On Thu, Feb 23, 2017 at 06:28:01PM +0100, Peter Zijlstra wrote:
> I think Boris and Thomas were talking about moving the entire c1e
> detection out of there.
Yeah, see
07c94a381253 ("x86/amd: Simplify AMD E400 aware idle routine")
tglx and I got rid of reading that MSR in the idle routine.
--
On Thu, Feb 23, 2017 at 06:28:01PM +0100, Peter Zijlstra wrote:
> I think Boris and Thomas were talking about moving the entire c1e
> detection out of there.
Yeah, see
07c94a381253 ("x86/amd: Simplify AMD E400 aware idle routine")
tglx and I got rid of reading that MSR in the idle routine.
--
On Thu, Feb 23, 2017 at 01:24:34PM +0100, Jiri Olsa wrote:
> On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> >
> > I think I prefer something like the below, that only annotates the one
> > RDMSR in question, instead of all of them.
> >
> >
> > diff --git
On Thu, Feb 23, 2017 at 01:24:34PM +0100, Jiri Olsa wrote:
> On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> >
> > I think I prefer something like the below, that only annotates the one
> > RDMSR in question, instead of all of them.
> >
> >
> > diff --git
> > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > index 0888a879120f..d6c6aa80675f 100644
> > --- a/arch/x86/kernel/process.c
> > +++ b/arch/x86/kernel/process.c
> > @@ -357,7 +357,7 @@ static void amd_e400_idle(void)
> > if (!amd_e400_c1e_detected) {
> >
> > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > index 0888a879120f..d6c6aa80675f 100644
> > --- a/arch/x86/kernel/process.c
> > +++ b/arch/x86/kernel/process.c
> > @@ -357,7 +357,7 @@ static void amd_e400_idle(void)
> > if (!amd_e400_c1e_detected) {
> >
On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> > hi,
> > Jan hit following output when msr tracepoints are enabled on amd server:
> >
> > [ 91.585653] ===
> > [ 91.589840] [ INFO:
On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> > hi,
> > Jan hit following output when msr tracepoints are enabled on amd server:
> >
> > [ 91.585653] ===
> > [ 91.589840] [ INFO:
On Wed, 30 Nov 2016, Borislav Petkov wrote:
> On Wed, Nov 30, 2016 at 09:54:58AM +0100, Thomas Gleixner wrote:
> > Right, that's the safe bet. But I'm quite sure that the C1E crap only
> > starts to work _after_ ACPI initialization.
>
> Yap, I think it is an ACPI decision whether to enter C1E or
On Wed, 30 Nov 2016, Borislav Petkov wrote:
> On Wed, Nov 30, 2016 at 09:54:58AM +0100, Thomas Gleixner wrote:
> > Right, that's the safe bet. But I'm quite sure that the C1E crap only
> > starts to work _after_ ACPI initialization.
>
> Yap, I think it is an ACPI decision whether to enter C1E or
On Wed, Nov 30, 2016 at 09:54:58AM +0100, Thomas Gleixner wrote:
> Right, that's the safe bet. But I'm quite sure that the C1E crap only
> starts to work _after_ ACPI initialization.
Yap, I think it is an ACPI decision whether to enter C1E or not. And all
those boxes which are unaffected - they
On Wed, Nov 30, 2016 at 09:54:58AM +0100, Thomas Gleixner wrote:
> Right, that's the safe bet. But I'm quite sure that the C1E crap only
> starts to work _after_ ACPI initialization.
