| Message ID | 20150304152243.GG3122@htj.duckdns.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed 04-03-15 10:22:43, Tejun Heo wrote: > global_update_bandwidth() uses static variable update_time as the > timestamp for the last update but forgets to initialize it to > INITIALIZE_JIFFIES. > > This means that global_dirty_limit will be 5 mins into the future on > 32bit and some large amount jiffies into the past on 64bit. This > isn't critical as the only effect is that global_dirty_limit won't be > updated for the first 5 mins after booting on 32bit machines, > especially given the auxiliary nature of global_dirty_limit's role - > protecting against global dirty threshold's sudden dips; however, it > does lead to unintended suboptimal behavior. Fix it. Looks good. You can add: Reviewed-by: Jan Kara <jack@suse.cz> Honza > > Signed-off-by: Tejun Heo <tj@kernel.org> > Cc: Wu Fengguang <fengguang.wu@intel.com> > Cc: Jan Kara <jack@suse.cz> > Cc: Jens Axboe <axboe@kernel.dk> > Cc: stable@vger.kernel.org > --- > mm/page-writeback.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > --- a/mm/page-writeback.c > +++ b/mm/page-writeback.c > @@ -922,7 +922,7 @@ static void global_update_bandwidth(unsi > unsigned long now) > { > static DEFINE_SPINLOCK(dirty_lock); > - static unsigned long update_time; > + static unsigned long update_time = INITIAL_JIFFIES; > > /* > * check locklessly first to optimize away locking for the most time
On 03/04/2015 08:30 AM, Jan Kara wrote: > On Wed 04-03-15 10:22:43, Tejun Heo wrote: >> global_update_bandwidth() uses static variable update_time as the >> timestamp for the last update but forgets to initialize it to >> INITIALIZE_JIFFIES. >> >> This means that global_dirty_limit will be 5 mins into the future on >> 32bit and some large amount jiffies into the past on 64bit. This >> isn't critical as the only effect is that global_dirty_limit won't be >> updated for the first 5 mins after booting on 32bit machines, >> especially given the auxiliary nature of global_dirty_limit's role - >> protecting against global dirty threshold's sudden dips; however, it >> does lead to unintended suboptimal behavior. Fix it. > Looks good. You can add: > Reviewed-by: Jan Kara <jack@suse.cz> We should add that it fixes c42843f2f0bbc (from 2011!) as well.
--- a/mm/page-writeback.c +++ b/mm/page-writeback.c @@ -922,7 +922,7 @@ static void global_update_bandwidth(unsi unsigned long now) { static DEFINE_SPINLOCK(dirty_lock); - static unsigned long update_time; + static unsigned long update_time = INITIAL_JIFFIES; /* * check locklessly first to optimize away locking for the most time
global_update_bandwidth() uses static variable update_time as the timestamp for the last update but forgets to initialize it to INITIALIZE_JIFFIES. This means that global_dirty_limit will be 5 mins into the future on 32bit and some large amount jiffies into the past on 64bit. This isn't critical as the only effect is that global_dirty_limit won't be updated for the first 5 mins after booting on 32bit machines, especially given the auxiliary nature of global_dirty_limit's role - protecting against global dirty threshold's sudden dips; however, it does lead to unintended suboptimal behavior. Fix it. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Jan Kara <jack@suse.cz> Cc: Jens Axboe <axboe@kernel.dk> Cc: stable@vger.kernel.org --- mm/page-writeback.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html