lkml.org 
[lkml]   [2013]   [Dec]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    Patch in this message
    /
    SubjectRe: process 'stuck' at exit.
    From
    Date
    On Tue, 2013-12-10 at 23:42 +0100, Thomas Gleixner wrote:
    > On Tue, 10 Dec 2013, Linus Torvalds wrote:
    >
    > > On Tue, Dec 10, 2013 at 1:57 PM, Linus Torvalds
    > > <torvalds@linux-foundation.org> wrote:
    > > >
    > > > So it looks like __get_user_pages_fast() fails, and keeps failing.
    > >
    > > Hmm.. Is any of the addresses unchecked, perhaps?
    > > __get_user_pages_fast() does an access_ok() check, while
    > > get_user_pages_fast() does *not* seem to do one.
    > >
    > > That looks a bit dangerous. Yeah, users should have checked the
    > > address range, but there really is no reason not to do it in
    > > get_user_pages_fast().
    > >
    > > And it looks like the futex code is actually seriously buggered. It
    > > only does the access_ok() check for the non-shared case.
    > >
    > > Why?
    >
    > The !fshared case is the fast path which does not even reach
    > get_user_pages_fast().
    >
    > We had this discussion some time ago already, where the access_ok()
    > check was missing in the !fshared case or the check was buggered for
    > some reason. Need to dig up the gory details.
    >
    > And yes, I remember that we do not do an extra check for the fshared
    > case, because get_user_pages_fast() should do it for us already. If
    > not we are fubared not only in the futex code.
    >
    > But there is a subtle detail:
    >
    > err = get_user_pages_fast(address, 1, 1, &page);
    >
    > So we ask for write access as the write argument is 1. In case that
    > fails we have that fallback path:
    >
    > /*
    > * If write access is not required (eg. FUTEX_WAIT), try
    > * and get read-only access.
    > */
    > if (err == -EFAULT && rw == VERIFY_READ) {
    > err = get_user_pages_fast(address, 1, 0, &page);
    >
    > That's a legitimate use case. And futex_requeue only requests
    > VERIFY_READ for the !requeue_pi case.
    >
    > Now, if that map is RO, i.e. we took the fallback path then the THP
    > one will fail as it has write=1 unconditionally.
    >
    > if (likely(__get_user_pages_fast(address, 1, 1, &page) == 1))
    >

    Is there a reason THP requires unconditional rw? Andrea?

    Or is the following actually the answer here?

    diff --git a/kernel/futex.c b/kernel/futex.c
    index 80ba086..02febad 100644
    --- a/kernel/futex.c
    +++ b/kernel/futex.c
    @@ -288,7 +288,7 @@ again:
    put_page(page);
    /* serialize against __split_huge_page_splitting() */
    local_irq_disable();
    - if (likely(__get_user_pages_fast(address, 1, 1, &page) == 1)) {
    + if (likely(__get_user_pages_fast(address, 1, !ro, &page) == 1)) {
    page_head = compound_head(page);
    /*
    * page_head is valid pointer but we must pin



    --
    Darren Hart
    Intel Open Source Technology Center
    Yocto Project - Linux Kernel




    \
     
     \ /
      Last update: 2013-12-11 01:01    [W:2.088 / U:0.024 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site