Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-14 Thread Paul E. McKenney
On Tue, May 14, 2013 at 07:37:39PM +0200, Joerg Roedel wrote: > Hi Paul, > > On Mon, May 13, 2013 at 07:00:05AM -0700, Paul E. McKenney wrote: > > rcu: Fix comparison sense in rcu_needs_cpu() > > > > Commit c0f4dfd4f (rcu: Make RCU_FAST_NO_HZ take advantage of numbered > > callbacks) introduced

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-14 Thread Joerg Roedel
Hi Paul, On Mon, May 13, 2013 at 07:00:05AM -0700, Paul E. McKenney wrote: > rcu: Fix comparison sense in rcu_needs_cpu() > > Commit c0f4dfd4f (rcu: Make RCU_FAST_NO_HZ take advantage of numbered > callbacks) introduced a bug that can result in excessively long grace > periods. This bug reverse

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-14 Thread Paul E. McKenney
On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: > "Paul E. McKenney" writes: > > > We don't want to back out the RCU_FAST_NO_HZ changes due to their > > energy-efficiency benefits. > > Yes, that's what I assumed. Just didn't know where to start dissecting > it... > > > So could

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-14 Thread Paul E. McKenney
On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: Paul E. McKenney paul...@linux.vnet.ibm.com writes: We don't want to back out the RCU_FAST_NO_HZ changes due to their energy-efficiency benefits. Yes, that's what I assumed. Just didn't know where to start dissecting it...

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-14 Thread Joerg Roedel
Hi Paul, On Mon, May 13, 2013 at 07:00:05AM -0700, Paul E. McKenney wrote: rcu: Fix comparison sense in rcu_needs_cpu() Commit c0f4dfd4f (rcu: Make RCU_FAST_NO_HZ take advantage of numbered callbacks) introduced a bug that can result in excessively long grace periods. This bug reverse the

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-14 Thread Paul E. McKenney
On Tue, May 14, 2013 at 07:37:39PM +0200, Joerg Roedel wrote: Hi Paul, On Mon, May 13, 2013 at 07:00:05AM -0700, Paul E. McKenney wrote: rcu: Fix comparison sense in rcu_needs_cpu() Commit c0f4dfd4f (rcu: Make RCU_FAST_NO_HZ take advantage of numbered callbacks) introduced a bug that

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-13 Thread Paul E. McKenney
On Sun, May 12, 2013 at 06:35:18PM -0700, Paul E. McKenney wrote: > On Sun, May 12, 2013 at 11:34:30PM +0200, Joerg Roedel wrote: > > On Sun, May 12, 2013 at 01:57:48PM -0700, Paul E. McKenney wrote: > > > That does look pretty extreme! If you build with CONFIG_RCU_NO_HZ=n, > > > but without the

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-13 Thread Rafael J. Wysocki
On Sunday, May 12, 2013 07:20:58 PM Borislav Petkov wrote: > On Sun, May 12, 2013 at 06:56:40PM +0200, Bjørn Mork wrote: > > Borislav Petkov writes: > > > On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: > > >> diff --git a/kernel/rcutree.c b/kernel/rcutree.c > > >> index

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-13 Thread Joerg Roedel
On Sun, May 12, 2013 at 06:35:18PM -0700, Paul E. McKenney wrote: > On Sun, May 12, 2013 at 11:34:30PM +0200, Joerg Roedel wrote: > > With CONFIG_RCU_NO_HZ=n (and no revert) the system boots fine, no boot > > delay. I also enabled some RCU debugging options (with > > CONFIG_RCU_NO_HZ=y), but

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-13 Thread Joerg Roedel
On Sun, May 12, 2013 at 06:35:18PM -0700, Paul E. McKenney wrote: On Sun, May 12, 2013 at 11:34:30PM +0200, Joerg Roedel wrote: With CONFIG_RCU_NO_HZ=n (and no revert) the system boots fine, no boot delay. I also enabled some RCU debugging options (with CONFIG_RCU_NO_HZ=y), but didn't see

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-13 Thread Rafael J. Wysocki
On Sunday, May 12, 2013 07:20:58 PM Borislav Petkov wrote: On Sun, May 12, 2013 at 06:56:40PM +0200, Bjørn Mork wrote: Borislav Petkov b...@alien8.de writes: On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: diff --git a/kernel/rcutree.c b/kernel/rcutree.c index

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-13 Thread Paul E. McKenney
On Sun, May 12, 2013 at 06:35:18PM -0700, Paul E. McKenney wrote: On Sun, May 12, 2013 at 11:34:30PM +0200, Joerg Roedel wrote: On Sun, May 12, 2013 at 01:57:48PM -0700, Paul E. McKenney wrote: That does look pretty extreme! If you build with CONFIG_RCU_NO_HZ=n, but without the revert,

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Paul E. McKenney
On Sun, May 12, 2013 at 11:34:30PM +0200, Joerg Roedel wrote: > On Sun, May 12, 2013 at 01:57:48PM -0700, Paul E. McKenney wrote: > > That does look pretty extreme! If you build with CONFIG_RCU_NO_HZ=n, > > but without the revert, do you still get the delays? > > With CONFIG_RCU_NO_HZ=n (and no

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Joerg Roedel
On Sun, May 12, 2013 at 01:57:48PM -0700, Paul E. McKenney wrote: > That does look pretty extreme! If you build with CONFIG_RCU_NO_HZ=n, > but without the revert, do you still get the delays? With CONFIG_RCU_NO_HZ=n (and no revert) the system boots fine, no boot delay. I also enabled some RCU

