Re: [Xen-devel] pci-passthrough not working on 4.0.1-rc1-pre: mm.c:3847:d10 Bad page 00000000001beab4: ed=ffff830151210000(10), sd=ffff830151210000, caf=8000000000000002, taf=7400000000000001

[Sander Eikelenboom <linux@xxxxxxxxxxxxxx>]

Hi Konrad,

Think i have found it, these domains where running old debian 2.6.26 kernels, 
based on 2.6.18.
Perhaps something recently on pciback has broken compatibility with that ?
Will try the other guest with a pvops kernel as well shortly.

--
Sander



Monday, April 19, 2010, 7:09:23 PM, you wrote:

> Hi Konrad,

> I don't use the framebuffer with "vfb", just use xm console or ssh to access 
> the guests.
> The strange thing is, i'm trying to passingthrough 1 soundcard to 1 guest, 
> and 2 USB cards to another.

> The bad page shows only for the domain with the 2 USB cards.
> The guest with the soundcard doesn't give a badpage error, but with lspci 
> there still is now pci device visible in the guest.
> After rebuilding what should be my last working setup, and enabling debug for 
> passthrough in the kernel dmesg shows:

> for the domain with the sound card:
> [  163.544664] blkback: ring-ref 2047, event-channel 4, protocol 1 
> (x86_64-abi)
> [  163.560049] pciback pci-7-0: fe state changed 3
> [  163.560405] pciback pci-7-0: Reading frontend config
> [  163.560787] pciback pci-7-0: Attaching to frontend resources - 
> gnt_ref=2046 evtchn=5
> [  163.560839] pciback pci-7-0: Attached!
> [  163.560842] pciback pci-7-0: Connecting...
> [  163.561234] pciback pci-7-0: Connected? 0
> [  163.562391] blkback: ring-ref 2045, event-channel 6, protocol 1 
> (x86_64-abi)
> [  163.568221] pciback pci-7-0: fe state changed 4
> [  169.727473] frontend_changed: backend/vbd/7/2049: prepare for reconnect
> [  169.733315] frontend_changed: backend/vbd/7/2050: prepare for reconnect
> [  169.922523] blkback: ring-ref 8, event-channel 8, protocol 1 (x86_64-abi)
> [  169.929428] blkback: ring-ref 9, event-channel 9, protocol 1 (x86_64-abi)
> [  173.880008] vif7.0: no IPv6 routers present
> [  178.148010] xen_bridge: port 7(vif7.0) entering forwarding state


> for the domain with the 2 usb cards:

