Currently the PowerTOP tool is x86 centric and does not work on other architectures, unless some changes are made.
This patchset removes the hardcoded values for the P and C states. It does that in two steps 1) Implements the OMAP3 support in PowerTOP. 2) Makes the code generic enough for all boards
I have tested this only on my desktop. I am yet to receive an ARM based board, and hence couldn't test it there. So, any kind of testing and/or feedback are most welcome! Thanks!
Amit Arora (2): Add OMAP3 support to PowerTOP Remove hardcoded c and p states
Makefile | 2 +- cpufreqstats.c | 3 +- display.c | 23 +++++++++----- generic_cstates.c | 63 ++++++++++++++++++++++++++++++++++++++++ powertop.c | 82 +++++++++++++++++++++++++++++++++++++++++++++++++---- powertop.h | 11 ++++--- 6 files changed, 162 insertions(+), 22 deletions(-) create mode 100644 generic_cstates.c
-- Regards, Amit Arora
On 10 Jul 29, Amit Arora wrote:
Currently the PowerTOP tool is x86 centric and does not work on other architectures, unless some changes are made.
This patchset removes the hardcoded values for the P and C states. It does that in two steps
- Implements the OMAP3 support in PowerTOP.
- Makes the code generic enough for all boards
I have tested this only on my desktop. I am yet to receive an ARM based board, and hence couldn't test it there. So, any kind of testing and/or feedback are most welcome! Thanks!
Amit Arora (2): Add OMAP3 support to PowerTOP Remove hardcoded c and p states
Makefile | 2 +- cpufreqstats.c | 3 +- display.c | 23 +++++++++----- generic_cstates.c | 63 ++++++++++++++++++++++++++++++++++++++++ powertop.c | 82 +++++++++++++++++++++++++++++++++++++++++++++++++---- powertop.h | 11 ++++--- 6 files changed, 162 insertions(+), 22 deletions(-) create mode 100644 generic_cstates.c
I understand that it is logical to keep the forward-ported TI patch separated from the generic code you wrote in the beginning.
But I think we should merge the patches since the bulk of the changes in the 2nd patch is just the renaming of the file.
More comments in other email.
/Amit
On Thu, Jul 29, 2010 at 5:42 PM, Amit Kucheria amit.kucheria@linaro.org wrote:
On 10 Jul 29, Amit Arora wrote:
Amit Arora (2): Add OMAP3 support to PowerTOP Remove hardcoded c and p states
Makefile | 2 +- cpufreqstats.c | 3 +- display.c | 23 +++++++++----- generic_cstates.c | 63 ++++++++++++++++++++++++++++++++++++++++ powertop.c | 82 +++++++++++++++++++++++++++++++++++++++++++++++++---- powertop.h | 11 ++++--- 6 files changed, 162 insertions(+), 22 deletions(-) create mode 100644 generic_cstates.c
I understand that it is logical to keep the forward-ported TI patch separated from the generic code you wrote in the beginning.
But I think we should merge the patches since the bulk of the changes in the 2nd patch is just the renaming of the file.
Ok. Agreed.
Thanks! -- Regards, Amit Arora
More comments in other email.
/Amit
On 10 Jul 29, Amit Arora wrote:
Currently the PowerTOP tool is x86 centric and does not work on other architectures, unless some changes are made.
This patchset removes the hardcoded values for the P and C states. It does that in two steps
- Implements the OMAP3 support in PowerTOP.
- Makes the code generic enough for all boards
I have tested this only on my desktop. I am yet to receive an ARM based board, and hence couldn't test it there. So, any kind of testing and/or feedback are most welcome! Thanks!
Amit Arora (2): Add OMAP3 support to PowerTOP Remove hardcoded c and p states
Makefile | 2 +- cpufreqstats.c | 3 +- display.c | 23 +++++++++----- generic_cstates.c | 63 ++++++++++++++++++++++++++++++++++++++++ powertop.c | 82 +++++++++++++++++++++++++++++++++++++++++++++++++---- powertop.h | 11 ++++--- 6 files changed, 162 insertions(+), 22 deletions(-) create mode 100644 generic_cstates.c
With these patches applied, I lose the entire bottom section of "Top causes for wakeups". Reverting the patches brings them back. I'm testing this on my PC yet.
/Amit
On Thu, Jul 29, 2010 at 6:38 PM, Amit Kucheria amit.kucheria@linaro.org wrote:
On 10 Jul 29, Amit Arora wrote:
Currently the PowerTOP tool is x86 centric and does not work on other architectures, unless some changes are made.
This patchset removes the hardcoded values for the P and C states. It does that in two steps
- Implements the OMAP3 support in PowerTOP.
- Makes the code generic enough for all boards
I have tested this only on my desktop. I am yet to receive an ARM based board, and hence couldn't test it there. So, any kind of testing and/or feedback are most welcome! Thanks!
Amit Arora (2): Add OMAP3 support to PowerTOP Remove hardcoded c and p states
Makefile | 2 +- cpufreqstats.c | 3 +- display.c | 23 +++++++++----- generic_cstates.c | 63 ++++++++++++++++++++++++++++++++++++++++ powertop.c | 82 +++++++++++++++++++++++++++++++++++++++++++++++++---- powertop.h | 11 ++++--- 6 files changed, 162 insertions(+), 22 deletions(-) create mode 100644 generic_cstates.c
With these patches applied, I lose the entire bottom section of "Top causes for wakeups". Reverting the patches brings them back. I'm testing this on my PC yet.
I tested it on my dual core desktop along with my laptop. Strange that I don't see this issue.
Can you please maximize the size of the window and try again ? This might give us some hint.
Thanks! -- Regards, Amit Arora
/Amit
Dnia czwartek, 29 lipca 2010 o 13:57:21 Amit Arora napisał(a):
I have tested this only on my desktop. I am yet to receive an ARM based board, and hence couldn't test it there. So, any kind of testing and/or feedback are most welcome! Thanks!
I fetched 9f4efd1c7cc0bc9a9078fa6b83a98763a7637e2f revision, applied both patches and built it for my desktop. Segfaulted on first run.
Desktop is Core2Quad q6600, powertop without your changes works. C-states are not available, frequencies are listed.
Regards,
On Thu, Jul 29, 2010 at 7:56 PM, Marcin Juszkiewicz marcin@juszkiewicz.com.pl wrote:
Dnia czwartek, 29 lipca 2010 o 13:57:21 Amit Arora napisał(a):
I have tested this only on my desktop. I am yet to receive an ARM based board, and hence couldn't test it there. So, any kind of testing and/or feedback are most welcome! Thanks!
I fetched 9f4efd1c7cc0bc9a9078fa6b83a98763a7637e2f revision, applied both patches and built it for my desktop. Segfaulted on first run.
Desktop is Core2Quad q6600, powertop without your changes works. C-states are not available, frequencies are listed.
Thanks for testing!
Is it possible for you to take a tar dump of the sysfs cpu directory (tar jcf sysfs-cpu.tar.bz2 /sys/devices/system/cpu) and send it to me ? It will help me debug the issue.
Thanks again! -- Regards, Amit Arora
Regards,
JID: hrw@jabber.org Website: http://marcin.juszkiewicz.com.pl/ LinkedIn: http://www.linkedin.com/in/marcinjuszkiewicz
linaro-dev mailing list linaro-dev@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-dev
-
Amit Arora -
Amit Kucheria -
Marcin Juszkiewicz