On Wed, Mar 12, 2014 at 07:15:03AM +0100, Peter Zijlstra wrote:
> On Wed, Mar 12, 2014 at 01:31:53PM +1100, Dave Chinner wrote:
> > With the queuing spinlock, I expected to see somewhat better
> > results, but I didn't at first. Turns out if you have any sort of
> > lock debugging turned on, then
On 03/12/2014 02:24 AM, Peter Zijlstra wrote:
On Tue, Mar 11, 2014 at 11:17:46PM -0400, Waiman Long wrote:
Except that I do a single atomic short integer write to switch the bits
instead of 2 byte write.
D'0h why didn't I think of that. A single short write is much better
than the 2 byte
On Wed, Mar 12, 2014 at 12:19:33PM -0400, Steven Rostedt wrote:
> On Wed, 12 Mar 2014 16:57:54 +0100
> Peter Zijlstra wrote:
>
>
> > I build and booted these patches against tip/master as of this morning.
> > It build ~50 defconfig kernels on my WSM-EP.
> >
> > I've not managed to reproduce
On Wed, 12 Mar 2014 16:57:54 +0100
Peter Zijlstra wrote:
> I build and booted these patches against tip/master as of this morning.
> It build ~50 defconfig kernels on my WSM-EP.
>
> I've not managed to reproduce the hang as reported by Dave, but then I
> doubt building kernels has anywhere
On Wed, Mar 12, 2014 at 8:57 AM, Peter Zijlstra wrote:
>
> I've not managed to reproduce the hang as reported by Dave, but then I
> doubt building kernels has anywhere near the lock contention he
> generated.
Building kernels has basically no lock contention at all.
Linus
--
To
On Wed, Mar 12, 2014 at 06:07:30AM -0400, Steven Rostedt wrote:
> On Wed, 12 Mar 2014 15:26:16 +1100
> Dave Chinner wrote:
>
>
> > I did apply them manually - but the patchset is based on the
> > tip/master, plus some patches from a previous patch set that aren't
> > in tip/locking to be able
On Wed, Mar 12, 2014 at 07:24:36AM +0100, Peter Zijlstra wrote:
> On Tue, Mar 11, 2014 at 11:17:46PM -0400, Waiman Long wrote:
> > Except that I do a single atomic short integer write to switch the bits
> > instead of 2 byte write.
>
> D'0h why didn't I think of that. A single short write is much
On Wed, 12 Mar 2014 15:26:16 +1100
Dave Chinner wrote:
> I did apply them manually - but the patchset is based on the
> tip/master, plus some patches from a previous patch set that aren't
> in tip/locking to be able to apply this patchset cleanly in the
> first place. So there's no "start froma
On Tue, Mar 11, 2014 at 11:17:46PM -0400, Waiman Long wrote:
> Except that I do a single atomic short integer write to switch the bits
> instead of 2 byte write.
D'0h why didn't I think of that. A single short write is much better
than the 2 byte writes.
In any case; can you try and keep the
On Wed, Mar 12, 2014 at 01:31:53PM +1100, Dave Chinner wrote:
> With the queuing spinlock, I expected to see somewhat better
> results, but I didn't at first. Turns out if you have any sort of
> lock debugging turned on, then the code doesn't ever go into the
> lock slow path and hence does not
On Wed, Mar 12, 2014 at 07:24:36AM +0100, Peter Zijlstra wrote:
On Tue, Mar 11, 2014 at 11:17:46PM -0400, Waiman Long wrote:
Except that I do a single atomic short integer write to switch the bits
instead of 2 byte write.
D'0h why didn't I think of that. A single short write is much better
On Wed, Mar 12, 2014 at 06:07:30AM -0400, Steven Rostedt wrote:
On Wed, 12 Mar 2014 15:26:16 +1100
Dave Chinner da...@fromorbit.com wrote:
I did apply them manually - but the patchset is based on the
tip/master, plus some patches from a previous patch set that aren't
in tip/locking to
On Wed, Mar 12, 2014 at 8:57 AM, Peter Zijlstra pet...@infradead.org wrote:
I've not managed to reproduce the hang as reported by Dave, but then I
doubt building kernels has anywhere near the lock contention he
generated.
Building kernels has basically no lock contention at all.
On Wed, 12 Mar 2014 16:57:54 +0100
Peter Zijlstra pet...@infradead.org wrote:
I build and booted these patches against tip/master as of this morning.
It build ~50 defconfig kernels on my WSM-EP.
I've not managed to reproduce the hang as reported by Dave, but then I
doubt building kernels
On Wed, Mar 12, 2014 at 12:19:33PM -0400, Steven Rostedt wrote:
On Wed, 12 Mar 2014 16:57:54 +0100
Peter Zijlstra pet...@infradead.org wrote:
I build and booted these patches against tip/master as of this morning.
It build ~50 defconfig kernels on my WSM-EP.
I've not managed to
On 03/12/2014 02:24 AM, Peter Zijlstra wrote:
On Tue, Mar 11, 2014 at 11:17:46PM -0400, Waiman Long wrote:
Except that I do a single atomic short integer write to switch the bits
instead of 2 byte write.
D'0h why didn't I think of that. A single short write is much better
than the 2 byte
On Wed, Mar 12, 2014 at 07:15:03AM +0100, Peter Zijlstra wrote:
On Wed, Mar 12, 2014 at 01:31:53PM +1100, Dave Chinner wrote:
With the queuing spinlock, I expected to see somewhat better
results, but I didn't at first. Turns out if you have any sort of
lock debugging turned on, then the
On Wed, Mar 12, 2014 at 01:31:53PM +1100, Dave Chinner wrote:
With the queuing spinlock, I expected to see somewhat better
results, but I didn't at first. Turns out if you have any sort of
lock debugging turned on, then the code doesn't ever go into the
lock slow path and hence does not ever
On Tue, Mar 11, 2014 at 11:17:46PM -0400, Waiman Long wrote:
Except that I do a single atomic short integer write to switch the bits
instead of 2 byte write.
