Re: [PATCH] block: fix 'kmem_cache of name 'bio-108' already exists'

[Ming Lei <ming.lei@redhat.com>]

On Mon, Mar 03, 2025 at 07:39:34AM -0800, Christoph Hellwig wrote:
> On Fri, Feb 28, 2025 at 09:26:56PM +0800, Ming Lei wrote:
> > Device mapper bioset often has big bio_slab size, which can be more than
> > 1000, then 8byte can't hold the slab name any more, cause the kmem_cache
> > allocation warning of 'kmem_cache of name 'bio-108' already exists'.
> > 
> > Fix the warning by extending bio_slab->name to 12 bytes, but fix output
> > of /proc/slabinfo
> > 
> > Reported-by: Guangwu Zhang <guazhang@redhat.com>
> > Signed-off-by: Ming Lei <ming.lei@redhat.com>
> > ---
> >  block/bio.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/block/bio.c b/block/bio.c
> > index f0c416e5931d..6ac5983ba51e 100644
> > --- a/block/bio.c
> > +++ b/block/bio.c
> > @@ -77,7 +77,7 @@ struct bio_slab {
> >  	struct kmem_cache *slab;
> >  	unsigned int slab_ref;
> >  	unsigned int slab_size;
> > -	char name[8];
> > +	char name[12];
> 
> Can you please turn this into a pointer and use kasprintf to fill
> it?  That way we fix the string overflow problem for real and don't
> need to doctor around it the next time someone uses names with a
> longer name.

There isn't the overflow problem, please see create_bio_slab(), which
calls snprintf() to fill the ->name[].

Also extra 4byte can support bio size of ~10^7, which is big enough as
block layer API.


Thanks,
Ming