[Fedora 02/19] mm: vm_brk(), align the length to page boundary

[Vivek Goyal]

On Thu, Sep 05, 2013 at 08:26:26AM -0400, Josh Boyer wrote:
> On Wed, Sep 4, 2013 at 5:24 PM, Vivek Goyal <vgoyal at redhat.com> wrote:
> > I was writing some code where I was locking all pages of a process
> > during exec() time by setting VM_LOCKED flag in mm->def_flags. But
> > that lead to errors because length of mapping is not page aligned.
> >
> > login: [  174.669002] INFO: rcu_sched self-detected stall on CPU { 2}  (t=60000
> > jiffies g=2580 c=2579 q=1085)
> > [  174.669002] Pid: 4894, comm: kexec Not tainted 3.9.0-rc6+ #243
> > [  174.669002] Call Trace:
> > [  174.669002]  <IRQ>  [<ffffffff810c413a>] rcu_check_callbacks+0x21a/0x760
> > [  174.669002]  [<ffffffff810c7c0c>] ? acct_account_cputime+0x1c/0x20
> > [  174.669002]  [<ffffffff8104fd08>] update_process_times+0x48/0x80
> > [  174.669002]  [<ffffffff810913dd>] tick_sched_handle+0x3d/0x50
> > [  174.669002]  [<ffffffff810915e5>] tick_sched_timer+0x45/0x70
> > [  174.669002]  [<ffffffff81066951>] __run_hrtimer+0x81/0x220
> > [  174.669002]  [<ffffffff810915a0>] ? tick_nohz_handler+0xa0/0xa0
> > [  174.669002]  [<ffffffff8108ae0c>] ? ktime_get_update_offsets+0x4c/0xd0
> > [  174.669002]  [<ffffffff81067297>] hrtimer_interrupt+0xf7/0x250
> > [  174.669002]  [<ffffffff81886739>] smp_apic_timer_interrupt+0x69/0x99
> > [  174.669002]  [<ffffffff818859ca>] apic_timer_interrupt+0x6a/0x70
> > [  174.669002]  <EOI>  [<ffffffff8111e557>] ?  __mlock_vma_pages_range+0x57/0x70
> > [  174.669002]  [<ffffffff8111e568>] ? __mlock_vma_pages_range+0x68/0x70
> > [  174.669002]  [<ffffffff8111ea01>] __mm_populate+0x71/0x140
> > [  174.669002]  [<ffffffff81121b5f>] vm_brk+0x7f/0xa0
> > [  174.669002]  [<ffffffff81199633>] load_elf_binary+0x1a73/0x1b10
> > [  174.669002]  [<ffffffff812d25a5>] ? ima_bprm_check+0x55/0x70
> > [  174.669002]  [<ffffffff8114890a>] search_binary_handler+0x12a/0x3b0
> > [  174.669002]  [<ffffffff81197bc0>] ? load_elf_library+0x210/0x210
> > [  174.669002]  [<ffffffff8114aa00>] do_execve_common+0x500/0x5c0
> > [  174.669002]  [<ffffffff8114aaf7>] do_execve+0x37/0x40
> > [  174.669002]  [<ffffffff8114ad9d>] sys_execve+0x3d/0x60
> > [  174.669002]  [<ffffffff81885379>] stub_execve+0x69/0xa0
> >
> > Thanks to Michel and Hugh Dickens that they identified that __mm_populate()
> > will loop forever if passed in length is not page aligned. Similar
> > issues related to mmap() have already been fixed. This patch fixes
> > vm_brk().
> >
> > sys_brk() seems to be only other caller of do_brk() and sys_brk()
> > already aligns lenth to page boundary. So looks like page alignment
> > logic can be removed from do_brk().
> >
> > Signed-off-by: Michel Lespinasse <walken at google.com>
> > Signed-off-by: Vivek Goyal <vgoyal at redhat.com>
> 
> I'm surprised this hasn't been submitted upstream yet.  It seems
> orthogonal to the rest of them, and like something we'd want fixed
> upstream ASAP.
> 
> Did Michel or Hugh think there was an issue with doing the changes
> this way?  I can't seem to find anything on the lists, but I'm curious
> why they didn't submit the fix themselves.

Michel was supposed to send it upstream. I had pinged him twice and
nothing happened. 

There was no urgency on this as none of the other path hit this condition.

Thanks
Vivek