Yap, I think it is an ACPI decision whether to enter C1E or not. And all
those boxes which are unaffected - they
On Wed, 30 Nov 2016, Borislav Petkov wrote:
> On Tue, Nov 29, 2016 at 02:59:01PM +0100, Thomas Gleixner wrote:
> > The issue is that you obvioulsy start with the assumption, that the machine
> > has this bug. As a consequence the machine is brute forced into tick
> > broadcast mode, which cannot
On Wed, 30 Nov 2016, Borislav Petkov wrote:
> On Tue, Nov 29, 2016 at 02:59:01PM +0100, Thomas Gleixner wrote:
> > The issue is that you obvioulsy start with the assumption, that the machine
> > has this bug. As a consequence the machine is brute forced into tick
> > broadcast mode, which cannot
On Tue, Nov 29, 2016 at 02:59:01PM +0100, Thomas Gleixner wrote:
> The issue is that you obvioulsy start with the assumption, that the machine
> has this bug. As a consequence the machine is brute forced into tick
> broadcast mode, which cannot be reverted when you clear that misfeature
> after
On Tue, Nov 29, 2016 at 02:59:01PM +0100, Thomas Gleixner wrote:
> The issue is that you obvioulsy start with the assumption, that the machine
> has this bug. As a consequence the machine is brute forced into tick
> broadcast mode, which cannot be reverted when you clear that misfeature
> after
On Tue, Nov 29, 2016 at 02:16:49PM +0100, Borislav Petkov wrote:
> On Mon, Nov 21, 2016 at 05:06:54PM +0100, Borislav Petkov wrote:
> > IOW, what's the worst thing that can happen if we did this below?
> >
> > We basically get rid of the detection and switch the timer to broadcast
> > mode
On Tue, Nov 29, 2016 at 02:16:49PM +0100, Borislav Petkov wrote:
> On Mon, Nov 21, 2016 at 05:06:54PM +0100, Borislav Petkov wrote:
> > IOW, what's the worst thing that can happen if we did this below?
> >
> > We basically get rid of the detection and switch the timer to broadcast
> > mode
On Tue, 29 Nov 2016, Borislav Petkov wrote:
> On Mon, Nov 21, 2016 at 05:06:54PM +0100, Borislav Petkov wrote:
> > IOW, what's the worst thing that can happen if we did this below?
> >
> > We basically get rid of the detection and switch the timer to broadcast
> > mode immediately on the halting
On Tue, 29 Nov 2016, Borislav Petkov wrote:
> On Mon, Nov 21, 2016 at 05:06:54PM +0100, Borislav Petkov wrote:
> > IOW, what's the worst thing that can happen if we did this below?
> >
> > We basically get rid of the detection and switch the timer to broadcast
> > mode immediately on the halting
On Mon, Nov 21, 2016 at 05:06:54PM +0100, Borislav Petkov wrote:
> IOW, what's the worst thing that can happen if we did this below?
>
> We basically get rid of the detection and switch the timer to broadcast
> mode immediately on the halting CPU.
>
> amd_e400_idle() is behind an "if
On Mon, Nov 21, 2016 at 05:06:54PM +0100, Borislav Petkov wrote:
> IOW, what's the worst thing that can happen if we did this below?
>
> We basically get rid of the detection and switch the timer to broadcast
> mode immediately on the halting CPU.
>
> amd_e400_idle() is behind an "if
> Well, actually it's been optimized a lot from the first instances.
> Note, tracepoints need to be quite generic, so where it can be optimized
> is not obvious.
There is quite a bit of stuff that doesn't change from trace point to
trace point. So you could just cache all these decisions into a
> Well, actually it's been optimized a lot from the first instances.
> Note, tracepoints need to be quite generic, so where it can be optimized
> is not obvious.
There is quite a bit of stuff that doesn't change from trace point to
trace point. So you could just cache all these decisions into a
On Mon, 28 Nov 2016 13:48:25 -0800
Andi Kleen wrote:
> > I took a look at this and forced some more functions to be inlined. I
> > did a little tweaking here and there. Could you pull my tree and see if
> > things are better? I don't currently have the hardware to run this
On Mon, 28 Nov 2016 13:48:25 -0800
Andi Kleen wrote:
> > I took a look at this and forced some more functions to be inlined. I
> > did a little tweaking here and there. Could you pull my tree and see if
> > things are better? I don't currently have the hardware to run this
> > myself.
