Date:	Sun, 03 Oct 2010 14:41:09 -0400
From:	Nitin Gupta <ngupta@...are.org>
To:	Dave Hansen <dave@...ux.vnet.ibm.com>
CC:	Pekka Enberg <penberg@...helsinki.fi>,
	Minchan Kim <minchan.kim@...il.com>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Greg KH <greg@...ah.com>,
	Linux Driver Project <devel@...uxdriverproject.org>,
	linux-mm <linux-mm@...ck.org>,
	linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: OOM panics with zram

Hi Dave,

Sorry for late reply. Since last month I couldn't get any chance to
work on this project.

On 9/9/2010 1:24 PM, Dave Hansen wrote:
> 
> I've been playing with using zram (from -staging) to back some qemu
> guest memory directly.  Basically mmap()'ing the device in instead of
> using anonymous memory.  The old code with the backing swap devices
> seemed to work pretty well, but I'm running into a problem with the new
> code.
> 
> I have plenty of swap on the system, and I'd been running with compcache
> nicely for a while.  But, I went to go tar up (and gzip) a pretty large
> directory in my qemu guest.  It panic'd the qemu host system:
> 
> [703826.003126] Kernel panic - not syncing: Out of memory and no killable processes...
> [703826.003127] 
> [703826.012350] Pid: 25508, comm: cat Not tainted 2.6.36-rc3-00114-g9b9913d #29
> [703826.019385] Call Trace:
> [703826.021928]  [<ffffffff8104032a>] panic+0xba/0x1e0
> [703826.026801]  [<ffffffff810bb4a1>] ? next_online_pgdat+0x21/0x50
> [703826.032799]  [<ffffffff810a7713>] ? find_lock_task_mm+0x23/0x60
> [703826.038795]  [<ffffffff810a79ab>] ? dump_header+0x19b/0x1b0
> [703826.044446]  [<ffffffff810a8157>] out_of_memory+0x297/0x2d0
> [703826.050098]  [<ffffffff810abbaf>] __alloc_pages_nodemask+0x72f/0x740
> [703826.056528]  [<ffffffff81110d4e>] ? __set_page_dirty+0x6e/0xc0
> [703826.062438]  [<ffffffff810da477>] alloc_pages_current+0x87/0xd0
> [703826.068438]  [<ffffffff810a533b>] __page_cache_alloc+0xb/0x10
> [703826.074263]  [<ffffffff810ae2ff>] __do_page_cache_readahead+0xdf/0x220
> [703826.080865]  [<ffffffff810ae45c>] ra_submit+0x1c/0x20
> [703826.085998]  [<ffffffff810ae5f8>] ondemand_readahead+0xa8/0x1d0
> [703826.091994]  [<ffffffff810ae797>] page_cache_async_readahead+0x77/0xc0
> [703826.098595]  [<ffffffff810a6489>] generic_file_aio_read+0x259/0x6d0
> [703826.104941]  [<ffffffff810eac21>] do_sync_read+0xd1/0x110
> [703826.110418]  [<ffffffff810eb3f6>] vfs_read+0xc6/0x170
> [703826.115547]  [<ffffffff810eb860>] sys_read+0x50/0x90
> [703826.120591]  [<ffffffff81002c2b>] system_call_fastpath+0x16/0x1b
> 
> I have the feeling that the compcache device all of a sudden lost its
> efficiency.  It can't do much about having non-compressible data stuck
> in it, of course.
> 
> But, it used to be able to write things out to backing storage.  It
> tries to return I/O errors when it runs out of space, but my system
> didn't get that far.  It panic'd before it got the chance.
> 
> This seems like an issue that will probably crop up when we use zram as
> a swap device too.  A panic seems like pretty undesirable behavior when
> you've simply changed the kind of data being used.  Have you run into
> this at all?
> 


Ability to write out zram (compressed) memory to a backing disk seems
really useful. However considering lkml reviews, I had to drop this
feature. Anyways, I guess I will try to push this feature again.

Also, please do not use linux-next/mainline version of compcache. Instead
just use version in the project repository here:
hg clone https://compcache.googlecode.com/hg/ compcache 

This is updated much more frequently and has many more bug fixes over
the mainline. It will also be easier to fix bugs/add features much more
quickly in this repo rather than sending them to lkml which can take
long time.

Thanks,
Nitin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists