mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Andrew Morton <akpm@osdl.org>
To: Steven Rostedt <rostedt@goodmis.org>
Cc: linux-kernel@vger.kernel.org, mingo@elte.hu
Subject: Re: [PATCH] protect remove_proc_entry
Date: Sat, 7 Jan 2006 03:36:37 -0800	[thread overview]
Message-ID: <20060107033637.458c4716.akpm@osdl.org> (raw)
In-Reply-To: <1135978110.6039.81.camel@localhost.localdomain>

Steven Rostedt <rostedt@goodmis.org> wrote:
>
> Working on a custom kernel that adds and removes proc entries quite a
> bit, I discovered that remove_proc_entry is not protected against
> multiple threads removing entries belonging to the same parent.  At
> first I thought that this is only a problem with my changes, but after
> inspecting the vanilla kernel, I see that there's several places that
> calls remove_proc_entry with the same parent (most noticeably
> /proc/drivers).
> 
> I've added a global remove_proc_lock to protect this section of code.  I
> was going to add a lock to proc_dir_entry so that the locking is only
> cut down to the same parent, but since this function is called so
> infrequently, why waste more memory then is needed.  One global lock
> should not cause too much of a headache here.
> 
> I'm not sure if remove_proc_entry is called from interrupt context, so I
> did a irqsave just in case.
> 
> -- Steve
> 
> 
> Index: linux-2.6.15-rc7/fs/proc/generic.c
> ===================================================================
> --- linux-2.6.15-rc7.orig/fs/proc/generic.c	2005-12-30 14:19:39.000000000 -0500
> +++ linux-2.6.15-rc7/fs/proc/generic.c	2005-12-30 16:18:42.000000000 -0500
> @@ -19,6 +19,7 @@
>  #include <linux/idr.h>
>  #include <linux/namei.h>
>  #include <linux/bitops.h>
> +#include <linux/spinlock.h>
>  #include <asm/uaccess.h>
>  
>  static ssize_t proc_file_read(struct file *file, char __user *buf,
> @@ -27,6 +28,8 @@
>  			       size_t count, loff_t *ppos);
>  static loff_t proc_file_lseek(struct file *, loff_t, int);
>  
> +static DEFINE_SPINLOCK(remove_proc_lock);
> +
>  int proc_match(int len, const char *name, struct proc_dir_entry *de)
>  {
>  	if (de->namelen != len)
> @@ -689,10 +692,13 @@
>  	struct proc_dir_entry *de;
>  	const char *fn = name;
>  	int len;
> +	unsigned long flags;
>  
>  	if (!parent && xlate_proc_name(name, &parent, &fn) != 0)
>  		goto out;
>  	len = strlen(fn);
> +
> +	spin_lock_irqsave(&remove_proc_lock, flags);
>  	for (p = &parent->subdir; *p; p=&(*p)->next ) {
>  		if (!proc_match(len, fn, *p))
>  			continue;
> @@ -713,6 +719,7 @@
>  		}
>  		break;
>  	}
> +	spin_unlock_irqrestore(&remove_proc_lock, flags);
>  out:
>  	return;
>  }

Aren't there other places where we need to take this lock?  Code which
traverses that list, code which adds things to it?



  parent reply	other threads:[~2006-01-07 11:37 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-12-30 20:04 [Question] race condition with remove_proc_entry Steven Rostedt
2005-12-30 21:28 ` [PATCH] protect remove_proc_entry Steven Rostedt
2005-12-30 21:34   ` Daniel Walker
2005-12-30 21:55     ` Steven Rostedt
2005-12-30 21:55   ` Mitchell Blank Jr
2005-12-30 22:09     ` Steven Rostedt
2005-12-30 22:18       ` Steven Rostedt
2006-01-04  9:21         ` Andrew Morton
2006-01-04 12:18           ` Steven Rostedt
2006-01-05  1:48             ` Mitchell Blank Jr
2006-01-07 11:25       ` Andrew Morton
2005-12-30 22:11     ` Steven Rostedt
2005-12-30 23:46   ` Andrew Morton
2005-12-31  6:58     ` Steven Rostedt
2005-12-31  8:34       ` Arjan van de Ven
2005-12-31  8:53     ` Kirill Korotaev
2006-01-04  9:36       ` Andrew Morton
2006-01-04 11:27         ` Kirill Korotaev
2006-01-02 13:02     ` Steven Rostedt
2006-01-07 11:36   ` Andrew Morton [this message]
2006-01-07 12:04     ` Steven Rostedt
2006-01-09 19:16     ` Steven Rostedt
2006-01-10  0:59       ` Steven Rostedt
2006-01-10  1:05         ` Ingo Molnar
2006-01-10 13:26       ` Steven Rostedt

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=20060107033637.458c4716.akpm@osdl.org \
    --to=akpm@osdl.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=rostedt@goodmis.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

Powered by JetHome