdev->devfn);
}
@@ -637,16 +622,13 @@ static void deallocate_iommu_page_tables
hd ->root_table = NULL;
}
-void amd_iommu_domain_destroy(struct domain *d)
-{
- if ( !amd_iommu_enabled )
- return;
-
+static void amd_iommu_domain_destroy(struct domain *d)
+{
deallocate_iommu_page_tables(d);
release_domain_devices(d);
}
-void amd_iommu_return_device(
+static void amd_iommu_return_device(
struct domain *s, struct domain *t, u8 bus, u8 devfn)
{
pdev_flr(bus, devfn);
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/iommu.c
--- a/xen/drivers/passthrough/iommu.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/passthrough/iommu.c Thu Apr 24 14:08:29 2008 -0600
@@ -18,6 +18,11 @@
extern struct iommu_ops intel_iommu_ops;
extern struct iommu_ops amd_iommu_ops;
+int intel_vtd_setup(void);
+int amd_iov_detect(void);
+
+int iommu_enabled = 1;
+boolean_param("iommu", iommu_enabled);
int iommu_domain_init(struct domain *domain)
{
@@ -134,3 +139,28 @@ void deassign_device(struct domain *d, u
return hd->platform_ops->reassign_device(d, dom0, bus, devfn);
}
+
+static int iommu_setup(void)
+{
+ int rc = -ENODEV;
+
+ if ( !iommu_enabled )
+ goto out;
+
+ switch ( boot_cpu_data.x86_vendor )
+ {
+ case X86_VENDOR_INTEL:
+ rc = intel_vtd_setup();
+ break;
+ case X86_VENDOR_AMD:
+ rc = amd_iov_detect();
+ break;
+ }
+
+ iommu_enabled = (rc == 0);
+
+ out:
+ printk("I/O virtualisation %sabled\n", iommu_enabled ? "en" : "dis");
+ return rc;
+}
+__initcall(iommu_setup);
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/vtd/dmar.c
--- a/xen/drivers/passthrough/vtd/dmar.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/passthrough/vtd/dmar.c Thu Apr 24 14:08:29 2008 -0600
@@ -30,8 +30,7 @@
#include "dmar.h"
#include "../pci_regs.h"
-int vtd_enabled;
-boolean_param("vtd", vtd_enabled);
+int vtd_enabled = 1;
#undef PREFIX
#define PREFIX VTDPREFIX "ACPI DMAR:"
@@ -79,14 +78,9 @@ struct acpi_drhd_unit * ioapic_to_drhd(u
struct acpi_drhd_unit * ioapic_to_drhd(unsigned int apic_id)
{
struct acpi_drhd_unit *drhd;
- list_for_each_entry( drhd, &acpi_drhd_units, list ) {
- if ( acpi_ioapic_device_match(&drhd->ioapic_list, apic_id) ) {
- dprintk(XENLOG_INFO VTDPREFIX,
- "ioapic_to_drhd: drhd->address = %lx\n",
- drhd->address);
+ list_for_each_entry( drhd, &acpi_drhd_units, list )
+ if ( acpi_ioapic_device_match(&drhd->ioapic_list, apic_id) )
return drhd;
- }
- }
return NULL;
}
@@ -94,15 +88,9 @@ struct iommu * ioapic_to_iommu(unsigned
{
struct acpi_drhd_unit *drhd;
- list_for_each_entry( drhd, &acpi_drhd_units, list ) {
- if ( acpi_ioapic_device_match(&drhd->ioapic_list, apic_id) ) {
- dprintk(XENLOG_INFO VTDPREFIX,
- "ioapic_to_iommu: drhd->address = %lx\n",
- drhd->address);
+ list_for_each_entry( drhd, &acpi_drhd_units, list )
+ if ( acpi_ioapic_device_match(&drhd->ioapic_list, apic_id) )
return drhd->iommu;
- }
- }
- dprintk(XENLOG_INFO VTDPREFIX, "returning NULL\n");
return NULL;
}
@@ -150,21 +138,11 @@ struct acpi_drhd_unit * acpi_find_matche
if ( acpi_pci_device_match(drhd->devices,
drhd->devices_cnt, dev) )
- {
- dprintk(XENLOG_INFO VTDPREFIX,
- "acpi_find_matched_drhd_unit: drhd->address = %lx\n",
- drhd->address);
return drhd;
- }
}
if ( include_all_drhd )
- {
- dprintk(XENLOG_INFO VTDPREFIX,
- "acpi_find_matched_drhd_unit:include_all_drhd->addr = %lx\n",
- include_all_drhd->address);
return include_all_drhd;
- }
return NULL;
}
@@ -174,11 +152,9 @@ struct acpi_rmrr_unit * acpi_find_matche
struct acpi_rmrr_unit *rmrr;
list_for_each_entry ( rmrr, &acpi_rmrr_units, list )
- {
if ( acpi_pci_device_match(rmrr->devices,
rmrr->devices_cnt, dev) )
return rmrr;
- }
return NULL;
}
@@ -199,11 +175,7 @@ struct acpi_atsr_unit * acpi_find_matche
}
if ( all_ports_atsru )
- {
- dprintk(XENLOG_INFO VTDPREFIX,
- "acpi_find_matched_atsr_unit: all_ports_atsru\n");
return all_ports_atsru;;
- }
return NULL;
}
@@ -604,22 +576,24 @@ int acpi_dmar_init(void)
{
int rc;
- if ( !vtd_enabled )
- return -ENODEV;
+ rc = -ENODEV;
+ if ( !iommu_enabled )
+ goto fail;
if ( (rc = vtd_hw_check()) != 0 )
- return rc;
+ goto fail;
acpi_table_parse(ACPI_DMAR, acpi_parse_dmar);
+ rc = -ENODEV;
if ( list_empty(&acpi_drhd_units) )
- {
- dprintk(XENLOG_ERR VTDPREFIX, "No DMAR devices found\n");
- vtd_enabled = 0;
- return -ENODEV;
- }
+ goto fail;
printk("Intel VT-d has been enabled\n");
return 0;
-}
+
+ fail:
+ vtd_enabled = 0;
+ return -ENODEV;
+}
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/vtd/iommu.c
--- a/xen/drivers/passthrough/vtd/iommu.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/passthrough/vtd/iommu.c Thu Apr 24 14:08:29 2008 -0600
@@ -41,6 +41,9 @@ static int domid_bitmap_size;
static int domid_bitmap_size; /* domain id bitmap size in bits */
static unsigned long *domid_bitmap; /* iommu domain id bitmap */
+static void setup_dom0_devices(struct domain *d);
+static void setup_dom0_rmrr(struct domain *d);
+
#define DID_FIELD_WIDTH 16
#define DID_HIGH_OFFSET 8
static void context_set_domain_id(struct context_entry *context,
@@ -78,17 +81,12 @@ static struct intel_iommu *alloc_intel_i
struct intel_iommu *intel;
intel = xmalloc(struct intel_iommu);
- if ( !intel )
- {
- gdprintk(XENLOG_ERR VTDPREFIX,
- "Allocate intel_iommu failed.\n");
+ if ( intel == NULL )
return NULL;
- }
memset(intel, 0, sizeof(struct intel_iommu));
spin_lock_init(&intel->qi_ctrl.qinval_lock);
spin_lock_init(&intel->qi_ctrl.qinval_poll_lock);
-
spin_lock_init(&intel->ir_ctrl.iremap_lock);
return intel;
@@ -96,68 +94,22 @@ static struct intel_iommu *alloc_intel_i
static void free_intel_iommu(struct intel_iommu *intel)
{
- if ( intel )
- {
- xfree(intel);
- intel = NULL;
- }
+ xfree(intel);
}
struct qi_ctrl *iommu_qi_ctrl(struct iommu *iommu)
{
- if ( !iommu )
- return NULL;
-
- if ( !iommu->intel )
- {
- iommu->intel = alloc_intel_iommu();
- if ( !iommu->intel )
- {
