summaryrefslogtreecommitdiffstats
path: root/arch/s390/kvm
diff options
context:
space:
mode:
authorHeiko Carstens <hca@linux.ibm.com>2023-08-25 14:29:53 +0200
committerHeiko Carstens <hca@linux.ibm.com>2023-08-30 11:03:28 +0200
commita7eb28801b692df2496e21ff9af6ab0dc1969133 (patch)
tree567ad596b5fe74d42c24731db0bccc5c4856224c /arch/s390/kvm
parent850612c8e4fb048905af597cbd25dfbb40d9ffdf (diff)
downloadlinux-stable-a7eb28801b692df2496e21ff9af6ab0dc1969133.tar.gz
linux-stable-a7eb28801b692df2496e21ff9af6ab0dc1969133.tar.bz2
linux-stable-a7eb28801b692df2496e21ff9af6ab0dc1969133.zip
s390/mm: use __set_memory() variants where useful
Use the __set_memory_yy() variants instead of set_memory_yy() where useful. This allows to make the code a bit more readable. This also fixes the debug pagealloc case, where set_memory_4k() might be called for an area larger than 8TB which would lead to an overflow of the num_pages parameter of set_memory_4k(). However RELOC_HIDE() has to be used for the __set_memory_4k() case for the time being, to avoid compiler warnings because of performing pointer arithmetic on a NULL pointer, which has undefined behavior. This happens because __va(0) always translates to NULL. However this will change, and as soon as this happens the RELOC_HIDE() hack can be removed again. Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com> Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Diffstat (limited to 'arch/s390/kvm')
0 files changed, 0 insertions, 0 deletions