> On Oct 16, 2019, at 6:59 PM, Jerry Snitselaar wrote:
>
> I guess the mode level 6 check is really for other potential callers
> increase_address_space, none exist at the moment, and the condition
> of the while loop in alloc_pte should fail if the mode level is 6.
Because there is no
On Wed Oct 16 19, Jerry Snitselaar wrote:
On Wed Oct 16 19, Qian Cai wrote:
BTW, Joerg, this line from the commit "iommu/amd: Remove domain->updated" looks
suspicious. Not sure what the purpose of it.
*updated = increase_address_space(domain, gfp) || *updated;
Looking at it again I think
On Wed Oct 16 19, Qian Cai wrote:
After the commit 754265bcab78 ("iommu/amd: Fix race in
increase_address_space()"), it could still possible trigger a race
condition under some heavy memory pressure below. The race to trigger a
warning is,
CPU0: CPU1:
in alloc_pte():
After the commit 754265bcab78 ("iommu/amd: Fix race in
increase_address_space()"), it could still possible trigger a race
condition under some heavy memory pressure below. The race to trigger a
warning is,
CPU0: CPU1:
in alloc_pte(): in increase_address_space():
