linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Andi Kleen <andi@firstfloor.org>
To: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>,
	Wu Fengguang <fengguang.wu@intel.com>,
	Balbir Singh <balbir@linux.vnet.ibm.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	LKML <linux-kernel@vger.kernel.org>, Ingo Molnar <mingo@elte.hu>,
	Mel Gorman <mel@csn.ul.ie>, Thomas Gleixner <tglx@linutronix.de>,
	"H. Peter Anvin" <hpa@zytor.com>,
	Peter Zijlstra <a.p.zijlstra@chello.nl>,
	Nick Piggin <npiggin@suse.de>, Andi Kleen <andi@firstfloor.org>,
	"riel@redhat.com" <riel@redhat.com>,
	"chris.mason@oracle.com" <chris.mason@oracle.com>,
	"linux-mm@kvack.org" <linux-mm@kvack.org>
Subject: Re: [PATCH 00/22] HWPOISON: Intro (v5)
Date: Mon, 15 Jun 2009 15:29:34 +0200	[thread overview]
Message-ID: <20090615132934.GE31969@one.firstfloor.org> (raw)
In-Reply-To: <20090615140019.4e405d37@lxorguk.ukuu.org.uk>


I think you're wrong about killing processes decreasing
reliability. Traditionally we always tried to keep things running if possible
instead of panicing. That is why ext3 or block does not default to panic
on each IO error for example. Or oops does not panic by default like
on BSDs. Your argumentation would be good for a traditional early Unix
which likes to panic instead of handling errors, but that's not the
Linux way as I know it.

Also BTW in many cases (e.g. a lot of file cache pages) there 
is actually no process kill involved; just a reload of the page from
disk.

Then for example in a cluster you typically have a application level
heartbeat on the application and restarting the app is faster
if you don't need to reboot the box too. If you don't have a cluster
with failover then gracefull degradation is the best. In general
panic is a very nasty failure mode and should be avoided.

That said you can configure it anyways to panic if you want,
but it would be a very bad default.

See also Linus' or hpa's statement on the topic.

> no testing, 

There's an extensive test suite in mce-test.git 

We did a lot of testing with these separate test suites and also
some other tests. For much more it needs actual users pounding on it, and that 
can be only adequately done in mainline.

That said the real tests of this can be only done with test suites
really, these errors tend to not happen quickly. 

> integration shakedown, no builds on non-x86 boxes, no work with other
> arch maintainers who have similar abilities and needs.

We did build tests on ia64 and power and it was reviewed by Tony for IA64.
The ia64 specific code is not quite ready yet, but will come at some point.

I don't think it's a requirement for merging to have PPC64 support.

-Andi
-- 
ak@linux.intel.com -- Speaking for myself only.

--
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>

  reply	other threads:[~2009-06-15 13:20 UTC|newest]

