linux-kernel - Re: Linux 2.6.31 - very swap-happy with plenty of free RAM

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <c0c067900912022016u32e393cw3fe9e04ba1051253@mail.gmail.com>
Date:	Wed, 2 Dec 2009 23:16:02 -0500
From:	Dan Merillat <dan.merillat@...il.com>
To:	Rik van Riel <riel@...hat.com>
Cc:	KOSAKI Motohiro <kosaki.motohiro@...il.com>,
	linux-kernel@...r.kernel.org, Norbert Preining <preining@...ic.at>,
	Tomasz Chmielewski <mangoo@...g.org>,
	Sven-Haegar Koch <haegar@...net.de>,
	Dave Chinner <david@...morbit.com>
Subject: Re: Linux 2.6.31 - very swap-happy with plenty of free RAM

>
> Can you try out the patch from
> http://lkml.org/lkml/2009/11/25/467 ?

It's on the system now, I missed your reply for some reason.   Stupid gmail.

So far it's survived the 'open all in tabs' firefox test, which used
to throw it into complete disarray, threw a 1gb file into tmpfs and it
swapped out 150mb and didn't trash.

It's running all the things I normally do in a day all at once, so far
nothing, but it seems to take
a while before things go pear-shaped.

I'll give another update in a day or so.

As an aside, why is swapin so incredibly slow compared to swapout?   I
mean, I understand why (pagefault->swapin *wait* run process new
pagefault->swapin etc) but wouldn't much larger (4mb?) granularity of
swap help with modern high-speed drives?

I've seen a high of 2mb/sec swapin, and 100+mb/sec swapout.

swapoff /dev/swap shows this very clearly - theoretically it should be
"drop enough clean pages to fit everything into RAM, then read the
swap partition in linearly".  It's especially egregious in the case
where a memory-hog has terminated, leaving more free RAM than swapped
pages and swapoff still takes upwards of 10 minutes to complete what
should be a 2-second disk read.
--
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/