Skip to contentSkip to conversation

slub: don't panic for memcg kmem cache creation failure

DISCUSSION Alexey Dobriyan 2 participants 2 messages Jun 29, 2019

lore
Share & tools
lore
#0 lore raw

Review

Quoted 4 lines
(quoted diff) +0 -4
-       if (flags & SLAB_PANIC)
-               panic("Cannot create slab %s size=%u realsize=%u order=%u offset=%u flags=%lx\n",
-                     s->name, s->size, s->size,
-                     oo_order(s->oo), s->offset, (unsigned long)flags);

This is wrong. Without SLAB_PANIC people will start to implement error checking out of habit and add all slightly different error messages. This simply increases text and rodata size.

If memcg kmem caches creation failure is OK, then SLAB_PANIC should not be passed.

The fact that SLAB doesn't implement SLAB_PANIC is SLAB bug.

#1 lore raw
Quoted 6 lines · Alexey Dobriyan

Review

Quoted 5 lines
(quoted diff) +0 -4
-       if (flags & SLAB_PANIC)
-               panic("Cannot create slab %s size=%u realsize=%u order=%u offset=%u flags=%lx\n",
-                     s->name, s->size, s->size,
-                     oo_order(s->oo), s->offset, (unsigned long)flags);
Quoted 8 lines

This is wrong. Without SLAB_PANIC people will start to implement error checking out of habit and add all slightly different error messages. This simply increases text and rodata size.

If memcg kmem caches creation failure is OK, then SLAB_PANIC should not be passed.

The fact that SLAB doesn't implement SLAB_PANIC is SLAB bug.

I do not agree with you. IMHO the kmem_cache_create_usercopy() is the right place to handle SLAB_PANIC which is handling it. If you want extra info here, you can add pr_warn for SLAB_PANIC here and the caller will still and rightfully do the panic().

thanks,
Shakeel