- dprintk(XENLOG_ERR VTDPREFIX,
- "iommu_qi_ctrl: Allocate iommu->intel failed.\n");
- return NULL;
- }
- }
-
- return &(iommu->intel->qi_ctrl);
+ return iommu ? &iommu->intel->qi_ctrl : NULL;
}
struct ir_ctrl *iommu_ir_ctrl(struct iommu *iommu)
{
- if ( !iommu )
- return NULL;
-
- if ( !iommu->intel )
- {
- iommu->intel = alloc_intel_iommu();
- if ( !iommu->intel )
- {
- dprintk(XENLOG_ERR VTDPREFIX,
- "iommu_ir_ctrl: Allocate iommu->intel failed.\n");
- return NULL;
- }
- }
-
- return &(iommu->intel->ir_ctrl);
+ return iommu ? &iommu->intel->ir_ctrl : NULL;
}
struct iommu_flush *iommu_get_flush(struct iommu *iommu)
{
- if ( !iommu )
- return NULL;
-
- if ( !iommu->intel )
- {
- iommu->intel = alloc_intel_iommu();
- if ( !iommu->intel )
- {
- dprintk(XENLOG_ERR VTDPREFIX,
- "iommu_get_flush: Allocate iommu->intel failed.\n");
- return NULL;
- }
- }
-
- return &(iommu->intel->flush);
+ return iommu ? &iommu->intel->flush : NULL;
}
unsigned int clflush_size;
@@ -276,11 +228,7 @@ static u64 addr_to_dma_page_maddr(struct
dma_set_pte_addr(*pte, maddr);
vaddr = map_vtd_domain_page(maddr);
if ( !vaddr )
- {
- unmap_vtd_domain_page(parent);
- spin_unlock_irqrestore(&hd->mapping_lock, flags);
- return 0;
- }
+ break;
/*
* high level table always sets r/w, last level
@@ -294,14 +242,9 @@ static u64 addr_to_dma_page_maddr(struct
{
vaddr = map_vtd_domain_page(pte->val);
if ( !vaddr )
- {
- unmap_vtd_domain_page(parent);
- spin_unlock_irqrestore(&hd->mapping_lock, flags);
- return 0;
- }
+ break;
}
- unmap_vtd_domain_page(parent);
if ( level == 2 )
{
pte_maddr = pte->val & PAGE_MASK_4K;
@@ -309,11 +252,13 @@ static u64 addr_to_dma_page_maddr(struct
break;
}
+ unmap_vtd_domain_page(parent);
parent = (struct dma_pte *)vaddr;
vaddr = NULL;
level--;
}
+ unmap_vtd_domain_page(parent);
spin_unlock_irqrestore(&hd->mapping_lock, flags);
return pte_maddr;
}
@@ -688,7 +633,7 @@ void dma_pte_free_pagetable(struct domai
struct dma_pte *page, *pte;
int total = agaw_to_level(hd->agaw);
int level;
- u32 tmp;
+ u64 tmp;
u64 pg_maddr;
drhd = list_entry(acpi_drhd_units.next, typeof(*drhd), list);
@@ -709,7 +654,10 @@ void dma_pte_free_pagetable(struct domai
{
pg_maddr = dma_addr_level_page_maddr(domain, tmp, level);
if ( pg_maddr == 0 )
- return;
+ {
+ tmp += level_size(level);
+ continue;
+ }
page = (struct dma_pte *)map_vtd_domain_page(pg_maddr);
pte = page + address_level_offset(tmp, level);
dma_clear_pte(*pte);
@@ -730,18 +678,11 @@ void dma_pte_free_pagetable(struct domai
}
}
-/* iommu handling */
static int iommu_set_root_entry(struct iommu *iommu)
{
u32 cmd, sts;
unsigned long flags;
-
- if ( iommu == NULL )
- {
- gdprintk(XENLOG_ERR VTDPREFIX,
- "iommu_set_root_entry: iommu == NULL\n");
- return -EINVAL;
- }
+ s_time_t start_time;
if ( iommu->root_maddr != 0 )
{
@@ -760,11 +701,14 @@ static int iommu_set_root_entry(struct i
dmar_writel(iommu->reg, DMAR_GCMD_REG, cmd);
/* Make sure hardware complete it */
+ start_time = NOW();
for ( ; ; )
{
sts = dmar_readl(iommu->reg, DMAR_GSTS_REG);
if ( sts & DMA_GSTS_RTPS )
break;
+ if ( NOW() > start_time + DMAR_OPERATION_TIMEOUT )
+ panic("DMAR hardware is malfunctional, please disable IOMMU\n");
cpu_relax();
}
@@ -777,6 +721,7 @@ static int iommu_enable_translation(stru
{
u32 sts;
unsigned long flags;
+ s_time_t start_time;
dprintk(XENLOG_INFO VTDPREFIX,
"iommu_enable_translation: iommu->reg = %p\n", iommu->reg);
@@ -784,11 +729,14 @@ static int iommu_enable_translation(stru
iommu->gcmd |= DMA_GCMD_TE;
dmar_writel(iommu->reg, DMAR_GCMD_REG, iommu->gcmd);
/* Make sure hardware complete it */
+ start_time = NOW();
for ( ; ; )
{
sts = dmar_readl(iommu->reg, DMAR_GSTS_REG);
if ( sts & DMA_GSTS_TES )
break;
+ if ( NOW() > start_time + DMAR_OPERATION_TIMEOUT )
+ panic("DMAR hardware is malfunctional, please disable IOMMU\n");
cpu_relax();
}
@@ -802,17 +750,21 @@ int iommu_disable_translation(struct iom
{
u32 sts;
unsigned long flags;
+ s_time_t start_time;
spin_lock_irqsave(&iommu->register_lock, flags);
iommu->gcmd &= ~ DMA_GCMD_TE;
dmar_writel(iommu->reg, DMAR_GCMD_REG, iommu->gcmd);
/* Make sure hardware complete it */
+ start_time = NOW();
for ( ; ; )
{
sts = dmar_readl(iommu->reg, DMAR_GSTS_REG);
if ( !(sts & DMA_GSTS_TES) )
break;
+ if ( NOW() > start_time + DMAR_OPERATION_TIMEOUT )
+ panic("DMAR hardware is malfunctional, please disable IOMMU\n");
cpu_relax();
}
spin_unlock_irqrestore(&iommu->register_lock, flags);
@@ -1039,69 +991,64 @@ int iommu_set_interrupt(struct iommu *io
return vector;
}
-struct iommu *iommu_alloc(void *hw_data)
-{
- struct acpi_drhd_unit *drhd = (struct acpi_drhd_unit *) hw_data;
+static int iommu_alloc(struct acpi_drhd_unit *drhd)
+{
struct iommu *iommu;
if ( nr_iommus > MAX_IOMMUS )
{
gdprintk(XENLOG_ERR VTDPREFIX,
"IOMMU: nr_iommus %d > MAX_IOMMUS\n", nr_iommus);
- return NULL;
+ return -ENOMEM;
}
iommu = xmalloc(struct iommu);
- if ( !iommu )
- return NULL;
+ if ( iommu == NULL )
+ return -ENOMEM;
memset(iommu, 0, sizeof(struct iommu));
+ iommu->intel = alloc_intel_iommu();
+ if ( iommu->intel == NULL )
+ {
+ xfree(iommu);
+ return -ENOMEM;
+ }
+
set_fixmap_nocache(FIX_IOMMU_REGS_BASE_0 + nr_iommus, drhd->address);
- iommu->reg = (void *) fix_to_virt(FIX_IOMMU_REGS_BASE_0 + nr_iommus);
-
- printk("iommu_alloc: iommu->reg = %p drhd->address = %lx\n",
- iommu->reg, drhd->address);
-
+ iommu->reg = (void *)fix_to_virt(FIX_IOMMU_REGS_BASE_0 + nr_iommus);
nr_iommus++;
-
- if ( !iommu->reg )
- {
- printk(KERN_ERR VTDPREFIX "IOMMU: can't mapping the region\n");
- goto error;
- }
iommu->cap = dmar_readq(iommu->reg, DMAR_CAP_REG);
iommu->ecap = dmar_readq(iommu->reg, DMAR_ECAP_REG);
- printk("iommu_alloc: cap = %"PRIx64"\n",iommu->cap);
- printk("iommu_alloc: ecap = %"PRIx64"\n", iommu->ecap);
-
spin_lock_init(&iommu->lock);
spin_lock_init(&iommu->register_lock);