D'0h why didn't I think of that. A single short write is much better
than the 2 byte writes.
In any case; can you try and keep the
On Wed, 12 Mar 2014 15:26:16 +1100
Dave Chinner da...@fromorbit.com wrote:
I did apply them manually - but the patchset is based on the
tip/master, plus some patches from a previous patch set that aren't
in tip/locking to be able to apply this patchset cleanly in the
first place. So there's
On Tue, Mar 11, 2014 at 11:11:59PM -0400, Steven Rostedt wrote:
> On Wed, 12 Mar 2014 13:31:53 +1100
> Dave Chinner wrote:
>
>
> > So either this patchset doesn't work right now, or there's something
> > else broken in the tip/master tree...
>
> I've found a few things broken with the tip
On 03/11/2014 06:45 AM, Ingo Molnar wrote:
* Peter Zijlstra wrote:
Hi Waiman,
I promised you this series a number of days ago; sorry for the delay
I've been somewhat unwell :/
That said, these few patches start with a (hopefully) simple and
correct form of the queue spinlock, and then
On Wed, 12 Mar 2014 13:31:53 +1100
Dave Chinner wrote:
> So either this patchset doesn't work right now, or there's something
> else broken in the tip/master tree...
I've found a few things broken with the tip master tree. Anyway, I'm
not sure the patches are even there. The last I saw, the
On Mon, Mar 10, 2014 at 04:42:36PM +0100, Peter Zijlstra wrote:
> Hi Waiman,
>
> I promised you this series a number of days ago; sorry for the delay I've been
> somewhat unwell :/
>
> That said, these few patches start with a (hopefully) simple and correct form
> of the queue spinlock, and then
* Peter Zijlstra wrote:
> On Tue, Mar 11, 2014 at 11:45:03AM +0100, Ingo Molnar wrote:
> >
> > * Peter Zijlstra wrote:
> >
> > > Hi Waiman,
> > >
> > > I promised you this series a number of days ago; sorry for the delay
> > > I've been somewhat unwell :/
> > >
> > > That said, these few
On Tue, Mar 11, 2014 at 11:45:03AM +0100, Ingo Molnar wrote:
>
> * Peter Zijlstra wrote:
>
> > Hi Waiman,
> >
> > I promised you this series a number of days ago; sorry for the delay
> > I've been somewhat unwell :/
> >
> > That said, these few patches start with a (hopefully) simple and
>
* Peter Zijlstra wrote:
> Hi Waiman,
>
> I promised you this series a number of days ago; sorry for the delay
> I've been somewhat unwell :/
>
> That said, these few patches start with a (hopefully) simple and
> correct form of the queue spinlock, and then gradually build upon
> it,
* Peter Zijlstra pet...@infradead.org wrote:
Hi Waiman,
I promised you this series a number of days ago; sorry for the delay
I've been somewhat unwell :/
That said, these few patches start with a (hopefully) simple and
correct form of the queue spinlock, and then gradually build upon
On Tue, Mar 11, 2014 at 11:45:03AM +0100, Ingo Molnar wrote:
* Peter Zijlstra pet...@infradead.org wrote:
Hi Waiman,
I promised you this series a number of days ago; sorry for the delay
I've been somewhat unwell :/
That said, these few patches start with a (hopefully) simple
* Peter Zijlstra pet...@infradead.org wrote:
On Tue, Mar 11, 2014 at 11:45:03AM +0100, Ingo Molnar wrote:
* Peter Zijlstra pet...@infradead.org wrote:
Hi Waiman,
I promised you this series a number of days ago; sorry for the delay
I've been somewhat unwell :/
That
On Mon, Mar 10, 2014 at 04:42:36PM +0100, Peter Zijlstra wrote:
Hi Waiman,
I promised you this series a number of days ago; sorry for the delay I've been
somewhat unwell :/
That said, these few patches start with a (hopefully) simple and correct form
of the queue spinlock, and then
On Wed, 12 Mar 2014 13:31:53 +1100
Dave Chinner da...@fromorbit.com wrote:
So either this patchset doesn't work right now, or there's something
else broken in the tip/master tree...
I've found a few things broken with the tip master tree. Anyway, I'm
not sure the patches are even there. The
On 03/11/2014 06:45 AM, Ingo Molnar wrote:
* Peter Zijlstrapet...@infradead.org wrote:
Hi Waiman,
I promised you this series a number of days ago; sorry for the delay
I've been somewhat unwell :/
That said, these few patches start with a (hopefully) simple and
correct form of the queue
On Tue, Mar 11, 2014 at 11:11:59PM -0400, Steven Rostedt wrote:
On Wed, 12 Mar 2014 13:31:53 +1100
Dave Chinner da...@fromorbit.com wrote:
So either this patchset doesn't work right now, or there's something
else broken in the tip/master tree...
I've found a few things broken with the
Hi Waiman,
I promised you this series a number of days ago; sorry for the delay I've been
somewhat unwell :/
That said, these few patches start with a (hopefully) simple and correct form
of the queue spinlock, and then gradually build upon it, explaining each
optimization as we go.
Having these
Hi Waiman,
I promised you this series a number of days ago; sorry for the delay I've been
somewhat unwell :/
That said, these few patches start with a (hopefully) simple and correct form
of the queue spinlock, and then gradually build upon it, explaining each
optimization as we go.
Having these
36 matches
Mail list logo
