Re: [PATCH] filp->f_pos not correctly updated in proc_task_readdir

[Al Viro <viro@ZenIV.linux.org.uk>]

On Mon, Mar 16, 2009 at 02:44:31PM +0800, Zhang Le wrote:
> filp->f_pos only get updated at the end of the function. Thus d_off of those
> dirents who are in the middle will be 0, and this will cause a problem in
> glibc's readdir implementation, specifically endless loop. Because when overflow
> occurs, f_pos will be set to next dirent to read, however it will be 0, unless
> the next one is the last one. So it will start over again and again.

Eh...  Here's what's really going on:

proc_..._fill_cache() API is rather kludgy and far too convoluted.  In
particular, it calls filldir() and passes file->f_pos to it, expecting
the caller to update file->f_pos between the calls.  proc_task_readdir()
uses that sucker, but doesn't update ->f_pos until the very end.  As
the result, d_off of direntries produced by it gets screwed.
Broken-by: commit 61a28784028e6d55755e4d0f39bee8d9bf2ee8d9
Author: Eric W. Biederman <ebiederm@xmission.com>
Date:   Mon Oct 2 02:18:49 2006 -0700

    [PATCH] proc: Remove the hard coded inode numbers

Patch is fine, but I'd rather rip the layers of ..._fill_cache() abstractions
out and see what falls out.  Anyway, for -rc your variant is definitely the
way to go.