- iommu->intel = alloc_intel_iommu();
-
drhd->iommu = iommu;
- return iommu;
- error:
- xfree(iommu);
- return NULL;
-}
-
-static void free_iommu(struct iommu *iommu)
-{
- if ( !iommu )
+ return 0;
+}
+
+static void iommu_free(struct acpi_drhd_unit *drhd)
+{
+ struct iommu *iommu = drhd->iommu;
+
+ if ( iommu == NULL )
return;
+
if ( iommu->root_maddr != 0 )
{
free_pgtable_maddr(iommu->root_maddr);
iommu->root_maddr = 0;
}
+
if ( iommu->reg )
iounmap(iommu->reg);
+
free_intel_iommu(iommu->intel);
free_irq(iommu->vector);
xfree(iommu);
+
+ drhd->iommu = NULL;
}
#define guestwidth_to_adjustwidth(gaw) ({ \
@@ -1111,22 +1058,21 @@ static void free_iommu(struct iommu *iom
agaw = 64; \
agaw; })
-int intel_iommu_domain_init(struct domain *domain)
-{
- struct hvm_iommu *hd = domain_hvm_iommu(domain);
+static int intel_iommu_domain_init(struct domain *d)
+{
+ struct hvm_iommu *hd = domain_hvm_iommu(d);
struct iommu *iommu = NULL;
int guest_width = DEFAULT_DOMAIN_ADDRESS_WIDTH;
- int adjust_width, agaw;
+ int i, adjust_width, agaw;
unsigned long sagaw;
struct acpi_drhd_unit *drhd;
- if ( !vtd_enabled || list_empty(&acpi_drhd_units) )
- return 0;
-
- for_each_drhd_unit ( drhd )
- iommu = drhd->iommu ? : iommu_alloc(drhd);
-
- /* calculate AGAW */
+ INIT_LIST_HEAD(&hd->pdev_list);
+
+ drhd = list_entry(acpi_drhd_units.next, typeof(*drhd), list);
+ iommu = drhd->iommu;
+
+ /* Calculate AGAW. */
if ( guest_width > cap_mgaw(iommu->cap) )
guest_width = cap_mgaw(iommu->cap);
adjust_width = guestwidth_to_adjustwidth(guest_width);
@@ -1142,6 +1088,26 @@ int intel_iommu_domain_init(struct domai
return -ENODEV;
}
hd->agaw = agaw;
+
+ if ( d->domain_id == 0 )
+ {
+ /* Set up 1:1 page table for dom0. */
+ for ( i = 0; i < max_page; i++ )
+ iommu_map_page(d, i, i);
+
+ setup_dom0_devices(d);
+ setup_dom0_rmrr(d);
+
+ iommu_flush_all();
+
+ for_each_drhd_unit ( drhd )
+ {
+ iommu = drhd->iommu;
+ if ( iommu_enable_translation(iommu) )
+ return -EIO;
+ }
+ }
+
return 0;
}
@@ -1153,28 +1119,15 @@ static int domain_context_mapping_one(
struct hvm_iommu *hd = domain_hvm_iommu(domain);
struct context_entry *context, *context_entries;
unsigned long flags;
- int ret = 0;
u64 maddr;
maddr = bus_to_context_maddr(iommu, bus);
context_entries = (struct context_entry *)map_vtd_domain_page(maddr);
context = &context_entries[devfn];
- if ( !context )
+
+ if ( context_present(*context) )
{
unmap_vtd_domain_page(context_entries);
- gdprintk(XENLOG_ERR VTDPREFIX,
- "domain_context_mapping_one:context == NULL:"
- "bdf = %x:%x:%x\n",
- bus, PCI_SLOT(devfn), PCI_FUNC(devfn));
- return -ENOMEM;
- }
-
- if ( context_present(*context) )
- {
- unmap_vtd_domain_page(context_entries);
- gdprintk(XENLOG_WARNING VTDPREFIX,
- "domain_context_mapping_one:context present:bdf=%x:%x:%x\n",
- bus, PCI_SLOT(devfn), PCI_FUNC(devfn));
return 0;
}
@@ -1203,12 +1156,6 @@ static int domain_context_mapping_one(
context_set_present(*context);
iommu_flush_cache_entry(iommu, context);
- gdprintk(XENLOG_INFO VTDPREFIX,
- "domain_context_mapping_one-%x:%x:%x-*context=%"PRIx64":%"PRIx64
- " hd->pgd_maddr=%"PRIx64"\n",
- bus, PCI_SLOT(devfn), PCI_FUNC(devfn),
- context->hi, context->lo, hd->pgd_maddr);
-
unmap_vtd_domain_page(context_entries);
if ( iommu_flush_context_device(iommu, domain_iommu_domid(domain),
@@ -1218,7 +1165,8 @@ static int domain_context_mapping_one(
else
iommu_flush_iotlb_dsi(iommu, domain_iommu_domid(domain), 0);
spin_unlock_irqrestore(&iommu->lock, flags);
- return ret;
+
+ return 0;
}
static int __pci_find_next_cap(u8 bus, unsigned int devfn, u8 pos, int cap)
@@ -1377,28 +1325,12 @@ static int domain_context_unmap_one(
maddr = bus_to_context_maddr(iommu, bus);
context_entries = (struct context_entry *)map_vtd_domain_page(maddr);
context = &context_entries[devfn];
- if ( !context )
+
+ if ( !context_present(*context) )
{
unmap_vtd_domain_page(context_entries);
- gdprintk(XENLOG_ERR VTDPREFIX,
- "domain_context_unmap_one-%x:%x:%x- context == NULL:return\n",
- bus, PCI_SLOT(devfn), PCI_FUNC(devfn));
- return -ENOMEM;
- }
-
- if ( !context_present(*context) )
- {
- unmap_vtd_domain_page(context_entries);
- gdprintk(XENLOG_WARNING VTDPREFIX,
- "domain_context_unmap_one-%x:%x:%x- "
- "context NOT present:return\n",
- bus, PCI_SLOT(devfn), PCI_FUNC(devfn));
return 0;
}
-
- gdprintk(XENLOG_INFO VTDPREFIX,
- "domain_context_unmap_one: bdf = %x:%x:%x\n",
- bus, PCI_SLOT(devfn), PCI_FUNC(devfn));
spin_lock_irqsave(&iommu->lock, flags);
context_clear_present(*context);
@@ -1431,24 +1363,12 @@ static int domain_context_unmap(
sub_bus = pci_conf_read8(
pdev->bus, PCI_SLOT(pdev->devfn),
PCI_FUNC(pdev->devfn), PCI_SUBORDINATE_BUS);
-
- gdprintk(XENLOG_INFO VTDPREFIX,
- "domain_context_unmap:BRIDGE:%x:%x:%x "
- "sec_bus=%x sub_bus=%x\n",
- pdev->bus, PCI_SLOT(pdev->devfn),
- PCI_FUNC(pdev->devfn), sec_bus, sub_bus);
break;
case DEV_TYPE_PCIe_ENDPOINT:
- gdprintk(XENLOG_INFO VTDPREFIX,
- "domain_context_unmap:PCIe : bdf = %x:%x:%x\n",
- pdev->bus, PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));
ret = domain_context_unmap_one(domain, iommu,
(u8)(pdev->bus), (u8)(pdev->devfn));
break;
case DEV_TYPE_PCI:
- gdprintk(XENLOG_INFO VTDPREFIX,
- "domain_context_unmap:PCI: bdf = %x:%x:%x\n",
- pdev->bus, PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));
if ( pdev->bus == 0 )
ret = domain_context_unmap_one(
domain, iommu,
@@ -1502,35 +1422,29 @@ void reassign_device_ownership(
int status;
unsigned long flags;
- gdprintk(XENLOG_INFO VTDPREFIX,
- "reassign_device-%x:%x:%x- source = %d target = %d\n",
- bus, PCI_SLOT(devfn), PCI_FUNC(devfn),
- source->domain_id, target->domain_id);
-
pdev_flr(bus, devfn);
for_each_pdev( source, pdev )
- {
- if ( (pdev->bus != bus) || (pdev->devfn != devfn) )
- continue;
-
- drhd = acpi_find_matched_drhd_unit(pdev);
- iommu = drhd->iommu;