> >
> >
> I took a look at this and forced some more functions to be inlined. I
> did a little tweaking here and there. Could you pull my tree and see if
> things are better? I don't currently have the hardware to run this
> myself.
>
>
> I took a look at this and forced some more functions to be inlined. I
> did a little tweaking here and there. Could you pull my tree and see if
> things are better? I don't currently have the hardware to run this
> myself.
>
>
On Mon, 21 Nov 2016 10:37:00 -0800
Andi Kleen wrote:
> > It tries to be optimized. I "unoptimized" it a while back to pull out
> > all the inlines that were done in the tracepoint itself. That is, the
> > trace_() function is inlined in the code itself. By
> > breaking that
On Mon, 21 Nov 2016 10:37:00 -0800
Andi Kleen wrote:
> > It tries to be optimized. I "unoptimized" it a while back to pull out
> > all the inlines that were done in the tracepoint itself. That is, the
> > trace_() function is inlined in the code itself. By
> > breaking that up a bit, I was able
On Tue, Nov 22, 2016 at 09:39:45AM -0500, Steven Rostedt wrote:
> On Mon, 21 Nov 2016 10:37:00 -0800
> Andi Kleen wrote:
>
>
> > Here is it again untruncated:
> >
> > http://halobates.de/tracepoint-trace
>
> BTW, what tool did you use to generate this?
perf and Intel PT,
On Tue, Nov 22, 2016 at 09:39:45AM -0500, Steven Rostedt wrote:
> On Mon, 21 Nov 2016 10:37:00 -0800
> Andi Kleen wrote:
>
>
> > Here is it again untruncated:
> >
> > http://halobates.de/tracepoint-trace
>
> BTW, what tool did you use to generate this?
perf and Intel PT, with the
On Mon, 21 Nov 2016 10:37:00 -0800
Andi Kleen wrote:
> Here is it again untruncated:
>
> http://halobates.de/tracepoint-trace
BTW, what tool did you use to generate this?
-- Steve
On Mon, 21 Nov 2016 10:37:00 -0800
Andi Kleen wrote:
> Here is it again untruncated:
>
> http://halobates.de/tracepoint-trace
BTW, what tool did you use to generate this?
-- Steve
On Mon, Nov 21, 2016 at 12:44:45PM -0800, Andi Kleen wrote:
> > > http://halobates.de/tracepoint-trace
> >
> > There's a lot of push and pop regs due to function calling. There's
> > places that inlines can still improve things, and perhaps even some
> > likely unlikelys well placed.
>
>
On Mon, Nov 21, 2016 at 12:44:45PM -0800, Andi Kleen wrote:
> > > http://halobates.de/tracepoint-trace
> >
> > There's a lot of push and pop regs due to function calling. There's
> > places that inlines can still improve things, and perhaps even some
> > likely unlikelys well placed.
>
>
> > http://halobates.de/tracepoint-trace
>
> There's a lot of push and pop regs due to function calling. There's
> places that inlines can still improve things, and perhaps even some
> likely unlikelys well placed.
Assuming you avoid all the push/pop and all the call/ret this would only be
~25%
> > http://halobates.de/tracepoint-trace
>
> There's a lot of push and pop regs due to function calling. There's
> places that inlines can still improve things, and perhaps even some
> likely unlikelys well placed.
Assuming you avoid all the push/pop and all the call/ret this would only be
~25%
On Mon, Nov 21, 2016 at 01:24:38PM -0500, Steven Rostedt wrote:
>
> Paul,
>
>
> On Mon, 21 Nov 2016 12:55:01 -0500
> Steven Rostedt wrote:
>
> > On Mon, 21 Nov 2016 18:18:53 +0100
> > Peter Zijlstra wrote:
> >
> > > Its not ftrace as such though,
On Mon, Nov 21, 2016 at 01:24:38PM -0500, Steven Rostedt wrote:
>
> Paul,
>
>
> On Mon, 21 Nov 2016 12:55:01 -0500
> Steven Rostedt wrote:
>
> > On Mon, 21 Nov 2016 18:18:53 +0100
> > Peter Zijlstra wrote:
> >
> > > Its not ftrace as such though, its RCU, ftrace simply uses RCU to avoid
> >
On Mon, 21 Nov 2016 10:37:00 -0800
Andi Kleen wrote:
> Just adding a few inlines won't fix the gigantic bloat that is currently
> there. See the PT trace I posted earlier (it was even truncated, it's
> actually worse). Just a single enabled trace point took about a us.
If
On Mon, 21 Nov 2016 10:37:00 -0800
Andi Kleen wrote:
> Just adding a few inlines won't fix the gigantic bloat that is currently
> there. See the PT trace I posted earlier (it was even truncated, it's
> actually worse). Just a single enabled trace point took about a us.
If you want to play with
On Mon, 21 Nov 2016 10:37:00 -0800
Andi Kleen wrote:
> Ok so how should tracing in idle code work then in your opinion?
As I suggested already. If we can get a light weight rcu_is_watching()
then we can do the rcu_idle work when needed, and not when we don't
need it. Sure
On Mon, 21 Nov 2016 10:37:00 -0800
Andi Kleen wrote:
> Ok so how should tracing in idle code work then in your opinion?
As I suggested already. If we can get a light weight rcu_is_watching()
then we can do the rcu_idle work when needed, and not when we don't
need it. Sure this will add more
> It tries to be optimized. I "unoptimized" it a while back to pull out
> all the inlines that were done in the tracepoint itself. That is, the
> trace_() function is inlined in the code itself. By
> breaking that up a bit, I was able to save a bunch of text because the
> tracepoints were bloating
> It tries to be optimized. I "unoptimized" it a while back to pull out
> all the inlines that were done in the tracepoint itself. That is, the
> trace_() function is inlined in the code itself. By
> breaking that up a bit, I was able to save a bunch of text because the
> tracepoints were bloating
Paul,
On Mon, 21 Nov 2016 12:55:01 -0500
Steven Rostedt wrote:
> On Mon, 21 Nov 2016 18:18:53 +0100
> Peter Zijlstra wrote:
>
> > Its not ftrace as such though, its RCU, ftrace simply uses RCU to avoid
> > locking, as one does.
>
> Just to be
Paul,
On Mon, 21 Nov 2016 12:55:01 -0500
Steven Rostedt wrote:
> On Mon, 21 Nov 2016 18:18:53 +0100
> Peter Zijlstra wrote:
>
> > Its not ftrace as such though, its RCU, ftrace simply uses RCU to avoid
> > locking, as one does.
>
> Just to be clear, as ftrace in the kernel mostly
On Mon, 21 Nov 2016 10:06:54 -0800
Andi Kleen wrote:
> > And a popf can be much more expensive than any of these. You should
> > know, not all instructions are equal.
> >
> > Using perf, I've seen popf take up almst 30% of a function the size of
> > this.
>
> In any case
On Mon, 21 Nov 2016 10:06:54 -0800
Andi Kleen wrote:
> > And a popf can be much more expensive than any of these. You should
> > know, not all instructions are equal.
> >
> > Using perf, I've seen popf take up almst 30% of a function the size of
> > this.
>
> In any case it's a small
> And a popf can be much more expensive than any of these. You should
> know, not all instructions are equal.
>
> Using perf, I've seen popf take up almst 30% of a function the size of
> this.
In any case it's a small fraction of the 600+ instructions which are currently
executed for every
> And a popf can be much more expensive than any of these. You should
> know, not all instructions are equal.
>
> Using perf, I've seen popf take up almst 30% of a function the size of
> this.
