[Xen-devel] Re: [PATCH] shadow-fixes.patch - Arun Sharma - com.xensource.lists.xen-devel

3 messages in com.xensource.lists.xen-devel[Xen-devel] Re: [PATCH] shadow-fixes....

From	Sent On	Attachments
Arun Sharma	28 May 2005 15:33
Michael A Fetterman	28 May 2005 18:02
Arun Sharma	30 May 2005 09:33

Subject:	[Xen-devel] Re: [PATCH] shadow-fixes.patch
From:	Arun Sharma (arun...@intel.com)
Date:	05/30/2005 09:33:05 AM
List:	com.xensource.lists.xen-devel

Michael A Fetterman wrote:

Re #1, below: can you provide a call stack when snapshot_entry_match faults? Seems like it's only ever called from __shadow_out_of_sync(), and that function tests for possible faulting conditions before invoking snapshot_entry_match(). I'm not sure I see how you could be seeing this.

This is what the stack trace looked like:

(XEN) EIP: e008:[<ff125037>] (XEN) EFLAGS: 00010246 CONTEXT: hypervisor (XEN) eax: fefa2000 ebx: fe31b000 ecx: ffbce0e0 edx: 00000000 (XEN) esi: c0000000 edi: ffbdb080 ebp: 000000b1 esp: ff103f04 (XEN) ds: e010 es: e010 fs: e010 gs: e010 ss: e010 cs: e008 (XEN) cr0: 00000004 cr3: 00000384 (XEN) Xen stack trace from esp=ff103f04: (XEN) 3d6e4000 00000c1b c0000000 ffbdb080 fe31b000 bff1b2c4 00000000 ffbdb080 (XEN) ffbda080 ffbcb080 c6cb1000 [ff124b20] ffbda080 c6cb1000 ffbda080 [ff12f601] (XEN) 07ccc063 00c1b063 01b05063 0000681e ffbda080 c6cb1000 ffbda080 [ff12f04d] (XEN) ffbda080 c6cb1000 00000002 ffbf7080 ffbda080 004c4b43 00000000 61402178 (XEN) bffff700 bffff6f8 bffff6d8 00000000 bffff6a0 00000001 00185000 c6cbd000 (XEN) c35e4448 00000100 c76daa84 [ff13600d] c6cbd000 ffffffff c6cbd000 c35e4448 (XEN) 00000100 c76daa84 c6cb1000 00000002 c36a5000 00130000 00000000 0000e008 (XEN) 00000202 00000010 00000010 00000000 00000000 00000000 ffbda080 (XEN) Xen call trace from esp=ff103f04: (XEN) [<ff124b20>] [<ff12f601>] [<ff12f04d>] [<ff13600d>]

I remember va=c6cb1000 and it was a vmexit due to invlpg -> shadow_invlpg -> __shadow_sync_va -> __shadow_out_of_sync()

I think the issue is, even though you've verified that it has a valid L2 entry, update_hl2e() hasn't been called. So linear pagetable can still fault.

-Arun