Re: [PATCH] Fix two potential mem leaks in MPT Fusion (mpt_attach())

[Andrew Morton <akpm@linux-foundation.org>]

On Fri, 3 Aug 2007 00:53:44 +0200
Jesper Juhl <jesper.juhl@gmail.com> wrote:

> On Thursday 02 August 2007 10:20:47 Jesper Juhl wrote:
> > On 02/08/07, Andrew Morton <akpm@linux-foundation.org> wrote:
> [snip]
> > > y'know, we could have a debug option which will spit warnings if someone
> > > does a !__GFP_WAIT allocation while !in_atomic() (only works if
> > > CONFIG_PREEMPT).
> > >
> > > But please, make it depend on !CONFIG_AKPM.  I shudder to think about all
> > > the stuff it would pick up.
> > >
> > 
> > I can try to cook up something like that tonight...
> > 
> 
> Ok, so I did a quick hack and I'm drowning in dmesg WARN_ON() traces 
> with my usual config.
> 
> This is what I added : 
> 
> diff --git a/mm/slub.c b/mm/slub.c
> index 6c6d74f..e60dd9e 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -20,6 +20,7 @@
>  #include <linux/mempolicy.h>
>  #include <linux/ctype.h>
>  #include <linux/kallsyms.h>
> +#include <linux/hardirq.h>
>  
>  /*
>   * Lock order:
> @@ -1568,6 +1569,10 @@ static void __always_inline *slab_alloc(struct kmem_cache *s,
>  
>  void *kmem_cache_alloc(struct kmem_cache *s, gfp_t gfpflags)
>  {
> +#ifdef CONFIG_PREEMPT
> +	WARN_ON( !in_atomic() && !(gfpflags & __GFP_WAIT) );
> +#endif
> +
>  	return slab_alloc(s, gfpflags, -1, __builtin_return_address(0));
>  }
>  EXPORT_SYMBOL(kmem_cache_alloc);
> @@ -2370,6 +2375,10 @@ void *__kmalloc(size_t size, gfp_t flags)
>  {
>  	struct kmem_cache *s = get_slab(size, flags);
>  
> +#ifdef CONFIG_PREEMPT
> +	WARN_ON( !in_atomic() && !(flags & __GFP_WAIT) );
> +#endif
> +
>  	if (ZERO_OR_NULL_PTR(s))
>  		return s;
>  
> 
> 
> And this is what I'm getting heaps of : 
> 
> ...
> [  165.128607]  =======================
> [  165.128609] WARNING: at mm/slub.c:1573 kmem_cache_alloc()
> [  165.128611]  [<c010400a>] show_trace_log_lvl+0x1a/0x30
> [  165.128614]  [<c0104cd2>] show_trace+0x12/0x20
> [  165.128616]  [<c0104cf6>] dump_stack+0x16/0x20
> [  165.128619]  [<c0175ad3>] kmem_cache_alloc+0xe3/0x110
> [  165.128622]  [<c015b10e>] mempool_alloc_slab+0xe/0x10
> [  165.128625]  [<c015b211>] mempool_alloc+0x31/0xf0

I said you would.

> So, where do we go from here?

Where I said ;) Add a new __GFP_ flag which suppresses the warning, add
that flag to known-to-be-OK callsites, such as mempool_alloc().