- domain_context_unmap(source, iommu, pdev);
-
- /* Move pci device from the source domain to target domain. */
- spin_lock_irqsave(&source_hd->iommu_list_lock, flags);
- spin_lock_irqsave(&target_hd->iommu_list_lock, flags);
- list_move(&pdev->list, &target_hd->pdev_list);
- spin_unlock_irqrestore(&target_hd->iommu_list_lock, flags);
- spin_unlock_irqrestore(&source_hd->iommu_list_lock, flags);
-
- status = domain_context_mapping(target, iommu, pdev);
- if ( status != 0 )
- gdprintk(XENLOG_ERR VTDPREFIX, "domain_context_mapping failed\n");
-
- break;
- }
+ if ( (pdev->bus == bus) && (pdev->devfn == devfn) )
+ goto found;
+
+ return;
+
+ found:
+ drhd = acpi_find_matched_drhd_unit(pdev);
+ iommu = drhd->iommu;
+ domain_context_unmap(source, iommu, pdev);
+
+ /* Move pci device from the source domain to target domain. */
+ spin_lock_irqsave(&source_hd->iommu_list_lock, flags);
+ spin_lock_irqsave(&target_hd->iommu_list_lock, flags);
+ list_move(&pdev->list, &target_hd->pdev_list);
+ spin_unlock_irqrestore(&target_hd->iommu_list_lock, flags);
+ spin_unlock_irqrestore(&source_hd->iommu_list_lock, flags);
+
+ status = domain_context_mapping(target, iommu, pdev);
+ if ( status != 0 )
+ gdprintk(XENLOG_ERR VTDPREFIX, "domain_context_mapping failed\n");
}
void return_devices_to_dom0(struct domain *d)
@@ -1541,9 +1455,6 @@ void return_devices_to_dom0(struct domai
while ( !list_empty(&hd->pdev_list) )
{
pdev = list_entry(hd->pdev_list.next, typeof(*pdev), list);
- dprintk(XENLOG_INFO VTDPREFIX,
- "return_devices_to_dom0: bdf = %x:%x:%x\n",
- pdev->bus, PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));
reassign_device_ownership(d, dom0, pdev->bus, pdev->devfn);
}
@@ -1600,7 +1511,7 @@ int intel_iommu_map_page(
return 0;
#endif
- pg_maddr = addr_to_dma_page_maddr(d, gfn << PAGE_SHIFT_4K);
+ pg_maddr = addr_to_dma_page_maddr(d, (paddr_t)gfn << PAGE_SHIFT_4K);
if ( pg_maddr == 0 )
return -ENOMEM;
page = (struct dma_pte *)map_vtd_domain_page(pg_maddr);
@@ -1643,11 +1554,11 @@ int intel_iommu_unmap_page(struct domain
}
int iommu_page_mapping(struct domain *domain, paddr_t iova,
- void *hpa, size_t size, int prot)
+ paddr_t hpa, size_t size, int prot)
{
struct acpi_drhd_unit *drhd;
struct iommu *iommu;
- unsigned long start_pfn, end_pfn;
+ u64 start_pfn, end_pfn;
struct dma_pte *page = NULL, *pte = NULL;
int index;
u64 pg_maddr;
@@ -1657,9 +1568,8 @@ int iommu_page_mapping(struct domain *do
if ( (prot & (DMA_PTE_READ|DMA_PTE_WRITE)) == 0 )
return -EINVAL;
iova = (iova >> PAGE_SHIFT_4K) << PAGE_SHIFT_4K;
- start_pfn = (unsigned long)(((unsigned long) hpa) >> PAGE_SHIFT_4K);
- end_pfn = (unsigned long)
- ((PAGE_ALIGN_4K(((unsigned long)hpa) + size)) >> PAGE_SHIFT_4K);
+ start_pfn = hpa >> PAGE_SHIFT_4K;
+ end_pfn = (PAGE_ALIGN_4K(hpa + size)) >> PAGE_SHIFT_4K;
index = 0;
while ( start_pfn < end_pfn )
{
@@ -1668,7 +1578,7 @@ int iommu_page_mapping(struct domain *do
return -ENOMEM;
page = (struct dma_pte *)map_vtd_domain_page(pg_maddr);
pte = page + (start_pfn & LEVEL_MASK);
- dma_set_pte_addr(*pte, start_pfn << PAGE_SHIFT_4K);
+ dma_set_pte_addr(*pte, (paddr_t)start_pfn << PAGE_SHIFT_4K);
dma_set_pte_prot(*pte, prot);
iommu_flush_cache_entry(iommu, pte);
unmap_vtd_domain_page(page);
@@ -1727,7 +1637,7 @@ static int iommu_prepare_rmrr_dev(
/* page table init */
size = rmrr->end_address - rmrr->base_address + 1;
ret = iommu_page_mapping(d, rmrr->base_address,
- (void *)rmrr->base_address, size,
+ rmrr->base_address, size,
DMA_PTE_READ|DMA_PTE_WRITE);
if ( ret )
return ret;
@@ -1743,37 +1653,15 @@ static int iommu_prepare_rmrr_dev(
return ret;
}
-void __init setup_dom0_devices(void)
-{
- struct hvm_iommu *hd = domain_hvm_iommu(dom0);
+static void setup_dom0_devices(struct domain *d)
+{
+ struct hvm_iommu *hd;
struct acpi_drhd_unit *drhd;
struct pci_dev *pdev;
int bus, dev, func, ret;
u32 l;
-#ifdef DEBUG_VTD_CONTEXT_ENTRY
- for ( bus = 0; bus < 256; bus++ )
- {
- for ( dev = 0; dev < 32; dev++ )
- {
- for ( func = 0; func < 8; func++ )
- {
- struct context_entry *context;
- struct pci_dev device;
-
- device.bus = bus;
- device.devfn = PCI_DEVFN(dev, func);
- drhd = acpi_find_matched_drhd_unit(&device);
- context = device_to_context_entry(drhd->iommu,
- bus, PCI_DEVFN(dev, func));
- if ( (context->lo != 0) || (context->hi != 0) )
- dprintk(XENLOG_INFO VTDPREFIX,
- "setup_dom0_devices-%x:%x:%x- context not 0\n",
- bus, dev, func);
- }
- }
- }
-#endif
+ hd = domain_hvm_iommu(d);
for ( bus = 0; bus < 256; bus++ )
{
@@ -1792,18 +1680,13 @@ void __init setup_dom0_devices(void)
list_add_tail(&pdev->list, &hd->pdev_list);
drhd = acpi_find_matched_drhd_unit(pdev);
- ret = domain_context_mapping(dom0, drhd->iommu, pdev);
+ ret = domain_context_mapping(d, drhd->iommu, pdev);
if ( ret != 0 )
gdprintk(XENLOG_ERR VTDPREFIX,
"domain_context_mapping failed\n");
}
}
}
-
- for_each_pdev ( dom0, pdev )
- dprintk(XENLOG_INFO VTDPREFIX,
- "setup_dom0_devices: bdf = %x:%x:%x\n",
- pdev->bus, PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));
}
void clear_fault_bits(struct iommu *iommu)
@@ -1850,13 +1733,6 @@ static int init_vtd_hw(void)
flush->context = flush_context_reg;
flush->iotlb = flush_iotlb_reg;
}
- return 0;
-}
-
-static int init_vtd2_hw(void)
-{
- struct acpi_drhd_unit *drhd;
- struct iommu *iommu;
for_each_drhd_unit ( drhd )
{
@@ -1873,52 +1749,38 @@ static int init_vtd2_hw(void)
dprintk(XENLOG_ERR VTDPREFIX,
"Interrupt Remapping hardware not found\n");
}
- return 0;
-}
-
-static int enable_vtd_translation(void)
-{
- struct acpi_drhd_unit *drhd;
- struct iommu *iommu;
-
- for_each_drhd_unit ( drhd )
- {
- iommu = drhd->iommu;
- if ( iommu_enable_translation(iommu) )
- return -EIO;
- }
- return 0;
-}
-
-static void setup_dom0_rmrr(void)
+
+ return 0;
+}
+
+static void setup_dom0_rmrr(struct domain *d)
{
struct acpi_rmrr_unit *rmrr;
