Hi Sergey,
On Wed, 8 Jun 2022 at 13:06, Sergey Senozhatsky senozhatsky@chromium.org wrote:
On (22/06/08 11:45), Sergey Senozhatsky wrote:
Something like this?
May be even something like below. Move static initializer to cpu up hook.
Thanks for providing a fix patch. As per your suggestion debug lock enabled.
CONFIG_DEBUG_LOCK_ALLOC=y
I have tested this patch and the reported issue got resolved [1].
Reported-by: Linux Kernel Functional Testing lkft@linaro.org Tested-by: Linux Kernel Functional Testing lkft@linaro.org
mm/zsmalloc.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c index 9152fbde33b5..6d3789d834e2 100644 --- a/mm/zsmalloc.c +++ b/mm/zsmalloc.c @@ -453,9 +453,7 @@ MODULE_ALIAS("zpool-zsmalloc"); #endif /* CONFIG_ZPOOL */
/* per-cpu VM mapping areas for zspage accesses that cross page boundaries */ -static DEFINE_PER_CPU(struct mapping_area, zs_map_area) = {
.lock = INIT_LOCAL_LOCK(lock),-}; +static DEFINE_PER_CPU(struct mapping_area, zs_map_area);
static __maybe_unused int is_first_page(struct page *page) { @@ -1113,6 +1111,7 @@ static inline int __zs_cpu_up(struct mapping_area *area) area->vm_buf = kmalloc(ZS_MAX_ALLOC_SIZE, GFP_KERNEL); if (!area->vm_buf) return -ENOMEM;
local_lock_init(&area->lock); return 0;}
2.36.1.255.ge46751e96f-goog
Build artifacts link [2]
-- Linaro LKFT https://lkft.linaro.org
[1] https://lkft.validation.linaro.org/scheduler/job/5160519#L828 [2] https://builds.tuxbuild.com/2ATpwm83jTe2wlUX33YeTpshh2u/