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Paul E. McKenney
On Sun, May 12, 2013 at 08:29:35PM +0200, Joerg Roedel wrote: > Hi Paul, > > On Sun, May 12, 2013 at 04:31:57AM -0700, Paul E. McKenney wrote: > > On Sat, May 11, 2013 at 08:04:50PM +0200, Bjørn Mork wrote: > > > Bisecting it ended up pointing to > > > > > > commit

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Joerg Roedel
Hi Paul, On Sun, May 12, 2013 at 04:31:57AM -0700, Paul E. McKenney wrote: > On Sat, May 11, 2013 at 08:04:50PM +0200, Bjørn Mork wrote: > > Bisecting it ended up pointing to > > > > commit c0f4dfd4f90f1667d234d21f15153ea09a2eaa66 > > Author: Paul E. McKenney > > Date: Fri Dec 28 11:30:36

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Borislav Petkov
On Sun, May 12, 2013 at 06:56:40PM +0200, Bjørn Mork wrote: > Borislav Petkov writes: > > On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: > >> diff --git a/kernel/rcutree.c b/kernel/rcutree.c > >> index 6934238..2dcbf84 100644 > >> --- a/kernel/rcutree.c > >> +++ b/kernel/rcutree.c >

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Bjørn Mork
Borislav Petkov writes: > On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: >> diff --git a/kernel/rcutree.c b/kernel/rcutree.c >> index 6934238..2dcbf84 100644 >> --- a/kernel/rcutree.c >> +++ b/kernel/rcutree.c >> @@ -3103,9 +3103,11 @@ static int rcu_pm_notify(struct notifier_block

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Borislav Petkov
On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: > diff --git a/kernel/rcutree.c b/kernel/rcutree.c > index 6934238..2dcbf84 100644 > --- a/kernel/rcutree.c > +++ b/kernel/rcutree.c > @@ -3103,9 +3103,11 @@ static int rcu_pm_notify(struct notifier_block *self, > { > switch

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Bjørn Mork
"Paul E. McKenney" writes: > We don't want to back out the RCU_FAST_NO_HZ changes due to their > energy-efficiency benefits. Yes, that's what I assumed. Just didn't know where to start dissecting it... > So could you please try out Borislav's > patch below? He ran into the same issue a few

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Paul E. McKenney
On Sat, May 11, 2013 at 08:04:50PM +0200, Bjørn Mork wrote: > Hello, > > resuming from system suspend is intolerably slow in current mainline. I > am not the most patient person around, and I actually started out > bisecting this believing it was hanging... Turned out it wasn't really > hanging.

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Paul E. McKenney
On Sat, May 11, 2013 at 08:04:50PM +0200, Bjørn Mork wrote: Hello, resuming from system suspend is intolerably slow in current mainline. I am not the most patient person around, and I actually started out bisecting this believing it was hanging... Turned out it wasn't really hanging. It

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Bjørn Mork
Paul E. McKenney paul...@linux.vnet.ibm.com writes: We don't want to back out the RCU_FAST_NO_HZ changes due to their energy-efficiency benefits. Yes, that's what I assumed. Just didn't know where to start dissecting it... So could you please try out Borislav's patch below? He ran into

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Borislav Petkov
On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 6934238..2dcbf84 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -3103,9 +3103,11 @@ static int rcu_pm_notify(struct notifier_block *self, { switch (action) {

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Bjørn Mork
Borislav Petkov b...@alien8.de writes: On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 6934238..2dcbf84 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -3103,9 +3103,11 @@ static int rcu_pm_notify(struct

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Borislav Petkov
On Sun, May 12, 2013 at 06:56:40PM +0200, Bjørn Mork wrote: Borislav Petkov b...@alien8.de writes: On Sun, May 12, 2013 at 06:13:34PM +0200, Bjørn Mork wrote: diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 6934238..2dcbf84 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Joerg Roedel
Hi Paul, On Sun, May 12, 2013 at 04:31:57AM -0700, Paul E. McKenney wrote: On Sat, May 11, 2013 at 08:04:50PM +0200, Bjørn Mork wrote: Bisecting it ended up pointing to commit c0f4dfd4f90f1667d234d21f15153ea09a2eaa66 Author: Paul E. McKenney paul.mcken...@linaro.org Date: Fri Dec 28

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Paul E. McKenney
On Sun, May 12, 2013 at 08:29:35PM +0200, Joerg Roedel wrote: Hi Paul, On Sun, May 12, 2013 at 04:31:57AM -0700, Paul E. McKenney wrote: On Sat, May 11, 2013 at 08:04:50PM +0200, Bjørn Mork wrote: Bisecting it ended up pointing to commit c0f4dfd4f90f1667d234d21f15153ea09a2eaa66

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Joerg Roedel
On Sun, May 12, 2013 at 01:57:48PM -0700, Paul E. McKenney wrote: That does look pretty extreme! If you build with CONFIG_RCU_NO_HZ=n, but without the revert, do you still get the delays? With CONFIG_RCU_NO_HZ=n (and no revert) the system boots fine, no boot delay. I also enabled some RCU

Re: Bisected post-3.9 regression: Resume takes 5 times as much time as with v3.9

2013-05-12 Thread Paul E. McKenney
On Sun, May 12, 2013 at 11:34:30PM +0200, Joerg Roedel wrote: On Sun, May 12, 2013 at 01:57:48PM -0700, Paul E. McKenney wrote: That does look pretty extreme! If you build with CONFIG_RCU_NO_HZ=n, but without the revert, do you still get the delays? With CONFIG_RCU_NO_HZ=n (and no revert)