struct pci_dev *pdev;
int ret;
for_each_rmrr_device ( rmrr, pdev )
- ret = iommu_prepare_rmrr_dev(dom0, rmrr, pdev);
+ ret = iommu_prepare_rmrr_dev(d, rmrr, pdev);
if ( ret )
gdprintk(XENLOG_ERR VTDPREFIX,
"IOMMU: mapping reserved region failed\n");
end_for_each_rmrr_device ( rmrr, pdev )
}
-int iommu_setup(void)
-{
- struct hvm_iommu *hd = domain_hvm_iommu(dom0);
+int intel_vtd_setup(void)
+{
struct acpi_drhd_unit *drhd;
struct iommu *iommu;
- unsigned long i;
if ( !vtd_enabled )
- return 0;
+ return -ENODEV;
spin_lock_init(&domid_bitmap_lock);
- INIT_LIST_HEAD(&hd->pdev_list);
-
- /* setup clflush size */
clflush_size = get_clflush_size();
+
+ for_each_drhd_unit ( drhd )
+ if ( iommu_alloc(drhd) != 0 )
+ goto error;
/* Allocate IO page directory page for the domain. */
drhd = list_entry(acpi_drhd_units.next, typeof(*drhd), list);
@@ -1933,27 +1795,15 @@ int iommu_setup(void)
memset(domid_bitmap, 0, domid_bitmap_size / 8);
set_bit(0, domid_bitmap);
- /* setup 1:1 page table for dom0 */
- for ( i = 0; i < max_page; i++ )
- iommu_map_page(dom0, i, i);
-
init_vtd_hw();
- setup_dom0_devices();
- setup_dom0_rmrr();
- iommu_flush_all();
- enable_vtd_translation();
- init_vtd2_hw();
return 0;
error:
- printk("iommu_setup() failed\n");
for_each_drhd_unit ( drhd )
- {
- iommu = drhd->iommu;
- free_iommu(iommu);
- }
- return -EIO;
+ iommu_free(drhd);
+ vtd_enabled = 0;
+ return -ENOMEM;
}
/*
@@ -1979,10 +1829,6 @@ int intel_iommu_assign_device(struct dom
if ( list_empty(&acpi_drhd_units) )
return ret;
-
- gdprintk(XENLOG_INFO VTDPREFIX,
- "assign_device: bus = %x dev = %x func = %x\n",
- bus, PCI_SLOT(devfn), PCI_FUNC(devfn));
reassign_device_ownership(dom0, d, bus, devfn);
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/vtd/utils.c
--- a/xen/drivers/passthrough/vtd/utils.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/passthrough/vtd/utils.c Thu Apr 24 14:08:29 2008 -0600
@@ -60,10 +60,10 @@ int vtd_hw_check(void)
dprintk(XENLOG_WARNING VTDPREFIX,
"*** vendor = %x device = %x revision = %x\n",
vendor, device, revision);
- vtd_enabled = 0;
return -ENODEV;
}
}
+
return 0;
}
diff -r 239b44eeb2d6 -r dc510776dd59 xen/drivers/passthrough/vtd/x86/vtd.c
--- a/xen/drivers/passthrough/vtd/x86/vtd.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/drivers/passthrough/vtd/x86/vtd.c Thu Apr 24 14:08:29 2008 -0600
@@ -114,8 +114,6 @@ void hvm_dpci_isairq_eoi(struct domain *
if ( --dpci->mirq[i].pending == 0 )
{
spin_unlock(&dpci->dirq_lock);
- gdprintk(XENLOG_INFO VTDPREFIX,
- "hvm_dpci_isairq_eoi:: mirq = %x\n", i);
stop_timer(&dpci->hvm_timer[irq_to_vector(i)]);
pirq_guest_eoi(d, i);
}
@@ -130,8 +128,6 @@ void iommu_set_pgd(struct domain *d)
{
struct hvm_iommu *hd = domain_hvm_iommu(d);
unsigned long p2m_table;
- int level = agaw_to_level(hd->agaw);
- l3_pgentry_t *l3e;
p2m_table = mfn_x(pagetable_get_mfn(d->arch.phys_table));
@@ -153,12 +149,12 @@ void iommu_set_pgd(struct domain *d)
return;
}
pgd_mfn = _mfn(dma_pte_addr(*dpte) >> PAGE_SHIFT_4K);
- hd->pgd_maddr = mfn_x(pgd_mfn) << PAGE_SHIFT_4K;
+ hd->pgd_maddr = (paddr_t)(mfn_x(pgd_mfn)) << PAGE_SHIFT_4K;
unmap_domain_page(dpte);
break;
case VTD_PAGE_TABLE_LEVEL_4:
pgd_mfn = _mfn(p2m_table);
- hd->pgd_maddr = mfn_x(pgd_mfn) << PAGE_SHIFT_4K;
+ hd->pgd_maddr = (paddr_t)(mfn_x(pgd_mfn)) << PAGE_SHIFT_4K;
break;
default:
gdprintk(XENLOG_ERR VTDPREFIX,
@@ -173,6 +169,8 @@ void iommu_set_pgd(struct domain *d)
int i;
u64 pmd_maddr;
unsigned long flags;
+ l3_pgentry_t *l3e;
+ int level = agaw_to_level(hd->agaw);
spin_lock_irqsave(&hd->mapping_lock, flags);
hd->pgd_maddr = alloc_pgtable_maddr();
@@ -236,6 +234,8 @@ void iommu_set_pgd(struct domain *d)
#elif CONFIG_PAGING_LEVELS == 4
mfn_t pgd_mfn;
+ l3_pgentry_t *l3e;
+ int level = agaw_to_level(hd->agaw);
switch ( level )
{
@@ -250,12 +250,12 @@ void iommu_set_pgd(struct domain *d)
}
pgd_mfn = _mfn(l3e_get_pfn(*l3e));
- hd->pgd_maddr = mfn_x(pgd_mfn) << PAGE_SHIFT_4K;
+ hd->pgd_maddr = (paddr_t)(mfn_x(pgd_mfn)) << PAGE_SHIFT_4K;
unmap_domain_page(l3e);
break;
case VTD_PAGE_TABLE_LEVEL_4:
pgd_mfn = _mfn(p2m_table);
- hd->pgd_maddr = mfn_x(pgd_mfn) << PAGE_SHIFT_4K;
+ hd->pgd_maddr = (paddr_t)(mfn_x(pgd_mfn)) << PAGE_SHIFT_4K;
break;
default:
gdprintk(XENLOG_ERR VTDPREFIX,
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/asm-x86/amd-iommu.h
--- a/xen/include/asm-x86/amd-iommu.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/asm-x86/amd-iommu.h Thu Apr 24 14:08:29 2008 -0600
@@ -28,10 +28,9 @@
#define iommu_found() (!list_empty(&amd_iommu_head))
-extern int amd_iommu_enabled;
extern struct list_head amd_iommu_head;
-extern int __init amd_iommu_detect(void);
+extern int __init amd_iov_detect(void);
struct table_struct {
void *buffer;
@@ -79,6 +78,9 @@ struct amd_iommu {
int exclusion_allow_all;
uint64_t exclusion_base;
uint64_t exclusion_limit;
+
+ int msi_cap;
+ int maskbit;
};
struct ivrs_mappings {
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/asm-x86/bitops.h
--- a/xen/include/asm-x86/bitops.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/asm-x86/bitops.h Thu Apr 24 14:08:29 2008 -0600
@@ -331,10 +331,9 @@ extern unsigned int __find_next_zero_bit
extern unsigned int __find_next_zero_bit(
const unsigned long *addr, unsigned int size, unsigned int offset);
-/* return index of first bit set in val or BITS_PER_LONG when no bit is set */
-static inline unsigned int __scanbit(unsigned long val)
-{
- asm ( "bsf %1,%0" : "=r" (val) : "r" (val), "0" (BITS_PER_LONG) );
+static inline unsigned int __scanbit(unsigned long val, unsigned long max)
+{
+ asm ( "bsf %1,%0 ; cmovz %2,%0" : "=&r" (val) : "r" (val), "r" (max) );
return (unsigned int)val;
}
@@ -346,9 +345,9 @@ static inline unsigned int __scanbit(uns
* Returns the bit-number of the first set bit, not the number of the byte
* containing a bit.
