Re: 2.2.10-ac10 nfs Oops (Another one for the slab poisoner) (fwd)
H.J. Lu (hjl@lucon.org)
Mon, 19 Jul 1999 17:14:01 -0700 (PDT)
>
> > Please send me patch to enable kmalloc and slab debugging.
>
> No problem. This patch is x86 only. If you want to use it on Alpha I need
> to send you some other fixes for kernel problems.
>
> Slab poison writes 0x5a5a5a and similar patterns over memory at kfree time
> so that it traps situations where you kfree something then reference it shortly
> afterwards
>
>
>
> --- linux.vanilla/mm/slab.c Thu Dec 3 13:48:38 1998
> +++ linux.ac/mm/slab.c Fri Jul 2 19:10:21 1999
> @@ -125,7 +125,7 @@
> * SLAB_SELFTEST - 1 to perform a few tests, mainly for development.
> */
> #define SLAB_MGMT_CHECKS 1
> -#define SLAB_DEBUG_SUPPORT 0
> +#define SLAB_DEBUG_SUPPORT 1
> #define SLAB_STATS 0
> #define SLAB_SELFTEST 0
>
> @@ -471,7 +471,7 @@
> * allow tighter packing of the smaller caches. */
> if (!(sizes->cs_cachep =
> kmem_cache_create(*names++, sizes->cs_size,
> - 0, SLAB_HWCACHE_ALIGN, NULL, NULL)))
> + 0, SLAB_HWCACHE_ALIGN|SLAB_POISON, NULL, NULL)))
> goto panic_time;
> if (!found) {
> /* Inc off-slab bufctl limit until the ceiling is hit. */
>
I tried this patch + 2.2.10 + patches from knfsd 1.4.6. It works just
fine for me. However, I did see the "Stale NFS file handle" problem.
My conclusions are
1. NFS client in 2.2.10 is broken.
2. Both NFS server and client in 2.2.10-ac10 are broken.
I will take a look at #1 and I will leave #2 to others.
--
H.J. Lu (hjl@gnu.org)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/