In any case it's a small fraction of the 600+ instructions which are currently
executed for every
On Mon, 21 Nov 2016 09:45:04 -0800
Andi Kleen wrote:
> On Mon, Nov 21, 2016 at 06:18:53PM +0100, Peter Zijlstra wrote:
> > On Mon, Nov 21, 2016 at 09:06:13AM -0800, Andi Kleen wrote:
> > > > > it got away with attached change.. but this rcu logic
> > > > > is far beyond
On Mon, 21 Nov 2016 09:45:04 -0800
Andi Kleen wrote:
> On Mon, Nov 21, 2016 at 06:18:53PM +0100, Peter Zijlstra wrote:
> > On Mon, Nov 21, 2016 at 09:06:13AM -0800, Andi Kleen wrote:
> > > > > it got away with attached change.. but this rcu logic
> > > > > is far beyond me, so it's just wild
On Mon, 21 Nov 2016 18:18:53 +0100
Peter Zijlstra wrote:
> Its not ftrace as such though, its RCU, ftrace simply uses RCU to avoid
> locking, as one does.
Just to be clear, as ftrace in the kernel mostly represents function
tracing, which doesn't use RCU. This is a
On Mon, 21 Nov 2016 18:18:53 +0100
Peter Zijlstra wrote:
> Its not ftrace as such though, its RCU, ftrace simply uses RCU to avoid
> locking, as one does.
Just to be clear, as ftrace in the kernel mostly represents function
tracing, which doesn't use RCU. This is a tracepoint feature.
>
>
On Mon, Nov 21, 2016 at 06:18:53PM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 09:06:13AM -0800, Andi Kleen wrote:
> > > > it got away with attached change.. but this rcu logic
> > > > is far beyond me, so it's just wild guess.. ;-)
> > >
> > > I think I prefer something like the
On Mon, Nov 21, 2016 at 06:18:53PM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 09:06:13AM -0800, Andi Kleen wrote:
> > > > it got away with attached change.. but this rcu logic
> > > > is far beyond me, so it's just wild guess.. ;-)
> > >
> > > I think I prefer something like the
On Mon, Nov 21, 2016 at 09:06:13AM -0800, Andi Kleen wrote:
> > > it got away with attached change.. but this rcu logic
> > > is far beyond me, so it's just wild guess.. ;-)
> >
> > I think I prefer something like the below, that only annotates the one
> > RDMSR in question, instead of all of
On Mon, Nov 21, 2016 at 09:06:13AM -0800, Andi Kleen wrote:
> > > it got away with attached change.. but this rcu logic
> > > is far beyond me, so it's just wild guess.. ;-)
> >
> > I think I prefer something like the below, that only annotates the one
> > RDMSR in question, instead of all of
> > it got away with attached change.. but this rcu logic
> > is far beyond me, so it's just wild guess.. ;-)
>
> I think I prefer something like the below, that only annotates the one
> RDMSR in question, instead of all of them.
It would be far better to just fix trace points that they always
> > it got away with attached change.. but this rcu logic
> > is far beyond me, so it's just wild guess.. ;-)
>
> I think I prefer something like the below, that only annotates the one
> RDMSR in question, instead of all of them.
It would be far better to just fix trace points that they always
+ tglx.
On Mon, Nov 21, 2016 at 04:41:04PM +0100, Peter Zijlstra wrote:
> Right, I just wondered about the !c1e present case. In that case we'll
> forever read the msr in the hope of finding that bit set, and we never
> will.
Well, erratum 400 flag is set on everything F10h from model 0x2
+ tglx.
On Mon, Nov 21, 2016 at 04:41:04PM +0100, Peter Zijlstra wrote:
> Right, I just wondered about the !c1e present case. In that case we'll
> forever read the msr in the hope of finding that bit set, and we never
> will.
Well, erratum 400 flag is set on everything F10h from model 0x2
On Mon, Nov 21, 2016 at 04:35:38PM +0100, Borislav Petkov wrote:
> On Mon, Nov 21, 2016 at 03:37:16PM +0100, Peter Zijlstra wrote:
> > Yeah, but this one does a printk() when it hits the contidion it checks
> > for, so not tracing it would be fine I think.
