lkml.org 
[lkml]   [2014]   [Jan]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH] mm/swap: fix race on swap_info reuse between swapoff and swapon
From
On Mon, Jan 13, 2014 at 11:27 AM, Andrew Morton
<akpm@linux-foundation.org> wrote:
> On Mon, 13 Jan 2014 11:08:58 +0800 Weijie Yang <weijie.yang.kh@gmail.com> wrote:
>
>> >> --- a/mm/swapfile.c
>> >> +++ b/mm/swapfile.c
>> >> @@ -1922,7 +1922,6 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
>> >> p->swap_map = NULL;
>> >> cluster_info = p->cluster_info;
>> >> p->cluster_info = NULL;
>> >> - p->flags = 0;
>> >> frontswap_map = frontswap_map_get(p);
>> >> spin_unlock(&p->lock);
>> >> spin_unlock(&swap_lock);
>> >> @@ -1948,6 +1947,16 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
>> >> mutex_unlock(&inode->i_mutex);
>> >> }
>> >> filp_close(swap_file, NULL);
>> >> +
>> >> + /*
>> >> + * clear SWP_USED flag after all resources freed
>> >> + * so that swapon can reuse this swap_info in alloc_swap_info() safely
>> >> + * it is ok to not hold p->lock after we cleared its SWP_WRITEOK
>> >> + */
>> >> + spin_lock(&swap_lock);
>> >> + p->flags = 0;
>> >> + spin_unlock(&swap_lock);
>> >> +
>> >> err = 0;
>> >> atomic_inc(&proc_poll_event);
>> >> wake_up_interruptible(&proc_poll_wait);
>> >
>> > I didn't look too closely, but this patch might also address the race
>> > which Krzysztof addressed with
>> > http://ozlabs.org/~akpm/mmots/broken-out/swap-fix-setting-page_size-blocksize-during-swapoff-swapon-race.patch.
>> > Can we please check that out?
>> >
>> > I do prefer fixing all these swapon-vs-swapoff races with some large,
>> > simple, wide-scope exclusion scheme. Perhaps SWP_USED is that scheme.
>> >
>> > An alternative would be to add another mutex and just make sys_swapon()
>> > and sys_swapoff() 100% exclusive. But that is plastering yet another
>> > lock over this mess to hide the horrors which lurk within :(
>> >
>>
>> Hi, Andrew. Thanks for your suggestion.
>>
>> I checked Krzysztof's patch, it use the global swapon_mutex to protect
>> race condition among
>> swapon, swapoff and swap_start(). It is a kind of correct method, but
>> a heavy method.
>
> But do you agree that your
> http://ozlabs.org/~akpm/mmots/broken-out/mm-swap-fix-race-on-swap_info-reuse-between-swapoff-and-swapon.patch
> makes Krzysztof's
> http://ozlabs.org/~akpm/mmots/broken-out/swap-fix-setting-page_size-blocksize-during-swapoff-swapon-race.patch
> obsolete?

Yes, I agree.

> I've been sitting on Krzysztof's
> swap-fix-setting-page_size-blocksize-during-swapoff-swapon-race.patch
> for several months - Hugh had issues with it so I put it on hold and
> nothing further happened.
>
>> I will try to resend a patchset to make lock usage in swapfile.c clear
>> and fine grit
>
> OK, thanks. In the meanwhile I'm planning on dropping Krzysztof's
> patch and merging your patch into 3.14-rc1, which is why I'd like
> confirmation that your patch addresses the issues which Krzysztof
> identified?
>

I think so, Krzysztof and I both try to fix the same issue(reuse
swap_info while its
previous resources are not cleared completely). The different is
Krzysztof's patch
uses a global swapon_mutex and its commit log only focuses on set_blocksize(),
while my patch try to maintain the fine grit lock usage.


\
 
 \ /
  Last update: 2014-01-13 05:21    [W:0.047 / U:0.328 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site