*/
-#define find_first_bit(addr,size) \
-((__builtin_constant_p(size) && (size) <= BITS_PER_LONG ? \
- (__scanbit(*(const unsigned long *)addr)) : \
+#define find_first_bit(addr,size) \
+((__builtin_constant_p(size) && (size) <= BITS_PER_LONG ? \
+ (__scanbit(*(const unsigned long *)addr, size)) : \
__find_first_bit(addr,size)))
/**
@@ -357,9 +356,9 @@ static inline unsigned int __scanbit(uns
* @offset: The bitnumber to start searching at
* @size: The maximum size to search
*/
-#define find_next_bit(addr,size,off) \
-((__builtin_constant_p(size) && (size) <= BITS_PER_LONG ? \
- ((off) + (__scanbit((*(const unsigned long *)addr) >> (off)))) : \
+#define find_next_bit(addr,size,off) \
+((__builtin_constant_p(size) && (size) <= BITS_PER_LONG ? \
+ ((off) + (__scanbit((*(const unsigned long *)addr) >> (off), size))) : \
__find_next_bit(addr,size,off)))
/**
@@ -370,9 +369,9 @@ static inline unsigned int __scanbit(uns
* Returns the bit-number of the first zero bit, not the number of the byte
* containing a bit.
*/
-#define find_first_zero_bit(addr,size) \
-((__builtin_constant_p(size) && (size) <= BITS_PER_LONG ? \
- (__scanbit(~*(const unsigned long *)addr)) : \
+#define find_first_zero_bit(addr,size) \
+((__builtin_constant_p(size) && (size) <= BITS_PER_LONG ? \
+ (__scanbit(~*(const unsigned long *)addr, size)) : \
__find_first_zero_bit(addr,size)))
/**
@@ -381,9 +380,9 @@ static inline unsigned int __scanbit(uns
* @offset: The bitnumber to start searching at
* @size: The maximum size to search
*/
-#define find_next_zero_bit(addr,size,off) \
-((__builtin_constant_p(size) && (size) <= BITS_PER_LONG ? \
- ((off)+(__scanbit(~(((*(const unsigned long *)addr)) >> (off))))) : \
+#define find_next_zero_bit(addr,size,off) \
+((__builtin_constant_p(size) && (size) <= BITS_PER_LONG ? \
+ ((off)+(__scanbit(~(((*(const unsigned long *)addr)) >> (off)), size))) : \
__find_next_zero_bit(addr,size,off)))
@@ -391,8 +390,7 @@ static inline unsigned int __scanbit(uns
* find_first_set_bit - find the first set bit in @word
* @word: the word to search
*
- * Returns the bit-number of the first set bit. If no bits are set then the
- * result is undefined.
+ * Returns the bit-number of the first set bit. The input must *not* be zero.
*/
static inline unsigned int find_first_set_bit(unsigned long word)
{
@@ -401,26 +399,10 @@ static inline unsigned int find_first_se
}
/**
- * ffz - find first zero in word.
- * @word: The word to search
- *
- * Undefined if no zero exists, so code should check against ~0UL first.
- */
-static inline unsigned long ffz(unsigned long word)
-{
- asm ( "bsf %1,%0"
- :"=r" (word)
- :"r" (~word));
- return word;
-}
-
-/**
* ffs - find first bit set
* @x: the word to search
*
- * This is defined the same way as
- * the libc and compiler builtin ffs routines, therefore
- * differs in spirit from the above ffz (man ffs).
+ * This is defined the same way as the libc and compiler builtin ffs routines.
*/
static inline int ffs(unsigned long x)
{
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/asm-x86/hvm/hvm.h
--- a/xen/include/asm-x86/hvm/hvm.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/asm-x86/hvm/hvm.h Thu Apr 24 14:08:29 2008 -0600
@@ -139,6 +139,8 @@ int hvm_vcpu_initialise(struct vcpu *v);
int hvm_vcpu_initialise(struct vcpu *v);
void hvm_vcpu_destroy(struct vcpu *v);
void hvm_vcpu_down(struct vcpu *v);
+int hvm_vcpu_cacheattr_init(struct vcpu *v);
+void hvm_vcpu_cacheattr_destroy(struct vcpu *v);
void hvm_send_assist_req(struct vcpu *v);
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/asm-x86/hvm/support.h
--- a/xen/include/asm-x86/hvm/support.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/asm-x86/hvm/support.h Thu Apr 24 14:08:29 2008 -0600
@@ -130,5 +130,7 @@ int hvm_set_cr0(unsigned long value);
int hvm_set_cr0(unsigned long value);
int hvm_set_cr3(unsigned long value);
int hvm_set_cr4(unsigned long value);
+int hvm_msr_read_intercept(struct cpu_user_regs *regs);
+int hvm_msr_write_intercept(struct cpu_user_regs *regs);
#endif /* __ASM_X86_HVM_SUPPORT_H__ */
diff -r 239b44eeb2d6 -r dc510776dd59
xen/include/asm-x86/hvm/svm/amd-iommu-defs.h
--- a/xen/include/asm-x86/hvm/svm/amd-iommu-defs.h Thu Apr 24 14:02:16
2008 -0600
+++ b/xen/include/asm-x86/hvm/svm/amd-iommu-defs.h Thu Apr 24 14:08:29
2008 -0600
@@ -35,6 +35,9 @@
/* IOMMU Command Buffer entries: in power of 2 increments, minimum of 256 */
#define IOMMU_CMD_BUFFER_DEFAULT_ENTRIES 512
+/* IOMMU Event Log entries: in power of 2 increments, minimum of 256 */
+#define IOMMU_EVENT_LOG_DEFAULT_ENTRIES 512
+
#define BITMAP_ENTRIES_PER_BYTE 8
#define PTE_PER_TABLE_SHIFT 9
@@ -303,6 +306,11 @@
#define IOMMU_EVENT_COMMAND_HW_ERROR 0x6
#define IOMMU_EVENT_IOTLB_INV_TIMEOUT 0x7
#define IOMMU_EVENT_INVALID_DEV_REQUEST 0x8
+
+#define IOMMU_EVENT_DOMAIN_ID_MASK 0x0000FFFF
+#define IOMMU_EVENT_DOMAIN_ID_SHIFT 0
+#define IOMMU_EVENT_DEVICE_ID_MASK 0x0000FFFF
+#define IOMMU_EVENT_DEVICE_ID_SHIFT 0
/* Control Register */
#define IOMMU_CONTROL_MMIO_OFFSET 0x18
@@ -427,4 +435,33 @@
#define IOMMU_IO_READ_ENABLED 1
#define HACK_BIOS_SETTINGS 0
+/* MSI interrupt */
+#define MSI_DATA_VECTOR_SHIFT 0
+#define MSI_DATA_VECTOR(v) (((u8)v) << MSI_DATA_VECTOR_SHIFT)
+
+#define MSI_DATA_DELIVERY_SHIFT 8
+#define MSI_DATA_DELIVERY_FIXED (0 << MSI_DATA_DELIVERY_SHIFT)
