linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Rusty Russell <rusty@rustcorp.com.au>
To: Johannes Stezenbach <js@linuxtv.org>
Cc: lkml - Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Gerd Knorr <kraxel@bytesex.org>
Subject: Re: modprobe + request_module() deadlock
Date: Thu, 18 Nov 2004 14:48:22 +1100	[thread overview]
Message-ID: <1100749702.5865.39.camel@localhost.localdomain> (raw)
In-Reply-To: <20041117222949.GA9006@linuxtv.org>

On Wed, 2004-11-17 at 23:29 +0100, Johannes Stezenbach wrote:
> Hi,
> 
> it seems that modprobe in newer versions of module-init-tools
> (here: 3.1-pre6) gets an exclusive lock on the module's .ko file:
> 
>                 struct flock lock;
>                 lock.l_type = F_WRLCK;
>                 lock.l_whence = SEEK_SET;
>                 lock.l_start = 0;
>                 lock.l_len = 1;
>                 fcntl(fd, F_SETLKW, &lock);
> 
> This leads to a deadlock when the loaded module calls
> request_module() in its module_init() function, to load
> a module which in turn depends on the first module.

My bug, I think.  Does this help?

Rusty.

--- modprobe.c.~70~	2004-09-30 20:16:19.000000000 +1000
+++ modprobe.c	2004-11-18 14:44:57.000000000 +1100
@@ -735,6 +735,11 @@
 		       strip_vermagic, strip_modversion);
 	}
 
+	/* Don't do ANYTHING if already in kernel. */
+	if (!ignore_proc
+	    && module_in_kernel(newname ?: mod->modname, NULL) == 1)
+		goto exists_error;
+
 	fd = lock_file(mod->filename);
 	if (fd < 0) {
 		error("Could not open '%s': %s\n",
@@ -742,11 +747,6 @@
 		goto out_optstring;
 	}
 
-	/* Don't do ANYTHING if already in kernel. */
-	if (!ignore_proc
-	    && module_in_kernel(newname ?: mod->modname, NULL) == 1)
-		goto exists_error;
-
 	command = find_command(mod->modname, commands);
 	if (command && !ignore_commands) {
 		/* It might recurse: unlock. */
@@ -799,7 +799,7 @@
 	if (first_time)
 		error("Module %s already in kernel.\n",
 		      newname ?: mod->modname);
-	goto out_unlock;
+	goto out_optstring;
 }
 
 /* Do recursive removal. */

-- 
A bad analogy is like a leaky screwdriver -- Richard Braakman


  reply	other threads:[~2004-11-18  3:48 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-11-17 22:29 modprobe + request_module() deadlock Johannes Stezenbach
2004-11-18  3:48 ` Rusty Russell [this message]
2004-11-18 13:55   ` Johannes Stezenbach
2004-11-18 19:05     ` Takashi Iwai
2004-11-19  4:04       ` Alexander E. Patrakov
2004-11-19 11:10         ` Takashi Iwai
2004-11-19 11:50           ` Gerd Knorr
2004-11-19 12:42             ` Alexander E. Patrakov
2004-11-21  8:39             ` Rusty Russell
2004-11-22 10:25               ` Gerd Knorr
2004-11-22 14:16                 ` Johannes Stezenbach
2004-11-22 14:44                   ` Gerd Knorr
2004-11-22 15:36                     ` Johannes Stezenbach
2004-11-22 16:52                       ` Gerd Knorr
2004-11-24  5:02                         ` Rusty Russell
2004-11-24 12:11                           ` Gerd Knorr
2004-11-25 16:03                           ` Gerd Knorr
2004-11-26  0:34                             ` Rusty Russell

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=1100749702.5865.39.camel@localhost.localdomain \
    --to=rusty@rustcorp.com.au \
    --cc=js@linuxtv.org \
    --cc=kraxel@bytesex.org \
    --cc=linux-kernel@vger.kernel.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;
as well as URLs for NNTP newsgroup(s).