lkml.org 
[lkml]   [2006]   [Oct]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC] Remove temp_priority
Martin J. Bligh wrote:
>> Coming from another angle, I am thinking about doing away with direct
>> reclaim completely. That means we don't need any GFP_IO or GFP_FS, and
>> solves the problem of large numbers of processes stuck in reclaim and
>> skewing aging and depleting the memory reserve.
>
>
> Last time I proposed that, the objection was how to throttle the heavy
> dirtiers so they don't fill up RAM with dirty pages?

Now that we have the dirty mmap accounting, page dirtiers should be
throttled pretty well via page writeback throttling.

> Also, how do you do atomic allocations? Create a huge memory pool and
> pray really hard?

Well, yes. Atomic allocations as of *today* cannot do any reclaim, and
thus they rely on kswapd to free their memory, and we keep a (not huge)
memory pool for them. They also have to be able to handle failures, and
by and large they do OK.

>> But that's tricky because we don't have enough kswapds to get maximum
>> reclaim throughput on many configurations (only single core opterons
>> and UP systems, really).
>
>
> It's not a question of enough kswapds. It's that we can dirty pages
> faster than they can possibly be written to disk.
>
> dd if=/dev/zero of=/tmp/foo

You can't catch that at the allocation side anyway because clean pagecache
may already exist for /tmp/foo.

We've always done pretty well (in 2.6) with correctly throttling and
limiting write(2) writes into pagecache, haven't we?

--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com

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

\
 
 \ /
  Last update: 2006-10-18 16:59    [W:0.035 / U:0.656 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site