+#define MSI_DATA_DELIVERY_LOWPRI (1 << MSI_DATA_DELIVERY_SHIFT)
+
+#define MSI_DATA_LEVEL_SHIFT 14
+#define MSI_DATA_LEVEL_DEASSERT (0 << MSI_DATA_LEVEL_SHIFT)
+#define MSI_DATA_LEVEL_ASSERT (1 << MSI_DATA_LEVEL_SHIFT)
+
+#define MSI_DATA_TRIGGER_SHIFT 15
+#define MSI_DATA_TRIGGER_EDGE (0 << MSI_DATA_TRIGGER_SHIFT)
+#define MSI_DATA_TRIGGER_LEVEL (1 << MSI_DATA_TRIGGER_SHIFT)
+
+#define MSI_TARGET_CPU_SHIFT 12
+#define MSI_ADDR_HEADER 0xfee00000
+#define MSI_ADDR_DESTID_MASK 0xfff0000f
+#define MSI_ADDR_DESTID_CPU(cpu) ((cpu) << MSI_TARGET_CPU_SHIFT)
+
+#define MSI_ADDR_DESTMODE_SHIFT 2
+#define MSI_ADDR_DESTMODE_PHYS (0 << MSI_ADDR_DESTMODE_SHIFT)
+#define MSI_ADDR_DESTMODE_LOGIC (1 << MSI_ADDR_DESTMODE_SHIFT)
+
+#define MSI_ADDR_REDIRECTION_SHIFT 3
+#define MSI_ADDR_REDIRECTION_CPU (0 << MSI_ADDR_REDIRECTION_SHIFT)
+#define MSI_ADDR_REDIRECTION_LOWPRI (1 << MSI_ADDR_REDIRECTION_SHIFT)
+
#endif /* _ASM_X86_64_AMD_IOMMU_DEFS_H */
diff -r 239b44eeb2d6 -r dc510776dd59
xen/include/asm-x86/hvm/svm/amd-iommu-proto.h
--- a/xen/include/asm-x86/hvm/svm/amd-iommu-proto.h Thu Apr 24 14:02:16
2008 -0600
+++ b/xen/include/asm-x86/hvm/svm/amd-iommu-proto.h Thu Apr 24 14:08:29
2008 -0600
@@ -35,6 +35,19 @@
#define DMA_32BIT_MASK 0x00000000ffffffffULL
#define PAGE_ALIGN(addr) (((addr) + PAGE_SIZE - 1) & PAGE_MASK)
+#ifdef AMD_IOV_DEBUG
+#define amd_iov_info(fmt, args...) \
+ printk(XENLOG_INFO "AMD_IOV: " fmt, ## args)
+#define amd_iov_warning(fmt, args...) \
+ printk(XENLOG_WARNING "AMD_IOV: " fmt, ## args)
+#define amd_iov_error(fmt, args...) \
+ printk(XENLOG_ERR "AMD_IOV: %s:%d: " fmt, __FILE__ , __LINE__ , ## args)
+#else
+#define amd_iov_info(fmt, args...)
+#define amd_iov_warning(fmt, args...)
+#define amd_iov_error(fmt, args...)
+#endif
+
typedef int (*iommu_detect_callback_ptr_t)(
u8 bus, u8 dev, u8 func, u8 cap_ptr);
@@ -49,6 +62,7 @@ void __init unmap_iommu_mmio_region(stru
void __init unmap_iommu_mmio_region(struct amd_iommu *iommu);
void __init register_iommu_dev_table_in_mmio_space(struct amd_iommu *iommu);
void __init register_iommu_cmd_buffer_in_mmio_space(struct amd_iommu *iommu);
+void __init register_iommu_event_log_in_mmio_space(struct amd_iommu *iommu);
void __init enable_iommu(struct amd_iommu *iommu);
/* mapping functions */
@@ -69,11 +83,6 @@ void invalidate_dev_table_entry(struct a
/* send cmd to iommu */
int send_iommu_command(struct amd_iommu *iommu, u32 cmd[]);
void flush_command_buffer(struct amd_iommu *iommu);
-
-/* iommu domain funtions */
-int amd_iommu_domain_init(struct domain *domain);
-void amd_iommu_setup_domain_device(struct domain *domain,
- struct amd_iommu *iommu, int bdf);
/* find iommu for bdf */
struct amd_iommu *find_iommu_for_device(int bus, int devfn);
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/asm-x86/hvm/vcpu.h
--- a/xen/include/asm-x86/hvm/vcpu.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/asm-x86/hvm/vcpu.h Thu Apr 24 14:08:29 2008 -0600
@@ -83,7 +83,16 @@ struct hvm_vcpu {
*/
unsigned long mmio_gva;
unsigned long mmio_gpfn;
+ /* Callback into x86_emulate when emulating FPU/MMX/XMM instructions. */
+ void (*fpu_exception_callback)(void *, struct cpu_user_regs *);
+ void *fpu_exception_callback_arg;
+ /* We may read up to m128 as a number of device-model transactions. */
+ paddr_t mmio_large_read_pa;
+ uint8_t mmio_large_read[16];
+ unsigned int mmio_large_read_bytes;
+ /* We may write up to m128 as a number of device-model transactions. */
+ paddr_t mmio_large_write_pa;
+ unsigned int mmio_large_write_bytes;
};
#endif /* __ASM_X86_HVM_VCPU_H__ */
-
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/asm-x86/mtrr.h
--- a/xen/include/asm-x86/mtrr.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/asm-x86/mtrr.h Thu Apr 24 14:08:29 2008 -0600
@@ -11,13 +11,6 @@
#define MTRR_TYPE_WRBACK 6
#define MTRR_NUM_TYPES 7
#define MEMORY_NUM_TYPES MTRR_NUM_TYPES
-
-#define MTRR_PHYSMASK_VALID_BIT 11
-#define MTRR_PHYSMASK_SHIFT 12
-
-#define MTRR_PHYSBASE_TYPE_MASK 0xff /* lowest 8 bits */
-#define MTRR_PHYSBASE_SHIFT 12
-#define MTRR_VCNT 8
#define NORMAL_CACHE_MODE 0
#define NO_FILL_CACHE_MODE 2
@@ -58,7 +51,6 @@ struct mtrr_state {
u64 mtrr_cap;
/* ranges in var MSRs are overlapped or not:0(no overlapped) */
bool_t overlapped;
- bool_t is_initialized;
};
extern void mtrr_save_fixed_ranges(void *);
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/asm-x86/paging.h
--- a/xen/include/asm-x86/paging.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/asm-x86/paging.h Thu Apr 24 14:08:29 2008 -0600
@@ -83,12 +83,14 @@ struct shadow_paging_mode {
unsigned long new,
unsigned int bytes,
struct sh_emulate_ctxt *sh_ctxt);
+#ifdef __i386__
int (*x86_emulate_cmpxchg8b )(struct vcpu *v, unsigned long va,
unsigned long old_lo,
unsigned long old_hi,
unsigned long new_lo,
unsigned long new_hi,
struct sh_emulate_ctxt *sh_ctxt);
+#endif
mfn_t (*make_monitor_table )(struct vcpu *v);
void (*destroy_monitor_table )(struct vcpu *v, mfn_t mmfn);
int (*guess_wrmap )(struct vcpu *v,
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/public/xsm/acm.h
--- a/xen/include/public/xsm/acm.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/public/xsm/acm.h Thu Apr 24 14:08:29 2008 -0600
@@ -91,7 +91,7 @@
* whenever the interpretation of the related
