Hi All,

Chages since v7:
- drop "unnecessary free pages" optimization
- fix error path page leak

Chages since v6:
- do not unnecessary free pages across iterations

Chages since v5:
- full error message included into commit description

Chages since v4:
- unused pages leak is avoided

Chages since v3:
- pfn_to_virt() changed to page_to_virt() due to compile error

Chages since v2:
- page allocation moved out of the atomic context

Chages since v1:
- Fixes: and -stable tags added to the patch description

Thanks!

Alexander Gordeev (1):
  kasan: Avoid sleepable page allocation from atomic context

 mm/kasan/shadow.c | 77 ++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 63 insertions(+), 14 deletions(-)

-- 
2.45.2

On 5/13/25 5:21 PM, Alexander Gordeev wrote:
> Hi All,
> 
> Chages since v7:
> - drop "unnecessary free pages" optimization
> - fix error path page leak
> 
> Chages since v6:
> - do not unnecessary free pages across iterations
> 


Have you looked at boot failure report from kernel test robot ?
https://lkml.kernel.org/r/202505121313.806a632c-lkp@intel.com

I think the report is for v6 version, but I don't see evidence that it was
addressed, so the v8 is probably affected as well?


> Chages since v5:
> - full error message included into commit description
> 
> Chages since v4:
> - unused pages leak is avoided
> 
> Chages since v3:
> - pfn_to_virt() changed to page_to_virt() due to compile error
> 
> Chages since v2:
> - page allocation moved out of the atomic context
> 
> Chages since v1:
> - Fixes: and -stable tags added to the patch description
> 
> Thanks!
> 
> Alexander Gordeev (1):
>   kasan: Avoid sleepable page allocation from atomic context
> 
>  mm/kasan/shadow.c | 77 ++++++++++++++++++++++++++++++++++++++---------
>  1 file changed, 63 insertions(+), 14 deletions(-)
>

On Tue, May 13, 2025 at 06:43:56PM +0200, Andrey Ryabinin wrote:
> Have you looked at boot failure report from kernel test robot ?
> https://lkml.kernel.org/r/202505121313.806a632c-lkp@intel.com
> 
> I think the report is for v6 version, but I don't see evidence that it was
> addressed, so the v8 is probably affected as well?

Yes. The problem is page_owner=on prevents bulk allcations.
I will send an updated version.

Thanks!