> [  230.323946] pciback pci-10-0: allocated pdev @ 0xffff88002eb60b40
> [  230.326866] pciback pci-10-0: getting be setup
> [  230.327668] pciback pci-10-0: exporting dom 0 bus 1 slot 8 func 0
> [  230.329067] pciback 0000:01:08.0: registering for 10
> [  230.330259] pciback pci-10-0: exporting dom 0 bus 1 slot 8 func 1
> [  230.331564] pciback 0000:01:08.1: registering for 10
> [  230.333510] pciback pci-10-0: exporting dom 0 bus 1 slot 8 func 2
> [  230.335060] pciback 0000:01:08.2: registering for 10
> [  230.337069] pciback pci-10-0: exporting dom 0 bus 9 slot 1 func 0
> [  230.339290] pciback 0000:09:01.0: registering for 10
> [  230.340986] pciback pci-10-0: exporting dom 0 bus 9 slot 1 func 1
> [  230.342140] pciback 0000:09:01.1: registering for 10
> [  230.343800] pciback pci-10-0: exporting dom 0 bus 9 slot 1 func 2
> [  230.344784] pciback 0000:09:01.2: registering for 10
> [  230.346331] pciback pci-10-0: Publishing pci roots
> [  230.346583] pciback pci-10-0: writing root 0 at 0000:01
> [  230.350401] pciback pci-10-0: Publishing pci roots
> [  230.351142] pciback pci-10-0: Publishing pci roots
> [  230.352203] pciback pci-10-0: Publishing pci roots
> [  230.352944] pciback pci-10-0: writing root 1 at 0000:09
> [  230.355628] pciback pci-10-0: Publishing pci roots
> [  230.356827] pciback pci-10-0: Publishing pci roots
> [  230.390018] pciback pci-10-0: fe state changed 1
> [  231.541480] blkback: ring-ref 2047, event-channel 4, protocol 1 
> (x86_64-abi)
> [  231.552080] pciback pci-10-0: fe state changed 3
> [  231.552306] pciback pci-10-0: Reading frontend config
> [  231.552678] pciback pci-10-0: Attaching to frontend resources - 
> gnt_ref=2046 evtchn=5
> [  231.552746] pciback pci-10-0: Attached!
> [  231.552749] pciback pci-10-0: Connecting...
> [  231.555412] pciback pci-10-0: Connected? 0
> [  231.558960] pciback pci-10-0: fe state changed 4
> [  231.561640] blkback: ring-ref 2045, event-channel 6, protocol 1 
> (x86_64-abi)
> [  231.570789] blkback: ring-ref 2044, event-channel 7, protocol 1 
> (x86_64-abi)
> [  237.811338] frontend_changed: backend/vbd/10/2049: prepare for reconnect
> [  237.818039] frontend_changed: backend/vbd/10/2050: prepare for reconnect
> [  237.824610] frontend_changed: backend/vbd/10/2065: prepare for reconnect
> [  238.243595] blkback: ring-ref 8, event-channel 12, protocol 1 (x86_64-abi)
> [  238.250822] blkback: ring-ref 9, event-channel 13, protocol 1 (x86_64-abi)
> [  238.258358] blkback: ring-ref 10, event-channel 14, protocol 1 (x86_64-abi)
> [  240.108012] vif10.0: no IPv6 routers present
> [  245.084015] xen_bridge: port 10(vif10.0) entering forwarding state


> and xm dmesg:

> (XEN)   _PSD: num_entries=5 rev=0 domain=0 coord_type=253 num_processors=2
> (XEN)   _PPC: 0
> (XEN) printk: 11 messages suppressed.
> (XEN) mm.c:3847:d10 Bad page 000000000013eaba: ed=ffff83013f7b0000(10), 
> sd=ffff83013f7b0000, caf=8000000000000002, taf=7400000000000001
> (XEN) mm.c:3847:d10 Bad page 000000000013eaba: ed=ffff83013f7b0000(10), 
> sd=ffff83013f7b0000, caf=8000000000000002, taf=7400000000000001
> (XEN) mm.c:3847:d10 Bad page 000000000013eaba: ed=ffff83013f7b0000(10), 
> sd=ffff83013f7b0000, caf=8000000000000002, taf=7400000000000001
> (XEN) mm.c:3847:d10 Bad page 000000000013eaba: ed=ffff83013f7b0000(10), 
> sd=ffff83013f7b0000, caf=8000000000000002, taf=7400000000000001
> (XEN) mm.c:3847:d10 Bad page 000000000013eaba: ed=ffff83013f7b0000(10), 
> sd=ffff83013f7b0000, caf=8000000000000002, taf=7400000000000001
> (XEN) mm.c:3847:d10 Bad page 000000000013eaba: ed=ffff83013f7b0000(10), 
> sd=ffff83013f7b0000, caf=8000000000000002, taf=7400000000000001

> --

> Sander




> Monday, April 19, 2010, 6:27:36 PM, you wrote:

>> On Mon, Apr 19, 2010 at 06:13:42PM +0200, Sander Eikelenboom wrote:
>>> Hi Konrad,
>>> 
>>> I'm using PV-domains and not HVM, does that use the QEMU fb as well ?

>> If you have 'vfb=[..]' then QEMU is involved - just to provide you with
>> a VNC version of a framebuffer.
>>> 
>>> What i did notice is CONFIG_XEN_GNTDEV seems to be missing from the "latest 
>>> working" kernel source .config
>>> Is this required for the passthrough ?

>> No. But it is required by QEMU.







-- 
Best regards,
 Sander                            mailto:linux@xxxxxxxxxxxxxx


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel