From: Anton Blanchard <anton@samba.org>
To: Andrew Morton <akpm@osdl.org>
Cc: Andi Kleen <ak@suse.de>,
raybry@sgi.com, lse-tech@lists.sourceforge.net,
linux-ia64@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [Lse-tech] Re: Hugetlbpages in very large memory machines.......
Date: Sun, 14 Mar 2004 04:06:34 +0000 [thread overview]
Message-ID: <20040314040634.GC19737@krispykreme> (raw)
In-Reply-To: <20040313184547.6e127b51.akpm@osdl.org>
> Demand-paging the hugepages is a decent feature to have, and ISTR resisting
> it before for this reason.
>
> Even though it's early in the 2.6 series I'd be a bit worried about
> breaking existing hugetlb users in this way. Yes, the pages are
> preallocated so it is unlikely that a working setup is suddenly going to
> break. Unless someone is using the return value from mmap to find out how
> many pages they can get.
Hmm what a coincidence, I was chasing a problem where large page
allocations would fail even though I clearly had enough large page memory
free.
It turns out we were tripping the overcommit logic in do_mmap. I had
30GB of large page and 2GB of small pages and of course cap_vm_enough_memory
was looking at the small page pool. Setting overcommit to 1 fixed it.
It seems we can solve both problems by having a separate hugetlb overcommit
policy. Make it strict and you wont have OOM problems on large pages
and I wont hit my 30GB / 2GB problem.
Anton
WARNING: multiple messages have this Message-ID (diff)
From: Anton Blanchard <anton@samba.org>
To: Andrew Morton <akpm@osdl.org>
Cc: Andi Kleen <ak@suse.de>,
raybry@sgi.com, lse-tech@lists.sourceforge.net,
linux-ia64@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [Lse-tech] Re: Hugetlbpages in very large memory machines.......
Date: Sun, 14 Mar 2004 15:06:34 +1100 [thread overview]
Message-ID: <20040314040634.GC19737@krispykreme> (raw)
In-Reply-To: <20040313184547.6e127b51.akpm@osdl.org>
> Demand-paging the hugepages is a decent feature to have, and ISTR resisting
> it before for this reason.
>
> Even though it's early in the 2.6 series I'd be a bit worried about
> breaking existing hugetlb users in this way. Yes, the pages are
> preallocated so it is unlikely that a working setup is suddenly going to
> break. Unless someone is using the return value from mmap to find out how
> many pages they can get.
Hmm what a coincidence, I was chasing a problem where large page
allocations would fail even though I clearly had enough large page memory
free.
It turns out we were tripping the overcommit logic in do_mmap. I had
30GB of large page and 2GB of small pages and of course cap_vm_enough_memory
was looking at the small page pool. Setting overcommit to 1 fixed it.
It seems we can solve both problems by having a separate hugetlb overcommit
policy. Make it strict and you wont have OOM problems on large pages
and I wont hit my 30GB / 2GB problem.
Anton
next prev parent reply other threads:[~2004-03-14 4:06 UTC|newest]
Thread overview: 64+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-03-13 3:44 Hugetlbpages in very large memory machines Ray Bryant
2004-03-13 3:44 ` Ray Bryant
2004-03-13 3:45 ` Ray Bryant
2004-03-13 3:45 ` Ray Bryant
2004-03-13 3:48 ` Andi Kleen
2004-03-13 3:48 ` Andi Kleen
2004-03-13 5:49 ` William Lee Irwin III
2004-03-13 5:49 ` William Lee Irwin III
2004-03-13 16:10 ` [Lse-tech] " Andi Kleen
2004-03-13 16:10 ` Andi Kleen
2004-03-14 0:05 ` William Lee Irwin III
2004-03-14 0:05 ` William Lee Irwin III
2004-03-14 5:22 ` Peter Chubb
2004-03-14 5:22 ` Peter Chubb
2004-03-15 23:31 ` Seth, Rohit
2004-03-15 23:31 ` Seth, Rohit
[not found] ` <844231526.20040313030948@adinet.com.uy>
[not found] ` <20040313061232.GB655@holomorphy.com>
2004-03-13 16:32 ` Re[2]: " Luis Mirabal
2004-03-14 2:45 ` Andrew Morton
2004-03-14 2:45 ` Andrew Morton
2004-03-14 4:06 ` Anton Blanchard [this message]
2004-03-14 4:06 ` [Lse-tech] " Anton Blanchard
2004-03-17 19:05 ` [Lse-tech] Re: Hugetlbpages in very large memory Andy Whitcroft
2004-03-17 19:05 ` [Lse-tech] Re: Hugetlbpages in very large memory machines Andy Whitcroft
2004-03-18 20:25 ` [Lse-tech] Re: Hugetlbpages in very large memory Andrew Morton
2004-03-18 20:25 ` [Lse-tech] Re: Hugetlbpages in very large memory machines Andrew Morton
2004-03-18 21:22 ` [Lse-tech] Re: Hugetlbpages in very large memory Stephen Smalley
2004-03-18 21:22 ` [Lse-tech] Re: Hugetlbpages in very large memory machines Stephen Smalley
2004-03-18 22:21 ` [Lse-tech] Re: Hugetlbpages in very large memory Andy Whitcroft
2004-03-18 22:21 ` [Lse-tech] Re: Hugetlbpages in very large memory machines Andy Whitcroft
2004-03-23 17:30 ` [Lse-tech] Re: Hugetlbpages in very large memory Andy Whitcroft
2004-03-23 17:30 ` [Lse-tech] Re: Hugetlbpages in very large memory machines Andy Whitcroft
2004-03-24 17:38 ` [Lse-tech] Re: Hugetlbpages in very large memory Andy Whitcroft
2004-03-24 17:38 ` [Lse-tech] Re: Hugetlbpages in very large memory machines Andy Whitcroft
2004-03-14 8:38 ` Ray Bryant
2004-03-14 8:38 ` Ray Bryant
2004-03-14 8:48 ` William Lee Irwin III
2004-03-14 8:48 ` William Lee Irwin III
2004-03-14 8:57 ` [Lse-tech] Re: Hugetlbpages in very large memory Andrew Morton
2004-03-14 8:57 ` [Lse-tech] Re: Hugetlbpages in very large memory machines Andrew Morton
2004-03-14 9:02 ` [Lse-tech] Re: Hugetlbpages in very large memory Andrew Morton
2004-03-14 9:02 ` [Lse-tech] Re: Hugetlbpages in very large memory machines Andrew Morton
2004-03-14 9:07 ` William Lee Irwin III
2004-03-14 9:07 ` William Lee Irwin III
2004-03-15 6:45 ` Ray Bryant
2004-03-15 6:45 ` Ray Bryant
2004-03-15 23:54 ` William Lee Irwin III
2004-03-15 23:54 ` William Lee Irwin III
2004-03-13 3:55 ` William Lee Irwin III
2004-03-13 3:55 ` William Lee Irwin III
2004-03-13 4:56 ` Hirokazu Takahashi
2004-03-13 4:56 ` Hirokazu Takahashi
2004-03-16 0:30 ` Nobuhiko Yoshida
2004-03-16 0:30 ` Nobuhiko Yoshida
2004-03-16 1:54 ` Andi Kleen
2004-03-16 1:54 ` Andi Kleen
2004-03-16 2:32 ` Hirokazu Takahashi
2004-03-16 2:32 ` Hirokazu Takahashi
2004-03-16 3:20 ` Hirokazu Takahashi
2004-03-16 3:20 ` Hirokazu Takahashi
2004-03-16 3:15 ` Nobuhiko Yoshida
2004-03-16 3:15 ` Nobuhiko Yoshida
2004-04-01 9:10 ` Nobuhiko Yoshida
2004-04-01 9:10 ` Nobuhiko Yoshida
2004-03-15 15:28 ` jlnance
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20040314040634.GC19737@krispykreme \
--to=anton@samba.org \
--cc=ak@suse.de \
--cc=akpm@osdl.org \
--cc=linux-ia64@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=lse-tech@lists.sourceforge.net \
--cc=raybry@sgi.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.