RE: [Xen-ia64-devel] RE: Patches for ia64 arch_vcpu_info_t merge

["Magenheimer, Dan (HP Labs Fort Collins)" <dan.magenheimer@xxxxxx>]

Some good news, some bad.  With your patch applied, domain0
boots to a login prompt.  However, Matt Chapman's recently
added multiple domain support crashes when creating a new
domain.  Here's the start of the error messages.  Perhaps
we can work on this together after the summit meeting.

arch_vcpu_info=f0000000040e0000
(XEN) ###allocating rid_range, domain f000000007ff9458:
starting_rid=80000, ending_rid=c0000
(XEN) arch_getdomaininfo_ctxt
(XEN) *** xen_handle_domain_access: exception table lookup failed,
iip=f00000000 

> -----Original Message-----
> From: Dong, Eddie [mailto:eddie.dong@xxxxxxxxx] 
> Sent: Wednesday, July 20, 2005 12:53 PM
> To: Magenheimer, Dan (HP Labs Fort Collins); Yang, Fred
> Cc: xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
> Subject: RE: [Xen-ia64-devel] RE: Patches for ia64 
> arch_vcpu_info_t merge
> 
> Magenheimer, Dan (HP Labs Fort Collins) wrote:
> > Is there a reason that the passing of the shared memory pointer
> > in r31 has been disabled by the patch?  It isn't used much right now
> > but I intend to use it more in the future.
> > 
> Dan:
>       Yes, no strong reason and only for simpilification as
> &((shared_info_t *)SHAREDINFO_ADDR)->vcpu_data[0].arch is not point to
> ipsr and ipsr is not 1st address of shared_archinfo now. I will let it
> there. But the key thing now is that I don't know why the 
> patch fail in
> your previous test tree, can u have a try base on the latest patch and
> let us know much detail failure message so that Kevin can 
> debug early in
> case the problem is still there.
>       I guess the problem will not be there :-)
> 
> Kevin:
>       Can you try following manual patch and send dan new patch?
> Eddei
> 
> diff -r ea8caf38c914 xen/arch/ia64/process.c
> --- a/xen/arch/ia64/process.c Wed Jul 13 20:18:04 2005
> +++ b/xen/arch/ia64/process.c Thu Jul 14 14:42:21 2005
> @@ -226,7 +226,7 @@
>  #ifdef CONFIG_SMP
>  #error "sharedinfo doesn't handle smp yet"
>  #endif
> -     regs->r31 = &((shared_info_t
> *)SHAREDINFO_ADDR)->vcpu_data[0].arch;
> +     regs->r31 = &((arch_vcpu_info_t *)SHARED_ARCHINFO_ADDR)->ipsr);
>  
>       PSCB(v,interrupt_delivery_enabled) = 0;
>       PSCB(v,interrupt_collection_enabled) = 0; 
> 
> 
> 
> 
> And following patch is not necessary now:
> diff -r 5fcf5a856e25 arch/ia64/xen/xenivt.S
> --- a/arch/ia64/xen/xenivt.S  Wed Jul 13 22:37:53 2005
> +++ b/arch/ia64/xen/xenivt.S  Thu Jul 14 14:34:31 2005
> @@ -407,6 +407,8 @@
>  ENTRY(alt_itlb_miss)
>       DBG_FAULT(3)
>  #ifdef CONFIG_XEN
> +        movl r31=XSI_IPSR
> +     ;;
>       ld8 r21=[r31],XSI_IFA-XSI_IPSR  // get ipsr, point to ifa
>       movl r17=PAGE_KERNEL
>       ;;
> @@ -471,6 +473,8 @@
>  ENTRY(alt_dtlb_miss)
>       DBG_FAULT(4)
>  #ifdef CONFIG_XEN
> +        movl r31=XSI_IPSR
> +     ;;
>       ld8 r21=[r31],XSI_ISR-XSI_IPSR  // get ipsr, point to isr
>       movl r17=PAGE_KERNEL
>       ;;
> @@ -964,6 +969,8 @@
>        */
>       DBG_FAULT(11)
>  #ifdef CONFIG_XEN
> +        movl r31=XSI_IPSR
> +     ;;
>       ld8 r29=[r31],XSI_IIP-XSI_IPSR          // get ipsr, point to
> iip
>       mov r18=__IA64_BREAK_SYSCALL
>       mov r21=ar.fpsr
> 

_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel