| Message ID | 20240821032637.1930-1-thunder.leizhen@huawei.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [1/1] mm/slab: remove duplicate check in create_cache() | expand |
On 8/21/24 05:26, Zhen Lei wrote: > The WARN_ON() check in static function create_cache() is done by its only > parent kmem_cache_create_usercopy() before calling it. > if (... > WARN_ON(size < usersize || size - usersize < useroffset)) > usersize = useroffset = 0; > ... > s = create_cache(..., size, ..., useroffset, usersize, ...); > > Therefore, the WARN_ON() check in create_cache() can be safely removed. > > Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com> Hi, sorry for the late reply. This code was heavily refactored as of 6.12-rc1. Can you check if it's still the case and send updated patch? Thanks, Vlastimil > --- > mm/slab_common.c | 3 --- > 1 file changed, 3 deletions(-) > > diff --git a/mm/slab_common.c b/mm/slab_common.c > index 40b582a014b8f2f..3069c9095523484 100644 > --- a/mm/slab_common.c > +++ b/mm/slab_common.c > @@ -210,9 +210,6 @@ static struct kmem_cache *create_cache(const char *name, > struct kmem_cache *s; > int err; > > - if (WARN_ON(useroffset + usersize > object_size)) > - useroffset = usersize = 0; > - > err = -ENOMEM; > s = kmem_cache_zalloc(kmem_cache, GFP_KERNEL); > if (!s)
On 2024/10/2 20:09, Vlastimil Babka wrote: > On 8/21/24 05:26, Zhen Lei wrote: >> The WARN_ON() check in static function create_cache() is done by its only >> parent kmem_cache_create_usercopy() before calling it. >> if (... >> WARN_ON(size < usersize || size - usersize < useroffset)) >> usersize = useroffset = 0; >> ... >> s = create_cache(..., size, ..., useroffset, usersize, ...); >> >> Therefore, the WARN_ON() check in create_cache() can be safely removed. >> >> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com> > > Hi, sorry for the late reply. It doesn't matter, because the compiler will optimize it. It just saves the reader a little time. > This code was heavily refactored as of 6.12-rc1. Can you check if it's still > the case and send updated patch? It is still there. I'll post v2 later. > Thanks, > Vlastimil > >> --- >> mm/slab_common.c | 3 --- >> 1 file changed, 3 deletions(-) >> >> diff --git a/mm/slab_common.c b/mm/slab_common.c >> index 40b582a014b8f2f..3069c9095523484 100644 >> --- a/mm/slab_common.c >> +++ b/mm/slab_common.c >> @@ -210,9 +210,6 @@ static struct kmem_cache *create_cache(const char *name, >> struct kmem_cache *s; >> int err; >> >> - if (WARN_ON(useroffset + usersize > object_size)) >> - useroffset = usersize = 0; >> - >> err = -ENOMEM; >> s = kmem_cache_zalloc(kmem_cache, GFP_KERNEL); >> if (!s) > > . >
diff --git a/mm/slab_common.c b/mm/slab_common.c index 40b582a014b8f2f..3069c9095523484 100644 --- a/mm/slab_common.c +++ b/mm/slab_common.c @@ -210,9 +210,6 @@ static struct kmem_cache *create_cache(const char *name, struct kmem_cache *s; int err; - if (WARN_ON(useroffset + usersize > object_size)) - useroffset = usersize = 0; - err = -ENOMEM; s = kmem_cache_zalloc(kmem_cache, GFP_KERNEL); if (!s)
The WARN_ON() check in static function create_cache() is done by its only parent kmem_cache_create_usercopy() before calling it. if (... WARN_ON(size < usersize || size - usersize < useroffset)) usersize = useroffset = 0; ... s = create_cache(..., size, ..., useroffset, usersize, ...); Therefore, the WARN_ON() check in create_cache() can be safely removed. Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com> --- mm/slab_common.c | 3 --- 1 file changed, 3 deletions(-)