This is a rework of the previously posted set of patches.
Patch 1 is the user level patch to add the SPU events to the user
OProfile tool.
Patch 2 is a kernel patch to do code clean up and restructuring to make
it easier to add the new SPU event profiling support. This patch makes
no functional changes.
Patch 3 is a kernel patch to add the SPU event profiling support.
Carl Love
On 01.12.08 16:18:26, Carl Love wrote:
> This is a rework of the previously posted set of patches.
>
> Patch 1 is the user level patch to add the SPU events to the user
> OProfile tool.
>
> Patch 2 is a kernel patch to do code clean up and restructuring to make
> it easier to add the new SPU event profiling support. This patch makes
> no functional changes.
>
> Patch 3 is a kernel patch to add the SPU event profiling support.
I applied patch 2 & 3 to:
git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile.git cell
The patches did not apply cleanly. I had to change the path to
arch/powerpc/include/asm/, fix cell/pr_util.h and did some whitespace
cleanups. Please run your tests on the cell branch.
Thanks,
-Robert
--
Advanced Micro Devices, Inc.
Operating System Research Center
email: [email protected]
On Thu, 2009-01-08 at 16:48 +0100, Robert Richter wrote:
> On 01.12.08 16:18:26, Carl Love wrote:
> > This is a rework of the previously posted set of patches.
> >
> > Patch 1 is the user level patch to add the SPU events to the user
> > OProfile tool.
> >
> > Patch 2 is a kernel patch to do code clean up and restructuring to make
> > it easier to add the new SPU event profiling support. This patch makes
> > no functional changes.
> >
> > Patch 3 is a kernel patch to add the SPU event profiling support.
>
> I applied patch 2 & 3 to:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile.git cell
>
> The patches did not apply cleanly. I had to change the path to
> arch/powerpc/include/asm/, fix cell/pr_util.h and did some whitespace
> cleanups. Please run your tests on the cell branch.
>
> Thanks,
>
> -Robert
>
I pulled down the git tree, compiled and installed it. I tested it
against the OProfile testsuite, which includes SPU event profiling
tests. Everything passed. The patch I submitted was against a 2.6.26
tree. You are now on a 2.6.28 tree so perhaps that is why the patch did
not apply cleanly. The patch has been out there for some time.
Carl Love
On 08.01.09 16:26:31, Carl Love wrote:
> > git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile.git cell
> >
> > The patches did not apply cleanly. I had to change the path to
> > arch/powerpc/include/asm/, fix cell/pr_util.h and did some whitespace
> > cleanups. Please run your tests on the cell branch.
Thanks Carl, could you also test the 'next' branch on your
architecture, please? This is the latest version of patches to go
upstream (containing your patches also).
Thanks,
-Robert
--
Advanced Micro Devices, Inc.
Operating System Research Center
email: [email protected]
On Thu, 2009-01-08 at 16:26 -0800, Carl Love wrote:
> I pulled down the git tree, compiled and installed it. I tested it
> against the OProfile testsuite, which includes SPU event profiling
> tests. Everything passed. The patch I submitted was against a 2.6.26
> tree. You are now on a 2.6.28 tree so perhaps that is why the patch did
> not apply cleanly. The patch has been out there for some time.
When you submited it on Dec 2, it should have been against whatever was
the latest upstream at the time, not 2.6.26.
Cheers,
Ben.
On Sun, 2009-01-11 at 10:31 +1100, Benjamin Herrenschmidt wrote:
> On Thu, 2009-01-08 at 16:26 -0800, Carl Love wrote:
> > I pulled down the git tree, compiled and installed it. I tested it
> > against the OProfile testsuite, which includes SPU event profiling
> > tests. Everything passed. The patch I submitted was against a 2.6.26
> > tree. You are now on a 2.6.28 tree so perhaps that is why the patch did
> > not apply cleanly. The patch has been out there for some time.
>
> When you submited it on Dec 2, it should have been against whatever was
> the latest upstream at the time, not 2.6.26.
>
> Cheers,
> Ben.
Ben, Arnd and Robert:
The patch was against the latest Arnd Cell Kernel tree at the time it
was posted. The OProfile for cell patches have all been submitted,
accepted by Arnd (CELL Kernel maintainer) and pushed up stream by Arnd.
Arnd has waited for the OProfile maintainer to review and approve the
change before Arnd would push it upstream.
So, at this point, Robert's (OProfile kernel maintainer) and Maynard's
(OProfile user space maintainer) approval for the patches.
The question now goes to Arnd and Robert, who is going to push the
patches upstream? Looks like Robert is ready to do it, so Arnd do you
approve the patches? Would you like to have Robert push the patches
upstream or would you prefer to do it? I think we need an
answer/agreement on this.
Carl Love
On 12.01.09 08:15:27, Carl Love wrote:
>
> On Sun, 2009-01-11 at 10:31 +1100, Benjamin Herrenschmidt wrote:
> > On Thu, 2009-01-08 at 16:26 -0800, Carl Love wrote:
> > > I pulled down the git tree, compiled and installed it. I tested it
> > > against the OProfile testsuite, which includes SPU event profiling
> > > tests. Everything passed. The patch I submitted was against a 2.6.26
> > > tree. You are now on a 2.6.28 tree so perhaps that is why the patch did
> > > not apply cleanly. The patch has been out there for some time.
> >
> > When you submited it on Dec 2, it should have been against whatever was
> > the latest upstream at the time, not 2.6.26.
> >
> > Cheers,
> > Ben.
>
> Ben, Arnd and Robert:
>
> The patch was against the latest Arnd Cell Kernel tree at the time it
> was posted. The OProfile for cell patches have all been submitted,
> accepted by Arnd (CELL Kernel maintainer) and pushed up stream by Arnd.
> Arnd has waited for the OProfile maintainer to review and approve the
> change before Arnd would push it upstream.
The patches are already upstream in v2.6.29-rc1. Hope this is ok. I
had added them to the oprofile tree and the tree was merged be Linus.
-Robert
> So, at this point, Robert's (OProfile kernel maintainer) and Maynard's
> (OProfile user space maintainer) approval for the patches.
>
> The question now goes to Arnd and Robert, who is going to push the
> patches upstream? Looks like Robert is ready to do it, so Arnd do you
> approve the patches? Would you like to have Robert push the patches
> upstream or would you prefer to do it? I think we need an
> answer/agreement on this.
Arnd, I thought the patches were ok for you. If there are still some
concerns, we have to make delta patches.
See also here:
git log -4 -p 25006644e6042aab4bb7cdc4bfc5777cd3141df7
-Robert
--
Advanced Micro Devices, Inc.
Operating System Research Center
email: [email protected]
On Monday 12 January 2009, Robert Richter wrote:
> Arnd, I thought the patches were ok for you. If there are still some
> concerns, we have to make delta patches.
Yes, that's fine with me, thanks for taking care of the patches!
Arnd <><
On Mon, 2009-01-12 at 18:59 +0100, Arnd Bergmann wrote:
> On Monday 12 January 2009, Robert Richter wrote:
> > Arnd, I thought the patches were ok for you. If there are still some
> > concerns, we have to make delta patches.
>
> Yes, that's fine with me, thanks for taking care of the patches!
>
> Arnd <><
Thanks guys for all your help on the patches.
Carl Love