* policy's data structure changes
*/
-#define ACM_POLICY_VERSION 3
+#define ACM_POLICY_VERSION 4
#define ACM_CHWALL_VERSION 1
#define ACM_STE_VERSION 1
@@ -131,6 +131,10 @@ typedef uint16_t domaintype_t;
/* high-16 = version, low-16 = check magic */
#define ACM_MAGIC 0x0001debc
+/* size of the SHA1 hash identifying the XML policy from which the
+ binary policy was created */
+#define ACM_SHA1_HASH_SIZE 20
+
/* each offset in bytes from start of the struct they
* are part of */
@@ -160,6 +164,7 @@ struct acm_policy_buffer {
uint32_t secondary_policy_code;
uint32_t secondary_buffer_offset;
struct acm_policy_version xml_pol_version; /* add in V3 */
+ uint8_t xml_policy_hash[ACM_SHA1_HASH_SIZE]; /* added in V4 */
};
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/xen/iommu.h
--- a/xen/include/xen/iommu.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/xen/iommu.h Thu Apr 24 14:08:29 2008 -0600
@@ -27,9 +27,8 @@
#include <public/domctl.h>
extern int vtd_enabled;
-extern int amd_iommu_enabled;
+extern int iommu_enabled;
-#define iommu_enabled ( amd_iommu_enabled || vtd_enabled )
#define domain_hvm_iommu(d) (&d->arch.hvm_domain.hvm_iommu)
#define domain_vmx_iommu(d) (&d->arch.hvm_domain.hvm_iommu.vmx_iommu)
@@ -72,7 +71,6 @@ struct iommu {
struct intel_iommu *intel;
};
-int iommu_setup(void);
int iommu_domain_init(struct domain *d);
void iommu_domain_destroy(struct domain *d);
int device_assigned(u8 bus, u8 devfn);
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/xen/serial.h
--- a/xen/include/xen/serial.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/xen/serial.h Thu Apr 24 14:08:29 2008 -0600
@@ -16,12 +16,10 @@ void serial_set_rx_handler(int handle, s
void serial_set_rx_handler(int handle, serial_rx_fn fn);
/* Number of characters we buffer for a polling receiver. */
-#define SERIAL_RXBUFSZ 32
-#define MASK_SERIAL_RXBUF_IDX(_i) ((_i)&(SERIAL_RXBUFSZ-1))
+#define serial_rxbufsz 32
/* Number of characters we buffer for an interrupt-driven transmitter. */
-#define SERIAL_TXBUFSZ 16384
-#define MASK_SERIAL_TXBUF_IDX(_i) ((_i)&(SERIAL_TXBUFSZ-1))
+extern unsigned int serial_txbufsz;
struct uart_driver;
@@ -39,7 +37,7 @@ struct serial_port {
/* Receiver callback functions (asynchronous receivers). */
serial_rx_fn rx_lo, rx_hi, rx;
/* Receive data buffer (polling receivers). */
- char rxbuf[SERIAL_RXBUFSZ];
+ char rxbuf[serial_rxbufsz];
unsigned int rxbufp, rxbufc;
/* Serial I/O is concurrency-safe. */
spinlock_t rx_lock, tx_lock;
diff -r 239b44eeb2d6 -r dc510776dd59 xen/include/xsm/acm/acm_core.h
--- a/xen/include/xsm/acm/acm_core.h Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/include/xsm/acm/acm_core.h Thu Apr 24 14:08:29 2008 -0600
@@ -34,6 +34,7 @@ struct acm_binary_policy {
u16 primary_policy_code;
u16 secondary_policy_code;
struct acm_policy_version xml_pol_version;
+ u8 xml_policy_hash[ACM_SHA1_HASH_SIZE];
};
struct chwall_binary_policy {
diff -r 239b44eeb2d6 -r dc510776dd59 xen/tools/Makefile
--- a/xen/tools/Makefile Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/tools/Makefile Thu Apr 24 14:08:29 2008 -0600
@@ -4,12 +4,12 @@ include $(XEN_ROOT)/Config.mk
.PHONY: default
default:
- $(MAKE) -C figlet
+ [ -d figlet ] && $(MAKE) -C figlet
$(MAKE) symbols
.PHONY: clean
clean:
- $(MAKE) -C figlet clean
+ [ -d figlet ] && $(MAKE) -C figlet clean
rm -f *.o symbols
symbols: symbols.c
diff -r 239b44eeb2d6 -r dc510776dd59 xen/tools/figlet/figlet.c
--- a/xen/tools/figlet/figlet.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/tools/figlet/figlet.c Thu Apr 24 14:08:29 2008 -0600
@@ -1488,18 +1488,7 @@ static void myputchar(unsigned char c)
putc(c, stderr);
- if ( nr_chars == 0 )
- putchar('"');
-
- putchar('\\');
- putchar('0' + ((c>>6)&7));
- putchar('0' + ((c>>3)&7));
- putchar('0' + ((c>>0)&7));
-
- if ( c == '\n' )
- startline = 1;
-
- if ( ++nr_chars == 18 )
+ if ( nr_chars == 18 )
{
nr_chars = 0;
putchar('"');
@@ -1507,6 +1496,17 @@ static void myputchar(unsigned char c)
putchar('\\');
putchar('\n');
}
+
+ if ( nr_chars++ == 0 )
+ putchar('"');
+
+ putchar('\\');
+ putchar('0' + ((c>>6)&7));
+ putchar('0' + ((c>>3)&7));
+ putchar('0' + ((c>>0)&7));
+
+ if ( c == '\n' )
+ startline = 1;
}
void putstring(string)
diff -r 239b44eeb2d6 -r dc510776dd59 xen/xsm/acm/acm_policy.c
--- a/xen/xsm/acm/acm_policy.c Thu Apr 24 14:02:16 2008 -0600
+++ b/xen/xsm/acm/acm_policy.c Thu Apr 24 14:08:29 2008 -0600
@@ -156,6 +156,10 @@ _acm_update_policy(void *buf, u32 buf_si
&pol->xml_pol_version,
sizeof(acm_bin_pol.xml_pol_version));
+ memcpy(&acm_bin_pol.xml_policy_hash,
+ pol->xml_policy_hash,
+ sizeof(acm_bin_pol.xml_policy_hash));
+
if ( acm_primary_ops->is_default_policy() &&
acm_secondary_ops->is_default_policy() )
require_update = 0;
@@ -257,6 +261,10 @@ acm_get_policy(XEN_GUEST_HANDLE_64(void)
memcpy(&bin_pol->xml_pol_version,
&acm_bin_pol.xml_pol_version,
sizeof(struct acm_policy_version));
+
+ memcpy(&bin_pol->xml_policy_hash,
+ &acm_bin_pol.xml_policy_hash,
+ sizeof(acm_bin_pol.xml_policy_hash));
ret = acm_dump_policy_reference(
policy_buffer + be32_to_cpu(bin_pol->policy_reference_offset),
_______________________________________________
Xen-changelog mailing list
Xen-changelog@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-changelog
|
Home