On Mon, 24 Sep 2007 11:44:35 +0400, Alexey Dobriyan said: > Interesting. Here is output from kernel with patch applied and leak > plugged into proc_dointvec() (I checked twice): > > $ grep kcalloc /proc/slab_allocators > $ grep proc_dointvec /proc/slab_allocators > size-64: 19 proc_dointvec+0x48/0xa0 Argh, went to "cancel" that previous msg and retry when I realized you meant it was with *your* patch installed, managed to send it anyhow.. I wonder if the only reason it's working for you is because gcc is doing some tail-recursion magic on the kcalloc() call (which calls to __kmalloc), which just happens to make it work, and if the tail-recursion wasn't happening it would be busticated.
Attachment:
pgpYQQd7bhNAJ.pgp
Description: PGP signature
- References:
- [PATCH] Uninline kcalloc()
- From: Alexey Dobriyan <[email protected]>
- Re: [PATCH] Uninline kcalloc()
- From: [email protected]
- Re: [PATCH] Uninline kcalloc()
- From: "Alexey Dobriyan" <[email protected]>
- [PATCH] Uninline kcalloc()
- Prev by Date: Re: 2.6.23-rc6-mm1: IPC: sleeping function called ...
- Next by Date: Re: [patches] [PATCH] [13/50] x86: Fix and reenable CLFLUSH support inchange_page_attr()
- Previous by thread: Re: [PATCH] Uninline kcalloc()
- Next by thread: Re: [PATCH] Uninline kcalloc()
- Index(es):