lkml.org 
[lkml]   [2002]   [Aug]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH] Linux-2.5 fix/improve get_pid()

    On 9 Aug 2002, Paul Larson wrote:
    > On Fri, 2002-08-09 at 16:42, Linus Torvalds wrote:
    > > Hmm.. Giving them a quick glance-over, the /proc issues look like they
    > > shouldn't be there in 2.5.x anyway, since the inode number really is
    > > largely just a random number in 2.5 and all the real information is
    > > squirelled away at path open time.

    It looks like the biggest impact on /proc would be that the /proc/<pid>
    inodes wouldn't be 10%% unique any more, which in turn means that an
    old-style /bin/pwd that actually walks the tree backwards and checks the
    inode number would occasionally fail.

    That in turn makes me suspect that we'd better off just biting the bullet
    and makign the inode numbers almost completely static, and forcing that
    particular failure mode early rather than hit it randomly due to unlucky
    timing.

    Doing a simple strace shows that all the systems I have regular access to
    use the "getcwd()" system call anyway, which gets this right on /proc (and
    other filesystems that do not guarantee unique inode numbers)

    > So for now then, should I dig out my original (minimal) patch that
    > *just* fixed the "loop forever even if we're out of pids" problem? Even
    > if we increase PID_MAX to something obscenely high, I think we should
    > still be checking for this.

    Ehh, considering that especially with a 30-bit pid, there's _no_ way we'd
    run out without some other serious problems hitting us long before (out of
    memory being the obvious one), I don't think even that is an issue.

    With a minimum of 8kB / pid for process overhead, you need to have at
    least 43 bits of physical address space completely populated just to cover
    the memory uses of that many pid's.

    Linus

    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

    \
     
     \ /
      Last update: 2005-03-22 13:27    [W:3.036 / U:0.532 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site