>
> FWIW, there's already:
>
> static
On Mon, Nov 21, 2016 at 04:35:38PM +0100, Borislav Petkov wrote:
> On Mon, Nov 21, 2016 at 03:37:16PM +0100, Peter Zijlstra wrote:
> > Yeah, but this one does a printk() when it hits the contidion it checks
> > for, so not tracing it would be fine I think.
>
> FWIW, there's already:
>
> static
On Mon, Nov 21, 2016 at 03:37:16PM +0100, Peter Zijlstra wrote:
> Yeah, but this one does a printk() when it hits the contidion it checks
> for, so not tracing it would be fine I think.
FWIW, there's already:
static inline void notrace
native_write_msr(unsigned int msr, u32 low, u32 high)
{
On Mon, Nov 21, 2016 at 03:37:16PM +0100, Peter Zijlstra wrote:
> Yeah, but this one does a printk() when it hits the contidion it checks
> for, so not tracing it would be fine I think.
FWIW, there's already:
static inline void notrace
native_write_msr(unsigned int msr, u32 low, u32 high)
{
On Mon, Nov 21, 2016 at 09:15:43AM -0500, Steven Rostedt wrote:
> On Mon, 21 Nov 2016 13:58:30 +0100
> Peter Zijlstra wrote:
>
> > On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> >
> > > > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> >
On Mon, Nov 21, 2016 at 09:15:43AM -0500, Steven Rostedt wrote:
> On Mon, 21 Nov 2016 13:58:30 +0100
> Peter Zijlstra wrote:
>
> > On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> >
> > > > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > > > index
On Mon, 21 Nov 2016 13:58:30 +0100
Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
>
> > > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > > index 0888a879120f..d6c6aa80675f 100644
> > > ---
On Mon, 21 Nov 2016 13:58:30 +0100
Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
>
> > > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > > index 0888a879120f..d6c6aa80675f 100644
> > > --- a/arch/x86/kernel/process.c
> > > +++
On Mon, 21 Nov 2016 13:58:30 +0100
Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
>
> > > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > > index 0888a879120f..d6c6aa80675f 100644
> > > ---
On Mon, 21 Nov 2016 13:58:30 +0100
Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
>
> > > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > > index 0888a879120f..d6c6aa80675f 100644
> > > --- a/arch/x86/kernel/process.c
> > > +++
On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > index 0888a879120f..d6c6aa80675f 100644
> > --- a/arch/x86/kernel/process.c
> > +++ b/arch/x86/kernel/process.c
> > @@ -357,7 +357,7 @@ static void
On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> > diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
> > index 0888a879120f..d6c6aa80675f 100644
> > --- a/arch/x86/kernel/process.c
> > +++ b/arch/x86/kernel/process.c
> > @@ -357,7 +357,7 @@ static void
On Mon, Nov 21, 2016 at 12:31:31PM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 12:22:29PM +0100, Jiri Olsa wrote:
> > On Mon, Nov 21, 2016 at 10:42:27AM +0100, Peter Zijlstra wrote:
> > > On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> > > > On Mon, Nov 21, 2016 at
On Mon, Nov 21, 2016 at 12:31:31PM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 12:22:29PM +0100, Jiri Olsa wrote:
> > On Mon, Nov 21, 2016 at 10:42:27AM +0100, Peter Zijlstra wrote:
> > > On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> > > > On Mon, Nov 21, 2016 at
On Mon, Nov 21, 2016 at 12:22:29PM +0100, Jiri Olsa wrote:
> On Mon, Nov 21, 2016 at 10:42:27AM +0100, Peter Zijlstra wrote:
> > On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> > > On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> >
> > > > I think I prefer something
On Mon, Nov 21, 2016 at 12:22:29PM +0100, Jiri Olsa wrote:
> On Mon, Nov 21, 2016 at 10:42:27AM +0100, Peter Zijlstra wrote:
> > On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> > > On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> >
> > > > I think I prefer something
