Re: [PATCH] VM: kswapd should not do blocking memory allocations

[Trond Myklebust <Trond.Myklebust@netapp.com>]

On Wed, 2010-08-18 at 12:24 -0700, Ram Pai wrote:
> 
> 
> On Wed, Aug 18, 2010 at 12:04 PM, Trond Myklebust
> <Trond.Myklebust@netapp.com> wrote:
>         From: Trond Myklebust <Trond.Myklebust@netapp.com>
>         
>         Allowing kswapd to do GFP_KERNEL memory allocations (or any
>         blocking memory
>         allocations) is wrong and can cause deadlocks in
>         try_to_release_page(), as
>         the filesystem believes it is safe to allocate new memory and
>         block,
>         whereas kswapd is there specifically to clear a low-memory
>         situation...
>         
>         Set the gfp_mask to GFP_IOFS instead.
>         
>         Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
>         ---
>         
>          mm/vmscan.c |    2 +-
>          1 files changed, 1 insertions(+), 1 deletions(-)
>         
>         
>         diff --git a/mm/vmscan.c b/mm/vmscan.c
>         index ec5ddcc..716dd16 100644
>         --- a/mm/vmscan.c
>         +++ b/mm/vmscan.c
>         @@ -2095,7 +2095,7 @@ static unsigned long
>         balance_pgdat(pg_data_t *pgdat, int order)
>                unsigned long total_scanned;
>                struct reclaim_state *reclaim_state =
>         current->reclaim_state;
>                struct scan_control sc = {
>         -               .gfp_mask = GFP_KERNEL,
>         +               .gfp_mask = GFP_IOFS,
>                        .may_unmap = 1,
>                        .may_swap = 1,
>                        /*
> 
> Trond,
> 
>            Has anyone hit this issue? Or is this based on code
> inspection?  
> 
>            The reason I  ask is we are seeing a problem, similar to
> the symptom described, on RH based kernel but have not been able to
> reproduce on 2.6.35.

Hi Ram,

I was seeing it on NFS until I put in the following kswapd-specific hack
into nfs_release_page():

	/* Only do I/O if gfp is a superset of GFP_KERNEL */
	if (mapping && (gfp & GFP_KERNEL) == GFP_KERNEL) {
		int how = FLUSH_SYNC;

		/* Don't let kswapd deadlock waiting for OOM RPC calls */
		if (current_is_kswapd())
			how = 0;
		nfs_commit_inode(mapping->host, how);
	}

Remove the 'if (current_is_kswapd())' line, and run an mmap() write
intensive workload, and it should hang pretty much every time.

Cheers
  Trond

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>