]> git.hungrycats.org Git - linux/commit
s390/mm: Fix VM_FAULT_HWPOISON handling in do_exception()
authorGerald Schaefer <gerald.schaefer@linux.ibm.com>
Mon, 15 Jul 2024 18:04:16 +0000 (20:04 +0200)
committerVasily Gorbik <gor@linux.ibm.com>
Wed, 17 Jul 2024 12:30:30 +0000 (14:30 +0200)
commitdf39038cd89525d465c2c8827eb64116873f141a
tree8cefc6903c2af5e7bea556844ed4a7e1ba15b5bf
parent275d05ce0680949085622bf38a365efcdd34fe11
s390/mm: Fix VM_FAULT_HWPOISON handling in do_exception()

There is no support for HWPOISON, MEMORY_FAILURE, or ARCH_HAS_COPY_MC on
s390. Therefore we do not expect to see VM_FAULT_HWPOISON in
do_exception().

However, since commit af19487f00f3 ("mm: make PTE_MARKER_SWAPIN_ERROR more
general"), it is possible to see VM_FAULT_HWPOISON in combination with
PTE_MARKER_POISONED, even on architectures that do not support HWPOISON
otherwise. In this case, we will end up on the BUG() in do_exception().

Fix this by treating VM_FAULT_HWPOISON the same as VM_FAULT_SIGBUS, similar
to x86 when MEMORY_FAILURE is not configured. Also print unexpected fault
flags, for easier debugging.

Note that VM_FAULT_HWPOISON_LARGE is not expected, because s390 cannot
support swap entries on other levels than PTE level.

Cc: stable@vger.kernel.org # 6.6+
Fixes: af19487f00f3 ("mm: make PTE_MARKER_SWAPIN_ERROR more general")
Reported-by: Yunseong Kim <yskelg@gmail.com>
Tested-by: Yunseong Kim <yskelg@gmail.com>
Acked-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Message-ID: <20240715180416.3632453-1-gerald.schaefer@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
arch/s390/mm/fault.c