On Mon, Nov 21, 2016 at 10:42:27AM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> > On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
>
> > > I think I prefer something like the below, that only annotates the one
> > > RDMSR in question,
On Mon, Nov 21, 2016 at 10:42:27AM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> > On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
>
> > > I think I prefer something like the below, that only annotates the one
> > > RDMSR in question,
On Mon, Nov 21, 2016 at 10:43:21AM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 01:02:25AM -0800, Paul E. McKenney wrote:
> > On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> > >
> > > it got away with attached change.. but this rcu logic
> > > is far beyond me, so it's just
On Mon, Nov 21, 2016 at 10:43:21AM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 01:02:25AM -0800, Paul E. McKenney wrote:
> > On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> > >
> > > it got away with attached change.. but this rcu logic
> > > is far beyond me, so it's just
On Mon, Nov 21, 2016 at 01:02:25AM -0800, Paul E. McKenney wrote:
> On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> >
> > it got away with attached change.. but this rcu logic
> > is far beyond me, so it's just wild guess.. ;-)
>
> If in idle, the _rcuidle() is needed, so:
Well,
On Mon, Nov 21, 2016 at 01:02:25AM -0800, Paul E. McKenney wrote:
> On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> >
> > it got away with attached change.. but this rcu logic
> > is far beyond me, so it's just wild guess.. ;-)
>
> If in idle, the _rcuidle() is needed, so:
Well,
On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> > I think I prefer something like the below, that only annotates the one
> > RDMSR in question, instead of all of them.
>
> I was wondering about that, but haven't found
On Mon, Nov 21, 2016 at 10:34:25AM +0100, Jiri Olsa wrote:
> On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> > I think I prefer something like the below, that only annotates the one
> > RDMSR in question, instead of all of them.
>
> I was wondering about that, but haven't found
On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> > hi,
> > Jan hit following output when msr tracepoints are enabled on amd server:
> >
> > [ 91.585653] ===
> > [ 91.589840] [ INFO:
On Mon, Nov 21, 2016 at 10:28:50AM +0100, Peter Zijlstra wrote:
> On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> > hi,
> > Jan hit following output when msr tracepoints are enabled on amd server:
> >
> > [ 91.585653] ===
> > [ 91.589840] [ INFO:
On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> hi,
> Jan hit following output when msr tracepoints are enabled on amd server:
>
> [ 91.585653] ===
> [ 91.589840] [ INFO: suspicious RCU usage. ]
> [ 91.594025] 4.9.0-rc1+ #1 Not tainted
> [
On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> hi,
> Jan hit following output when msr tracepoints are enabled on amd server:
>
> [ 91.585653] ===
> [ 91.589840] [ INFO: suspicious RCU usage. ]
> [ 91.594025] 4.9.0-rc1+ #1 Not tainted
> [
On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> hi,
> Jan hit following output when msr tracepoints are enabled on amd server:
>
> [ 91.585653] ===
> [ 91.589840] [ INFO: suspicious RCU usage. ]
> [ 91.594025] 4.9.0-rc1+ #1 Not tainted
> [
On Mon, Nov 21, 2016 at 01:53:43AM +0100, Jiri Olsa wrote:
> hi,
> Jan hit following output when msr tracepoints are enabled on amd server:
>
> [ 91.585653] ===
> [ 91.589840] [ INFO: suspicious RCU usage. ]
> [ 91.594025] 4.9.0-rc1+ #1 Not tainted
> [
hi,
Jan hit following output when msr tracepoints are enabled on amd server:
[ 91.585653] ===
[ 91.589840] [ INFO: suspicious RCU usage. ]
[ 91.594025] 4.9.0-rc1+ #1 Not tainted
[ 91.597691] ---
[ 91.601877]
hi,
Jan hit following output when msr tracepoints are enabled on amd server:
[ 91.585653] ===
[ 91.589840] [ INFO: suspicious RCU usage. ]
[ 91.594025] 4.9.0-rc1+ #1 Not tainted
[ 91.597691] ---
[ 91.601877]
96 matches
Mail list logo