public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Rusty Russell <rusty@rustcorp.com.au>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Brandon Philips <brandon@ifup.org>,
	"Rafael J. Wysocki" <rjw@sisk.pl>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Jon Masters <jonathan@jonmasters.org>,
	Tejun Heo <htejun@gmail.com>,
	Masami Hiramatsu <mhiramat@redhat.com>,
	Kay Sievers <kay.sievers@vrfy.org>
Subject: Re: [PATCH 2/2] module: fix bne2 "gave up waiting for init of module libcrc32c"
Date: Tue, 1 Jun 2010 10:34:46 +0930	[thread overview]
Message-ID: <201006011034.47739.rusty@rustcorp.com.au> (raw)
In-Reply-To: <20100531094834.c1a684d1.akpm@linux-foundation.org>

On Tue, 1 Jun 2010 02:18:34 am Andrew Morton wrote:
> On Mon, 31 May 2010 21:32:27 +0930 Rusty Russell <rusty@rustcorp.com.au> wrote:
> 
> > Problem: it's hard to avoid an init routine stumbling over a
> > request_module these days.  And it's not clear it's always a bad idea:
> > for example, a module like kvm with dynamic dependencies on kvm-intel
> > or kvm-amd would be neater if it could simply request_module the right
> > one.
> > 
> > In this particular case, it's libcrc32c:
> > 
> > 	libcrc32c_mod_init
> > 	 crypto_alloc_shash
> > 	  crypto_alloc_tfm
> > 	   crypto_find_alg
> > 	    crypto_alg_mod_lookup
> > 	     crypto_larval_lookup
> > 	      request_module
> > 
> > If another module is waiting for libcrc32c to finish initializing
> > (ie. bne2 depends on libcrc32c) then it does so holding the module
> > lock, and our request_module() can't make progress until that is
> > released.
> > 
> > Waiting without the lock isn't all that hard: we just need to pass the
> > -EBUSY up the call chain so we can sleep where we don't hold the lock.
> > Error reporting is a bit trickier: we need to copy the name of the
> > unfinished module before releasing the lock.
> 
> Who's returning -EBUSY?  request_module()?  If so, are you requiring
> that all code which might call request_module() be correctly
> propagating error codes back?  Please spell this all out?

Sorry if I was unclear.  It's all inside module.c.  Here's the updated
commentry:

 If another module is waiting inside resolve_symbol() for libcrc32c to
 finish initializing (ie. bne2 depends on libcrc32c) then it does so
 holding the module lock, and our request_module() can't make progress
 until that is released.

 Waiting inside resolve_symbol() without the lock isn't all that hard:
 we just need to pass the -EBUSY up the call chain so we can sleep
 where we don't hold the lock.  Error reporting is a bit trickier: we
 need to copy the name of the unfinished module before releasing the
 lock.

Hope that helps,
Rusty.

  parent reply	other threads:[~2010-06-01  1:04 UTC|newest]

Thread overview: 71+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-05-25 21:00 [Regression] Crash in load_module() while freeing args Rafael J. Wysocki
2010-05-25 22:54 ` Rafael J. Wysocki
2010-05-25 23:47   ` Linus Torvalds
2010-05-26  8:00     ` Rusty Russell
2010-05-26 11:57       ` Rusty Russell
2010-05-26 22:56         ` Rafael J. Wysocki
2010-05-26 23:07           ` Linus Torvalds
2010-05-27  5:26           ` Rusty Russell
2010-05-27 18:46             ` Brandon Philips
2010-05-31  9:40               ` Rusty Russell
2010-05-31 12:00                 ` [PATCH 0/2] kernel/module.c locking changes Rusty Russell
2010-05-31 12:01                   ` [PATCH 1/2] module: make locking more fine-grained Rusty Russell
2010-05-31 12:02                     ` [PATCH 2/2] module: fix bne2 "gave up waiting for init of module libcrc32c" Rusty Russell
2010-05-31 16:48                       ` Andrew Morton
2010-05-31 18:19                         ` Linus Torvalds
2010-05-31 20:15                           ` Linus Torvalds
2010-05-31 20:16                             ` [PATCH 1/2] Make the module 'usage' lists be two-way Linus Torvalds
2010-05-31 20:17                               ` [PATCH 2/2] module: wait for other modules after dropping the module_mutex Linus Torvalds
2010-06-01  1:37                               ` [PATCH 1/2] Make the module 'usage' lists be two-way Rusty Russell
2010-06-01  3:42                                 ` Rusty Russell
2010-06-01  4:00                                   ` Linus Torvalds
2010-06-01  4:05                                     ` Linus Torvalds
2010-06-01  2:44                               ` Américo Wang
2010-06-01  3:51                                 ` Linus Torvalds
2010-06-01  1:57                             ` [PATCH 2/2] module: fix bne2 "gave up waiting for init of module libcrc32c" Rusty Russell
2010-06-01  3:40                               ` Linus Torvalds
2010-06-01  4:27                                 ` Linus Torvalds
2010-06-01  5:19                                 ` Rusty Russell
2010-06-02  3:15                                   ` Rusty Russell
2010-06-01  1:21                           ` Rusty Russell
2010-06-01  3:24                             ` Linus Torvalds
2010-06-01  5:22                               ` Rusty Russell
2010-06-01 14:58                                 ` Linus Torvalds
2010-06-01 17:53                                   ` Linus Torvalds
2010-06-01 23:24                                     ` Brandon Philips
2010-06-01 23:51                                       ` Linus Torvalds
2010-06-02  2:10                                         ` Brandon Philips
2010-06-02  3:03                                           ` Rusty Russell
2010-06-02  4:35                                           ` Linus Torvalds
2010-06-02  4:44                                             ` Linus Torvalds
2010-06-02  6:35                                               ` Rusty Russell
2010-06-02  7:45                                                 ` Linus Torvalds
2010-06-02  8:12                                                   ` Linus Torvalds
2010-06-02  9:07                                                     ` Rusty Russell
2010-06-02  5:52                                             ` Rusty Russell
2010-06-02  7:21                                               ` Linus Torvalds
2010-06-02 14:06                                                 ` Rusty Russell
2010-06-02 14:50                                                   ` Linus Torvalds
2010-06-03 13:06                                                     ` Rusty Russell
2010-06-02 16:53                                                   ` Brandon Philips
2010-06-02 18:01                                                   ` Linus Torvalds
2010-06-03  5:20                                                     ` Rusty Russell
2010-06-03 16:24                                                       ` Linus Torvalds
2010-06-04  1:02                                                         ` Rusty Russell
2010-06-04  1:55                                                           ` Linus Torvalds
2010-06-04  5:20                                                             ` Rusty Russell
2010-06-04 22:48                                                               ` Linus Torvalds
2010-06-05  1:49                                                                 ` Rusty Russell
2010-06-02  3:09                                   ` Rusty Russell
2010-06-02  4:32                                     ` Linus Torvalds
2010-06-02  4:56                                     ` Linus Torvalds
2010-06-02  5:52                                       ` Rusty Russell
2010-06-02  6:59                                         ` Linus Torvalds
2010-06-01  1:04                         ` Rusty Russell [this message]
2010-06-01  5:38                     ` [PATCH 1/2] module: make locking more fine-grained Américo Wang
2010-06-01  5:55                       ` Rusty Russell
2010-05-27 21:57             ` [Regression] Crash in load_module() while freeing args Rafael J. Wysocki
2010-05-31  7:54               ` Rusty Russell
2010-05-31 10:23               ` [PATCH] module: fix reference to mod->percpu after freeing module Rusty Russell
2010-05-31 10:25                 ` Tejun Heo
2010-05-26 15:41       ` [Regression] Crash in load_module() while freeing args Linus Torvalds

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=201006011034.47739.rusty@rustcorp.com.au \
    --to=rusty@rustcorp.com.au \
    --cc=akpm@linux-foundation.org \
    --cc=brandon@ifup.org \
    --cc=htejun@gmail.com \
    --cc=jonathan@jonmasters.org \
    --cc=kay.sievers@vrfy.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mhiramat@redhat.com \
    --cc=rjw@sisk.pl \
    --cc=torvalds@linux-foundation.org \
    /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