Thread overview: 79+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-06-15  2:45 Wu Fengguang
2009-06-15  2:45 ` [PATCH 01/22] HWPOISON: Add page flag for poisoned pages Wu Fengguang
2009-06-15  2:45 ` [PATCH 02/22] HWPOISON: Export some rmap vma locking to outside world Wu Fengguang
2009-06-15  2:45 ` [PATCH 03/22] HWPOISON: Add support for poison swap entries v2 Wu Fengguang
2009-06-15  2:45 ` [PATCH 04/22] HWPOISON: Add new SIGBUS error codes for hardware poison signals Wu Fengguang
2009-06-15  2:45 ` [PATCH 05/22] HWPOISON: Add basic support for poisoned pages in fault handler v3 Wu Fengguang
2009-06-15  2:45 ` [PATCH 06/22] HWPOISON: x86: Add VM_FAULT_HWPOISON handling to x86 page fault handler v2 Wu Fengguang
2009-06-15  2:45 ` [PATCH 07/22] HWPOISON: define VM_FAULT_HWPOISON to 0 when feature is disabled Wu Fengguang
2009-06-15  2:45 ` [PATCH 08/22] HWPOISON: Use bitmask/action code for try_to_unmap behaviour Wu Fengguang
2009-06-15  2:45 ` [PATCH 09/22] HWPOISON: Handle hardware poisoned pages in try_to_unmap Wu Fengguang
2009-06-15 13:09   ` Minchan Kim
2009-06-15 15:26     ` Wu Fengguang
2009-06-16  0:03       ` Minchan Kim
2009-06-16 13:49         ` Wu Fengguang
2009-06-17  0:28           ` Minchan Kim
2009-06-17  7:23             ` Wu Fengguang
2009-06-17 13:27               ` Minchan Kim
2009-06-17 13:37                 ` Wu Fengguang
2009-06-17 13:43                   ` Minchan Kim
2009-06-17 14:03                     ` Wu Fengguang
2009-06-17 14:08                       ` Minchan Kim
2009-06-17 14:12                         ` Wu Fengguang
     [not found]               ` <28c262360906170644w65c08a8y2d2805fb08045804@mail.gmail.com>
     [not found]                 ` <20090617135543.GA8079@localhost>
     [not found]                   ` <28c262360906170703h3363b68dp74471358f647921e@mail.gmail.com>
2009-06-18 12:14                     ` Wu Fengguang
2009-06-18 13:31                       ` Minchan Kim
2009-06-19  1:58                         ` Wu Fengguang
2009-06-15  2:45 ` [PATCH 10/22] HWPOISON: check and isolate corrupted free pages v2 Wu Fengguang
2009-06-15  9:41   ` KAMEZAWA Hiroyuki
2009-06-15 10:16     ` Wu Fengguang
2009-06-15 23:52       ` KAMEZAWA Hiroyuki
2009-06-16  0:34         ` Wu Fengguang
2009-06-16 11:29           ` Hugh Dickins
2009-06-16 11:40             ` Wu Fengguang
2009-06-15  2:45 ` [PATCH 11/22] HWPOISON: Refactor truncate to allow direct truncating of page v3 Wu Fengguang
2009-06-15  2:45 ` [PATCH 12/22] HWPOISON: The high level memory error handler in the VM v7 Wu Fengguang
2009-06-15  2:45 ` [PATCH 13/22] HWPOISON: Add madvise() based injector for hardware poisoned pages v3 Wu Fengguang
2009-06-15  2:45 ` [PATCH 14/22] HWPOISON: Add simple debugfs interface to inject hwpoison on arbitary PFNs Wu Fengguang
2009-06-15  2:45 ` [PATCH 15/22] HWPOISON: early kill cleanups and fixes Wu Fengguang
2009-06-15  2:45 ` [PATCH 16/22] mm: move page flag numbers for user space to page-flags.h Wu Fengguang
2009-06-15  2:45 ` [PATCH 17/22] HWPOISON: introduce struct hwpoison_control Wu Fengguang
2009-06-15  2:45 ` [PATCH 18/22] HWPOISON: use compound head page Wu Fengguang
2009-06-15  2:45 ` [PATCH 19/22] HWPOISON: detect free buddy pages explicitly Wu Fengguang
2009-06-15  2:45 ` [PATCH 20/22] HWPOISON: collect infos that reflect the impact of the memory corruption Wu Fengguang
2009-06-15  2:45 ` [PATCH 21/22] HWPOISON: send uevent to report " Wu Fengguang
2009-06-15  6:29   ` Andi Kleen
2009-06-15  9:56     ` Wu Fengguang
2009-06-16  0:35   ` Greg KH
2009-06-15  2:45 ` [PATCH 22/22] HWPOISON: FOR TESTING: Enable memory failure code unconditionally Wu Fengguang
2009-06-15  3:18 ` [PATCH 00/22] HWPOISON: Intro (v5) Balbir Singh
2009-06-15  4:27   ` Wu Fengguang
2009-06-15  6:44     ` Nick Piggin
2009-06-15  7:09       ` Andi Kleen
2009-06-15  7:19         ` Nick Piggin
2009-06-15 12:10           ` Wu Fengguang
2009-06-15 12:25             ` Nick Piggin
2009-06-15 14:22               ` Wu Fengguang
2009-06-17  6:37                 ` [RFC][PATCH] HWPOISON: only early kill processes who installed SIGBUS handler Wu Fengguang
2009-06-17  8:04                   ` Nick Piggin
2009-06-17  9:55                     ` Wu Fengguang
2009-06-17 10:00                       ` Nick Piggin
2009-06-17 11:56                         ` Wu Fengguang
2009-06-18  9:56                     ` Wu Fengguang
2009-06-15  8:14       ` [PATCH 00/22] HWPOISON: Intro (v5) Nick Piggin
2009-06-15 10:09         ` Wu Fengguang
2009-06-15 10:36           ` Nick Piggin
2009-06-15 11:41             ` Wu Fengguang
2009-06-15 12:51     ` Hugh Dickins
2009-06-15 13:00       ` Alan Cox
2009-06-15 13:29         ` Andi Kleen [this message]
2009-06-15 13:28           ` H. Peter Anvin
2009-06-15 14:48           ` Alan Cox
2009-06-15 15:24             ` Andi Kleen
2009-06-15 15:28               ` Alan Cox
2009-06-15 16:19                 ` Andi Kleen
2009-06-15 16:28                   ` Alan Cox
2009-06-15 17:07                     ` Andi Kleen
2009-06-16 19:44           ` Russ Anderson
2009-06-16 20:28             ` H. Peter Anvin
2009-06-16 20:54               ` Russ Anderson
2009-06-16 20:58                 ` H. Peter Anvin

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=20090615132934.GE31969@one.firstfloor.org \
    --to=andi@firstfloor.org \
    --cc=a.p.zijlstra@chello.nl \
    --cc=akpm@linux-foundation.org \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=balbir@linux.vnet.ibm.com \
    --cc=chris.mason@oracle.com \
    --cc=fengguang.wu@intel.com \
    --cc=hpa@zytor.com \
    --cc=hugh.dickins@tiscali.co.uk \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mel@csn.ul.ie \
    --cc=mingo@elte.hu \
    --cc=npiggin@suse.de \
    --cc=riel@redhat.com \
    --cc=tglx@linutronix.de \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox