From: Qiujun Huang hqjagain@gmail.com
commit 99b3b837855b987563bcfb397cf9ddd88262814b upstream.
There is a case found when triggering a panic_on_oom, pstore fails to dump kmsg. Because psz_kmsg_write_record can't get the new buffer.
Handle this by using GFP_ATOMIC to allocate a buffer at lower watermark.
Signed-off-by: Qiujun Huang hqjagain@gmail.com Fixes: 335426c6dcdd ("pstore/zone: Provide way to skip "broken" zone for MTD devices") Cc: WeiXiong Liao gmpy.liaowx@gmail.com Cc: stable@vger.kernel.org Signed-off-by: Kees Cook keescook@chromium.org Link: https://lore.kernel.org/r/CAJRQjofRCF7wjrYmw3D7zd5QZnwHQq+F8U-mJDJ6NZ4bddYdL... Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- fs/pstore/zone.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
--- a/fs/pstore/zone.c +++ b/fs/pstore/zone.c @@ -761,7 +761,7 @@ static inline int notrace psz_kmsg_write /* avoid destroying old data, allocate a new one */ len = zone->buffer_size + sizeof(*zone->buffer); zone->oldbuf = zone->buffer; - zone->buffer = kzalloc(len, GFP_KERNEL); + zone->buffer = kzalloc(len, GFP_ATOMIC); if (!zone->buffer) { zone->buffer = zone->oldbuf; return -ENOMEM;