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.
Re #2: I think the bug is that external mode should not set the
SHM_write_all
flag... I'll commit that fix shortly.
Michael
-----Original Message-----
From: Arun Sharma [mailto:arun.sharma@xxxxxxxxx]
Sent: Saturday, May 28, 2005 3:34 PM
To: Ian Pratt; Keir Fraser
Cc: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: [PATCH] shadow-fixes.patch
1. Indexing guest_pt[] can fault. Need to use __copy_from_user. This was
preventing FreeBSD 5.4 and RHEL3 install kernel from booting.
2. Read-only page tables should remain read only on a VMX domain. Linux 2.6
depends on getting a write fault on a L2 page table page.
Signed-off-by: Arun Sharma <arun.sharma@xxxxxxxxx>
--- 1.114/xen/arch/x86/shadow.c 2005-05-25 03:36:57 -07:00
+++ edited/xen/arch/x86/shadow.c 2005-05-28 15:19:23 -07:00
@@ -1906,7 +1906,7 @@
unsigned long gpfn, unsigned index)
{
unsigned long smfn = __shadow_status(d, gpfn, PGT_snapshot);
- l1_pgentry_t *snapshot; // could be L1s or L2s or ...
+ l1_pgentry_t *snapshot, gpte; // could be L1s or L2s or ...
int entries_match;
perfc_incrc(snapshot_entry_matches_calls);
@@ -1916,10 +1916,14 @@
snapshot = map_domain_mem(smfn << PAGE_SHIFT);
+ if (__copy_from_user(&gpte, &guest_pt[index],
+ sizeof(gpte)))
+ return 0;
+
// This could probably be smarter, but this is sufficent for
// our current needs.
//
- entries_match = !l1e_has_changed(&guest_pt[index], &snapshot[index],
+ entries_match = !l1e_has_changed(&gpte, &snapshot[index],
PAGE_FLAG_MASK);
unmap_domain_mem(snapshot);
@@ -2600,7 +2604,8 @@
if ( unlikely(!(l1e_get_flags(gpte) & _PAGE_RW)) )
{
- if ( shadow_mode_page_writable(d, l1e_get_pfn(gpte)) )
+ if ( shadow_mode_page_writable(d, l1e_get_pfn(gpte))
+ && !shadow_mode_external(d))
{
allow_writes = 1;
l1e_add_flags(&gpte, _PAGE_RW);
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home