Relax assertion in VRAM tracking code
PATCH 1/1 Relax assertion in VRAM tracking code
The original assertion is too strict, as it includes the A/D bits of the PTE, which (by design) can change under our feet.
xen/arch/x86/mm/shadow/multi.c +2 -1
@@ -1117,7 +1117,8 @@ }else { -ASSERT (dirty_vram->sl1ma[i] == sl1ma); +ASSERT ((dirty_vram->sl1ma[i] &PAGE_MASK ) + == (sl1ma &PAGE_MASK )); dirty_vram->sl1ma[i] =INVALID_PADDR ;if ( flags & _PAGE_DIRTY ) dirty =1 ;
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel
Quoted 5 lines
Quoted 1 line
Quoted 1 line
Tim Deegan <[email protected]> 16.11.09 18:00 >>>
The original assertion is too strict, as it includes the A/D bits of the PTE, which (by design) can change under our feet.
Signed-off-by Tim Deegan <[email protected]>
Wouldn't the comparison a few lines down from the ASSERT() thus fixed also benefit from a similar adjustment?
Jan
Quoted 8 lines · Jan Beulich
Quoted 5 lines
Quoted 1 line
Quoted 1 line
Tim Deegan <[email protected]> 16.11.09 18:00 >>>
The original assertion is too strict, as it includes the A/D bits of the PTE, which (by design) can change under our feet.
Signed-off-by Tim Deegan <[email protected]>Wouldn't the comparison a few lines down from the
ASSERT()thus fixed also benefit from a similar adjustment?
Actually, on closer inspection my original fix was wrong; the bottom 12 bits of sl1ma are taken from the _pointer_ to the sl1e, not the contents, so the assertion is plausible.
Keir, can you please revert it?
Tim.
Signature · 3 lines
Tim Deegan <[email protected]> Principal Software Engineer, Citrix Systems (R&D) Ltd. [Company #02300071, SL9 0DZ, UK.]