Skip to contentSkip to conversation

Backport of 353d7a84c214f18 ("powerpc/64s/radix/kfence: map __kfence_pool at page granularity") to stable versions 6.1 causes build failure

regression 6.1.y Salvatore Bonaccorso 2 participants 2 messages Jan 16 – Jan 17

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi,

In v6.1.160 353d7a84c214f18 ("powerpc/64s/radix/kfence: map __kfence_pool at page granularity") was backported. But this now causes a build failure:

make KERNELRELEASE=6.1.159+ ARCH=powerpc KBUILD_BUILD_VERSION=3 -f ./Makefile

  CALL    scripts/checksyscalls.sh
  UPD     init/utsversion-tmp.h
  CC      init/version.o
  AR      init/built-in.a
  CC      arch/powerpc/mm/book3s64/radix_pgtable.o
In file included from arch/powerpc/mm/book3s64/radix_pgtable.c:35:
./arch/powerpc/include/asm/kfence.h: In function 'kfence_protect_page':
./arch/powerpc/include/asm/kfence.h:37:9: error: implicit declaration of function '__kernel_map_pages'; did you mean 'hash__kernel_map_pages'? [-Werror=implicit-function-declaration]
   37 |         __kernel_map_pages(page, 1, !protect);
      |         ^~~~~~~~~~~~~~~~~~
      |         hash__kernel_map_pages

cc1: some warnings being treated as errors
make[7]: *** [scripts/Makefile.build:250: arch/powerpc/mm/book3s64/radix_pgtable.o] Error 1 make[6]: *** [scripts/Makefile.build:503: arch/powerpc/mm/book3s64] Error 2

This is because 8f14a96386b2 ("mm: page_poison: always declare __kernel_map_pages() function") is missing from 6.1.y (it was only included in 6.5-rc1).

Cherry-picking 8f14a96386b2 fixes the build failure on powerpc.

#regzbot introduced 36c1dc122eb3b917cd2c343029ff60a366a00539

Marking as regression for 36c1dc122eb3 ("powerpc/64s/radix/kfence: map __kfence_pool at page granularity"), for the 6.1.y specific commit causing the regression.

Regards,
Salvatore

#1 lore raw
Quoted 26 lines · Salvatore Bonaccorso

Hi,

In v6.1.160 353d7a84c214f18 ("powerpc/64s/radix/kfence: map __kfence_pool at page granularity") was backported. But this now causes a build failure:

make KERNELRELEASE=6.1.159+ ARCH=powerpc KBUILD_BUILD_VERSION=3 -f ./Makefile

  CALL    scripts/checksyscalls.sh
  UPD     init/utsversion-tmp.h
  CC      init/version.o
  AR      init/built-in.a
  CC      arch/powerpc/mm/book3s64/radix_pgtable.o
In file included from arch/powerpc/mm/book3s64/radix_pgtable.c:35:
./arch/powerpc/include/asm/kfence.h: In function 'kfence_protect_page':
./arch/powerpc/include/asm/kfence.h:37:9: error: implicit declaration of function '__kernel_map_pages'; did you mean 'hash__kernel_map_pages'? [-Werror=implicit-function-declaration]
   37 |         __kernel_map_pages(page, 1, !protect);
      |         ^~~~~~~~~~~~~~~~~~
      |         hash__kernel_map_pages

cc1: some warnings being treated as errors
make[7]: *** [scripts/Makefile.build:250: arch/powerpc/mm/book3s64/radix_pgtable.o] Error 1 make[6]: *** [scripts/Makefile.build:503: arch/powerpc/mm/book3s64] Error 2

This is because 8f14a96386b2 ("mm: page_poison: always declare __kernel_map_pages() function") is missing from 6.1.y (it was only included in 6.5-rc1).

Quoted 7 lines

Cherry-picking 8f14a96386b2 fixes the build failure on powerpc.

#regzbot introduced 36c1dc122eb3b917cd2c343029ff60a366a00539

Marking as regression for 36c1dc122eb3 ("powerpc/64s/radix/kfence: map __kfence_pool at page granularity"), for the 6.1.y specific commit causing the regression.

Now queued up, thanks.

greg k-h