On Fri, Nov 03, 2006 at 03:41:32AM +0100, Adrian Bunk wrote:
> This email lists some known regressions in 2.6.19-rc4 compared to 2.6.18
> that are not yet fixed in Linus' tree.
>
> If you find your name in the Cc header, you are either submitter of one
> of the bugs, maintainer of an affectected subsystem or driver, a patch
> of you caused a breakage or I'm considering you in any other way possibly
> involved with one or more of these issues.
>
> Due to the huge amount of recipients, please trim the Cc when answering.
>
> Subject : cpufreq not working on AMD K8
> References : http://lkml.org/lkml/2006/10/10/114
> Submitter : Christian <[email protected]>
> Status : unknown
As Mark mentioned in his followup, powernow-k8 didn't change in .19 at all.
I'm suspecting an ACPI change meant that we no longer find the PST tables
correctly.
Christian, can you post the full dmesg's from the working/broken kernels.
It may be useful to enable CONFIG_ACPI_DEBUG too.
Dave
--
http://www.codemonkey.org.uk
Could this be a problem?
--------------------
...
CONFIG_ACPI_PROCESSOR=m
...
CONFIG_X86_POWERNOW_K8=y
...
Regards,
Alex.
Dave Jones wrote:
> On Fri, Nov 03, 2006 at 03:41:32AM +0100, Adrian Bunk wrote:
> > This email lists some known regressions in 2.6.19-rc4 compared to 2.6.18
> > that are not yet fixed in Linus' tree.
> >
> > If you find your name in the Cc header, you are either submitter of one
> > of the bugs, maintainer of an affectected subsystem or driver, a patch
> > of you caused a breakage or I'm considering you in any other way possibly
> > involved with one or more of these issues.
> >
> > Due to the huge amount of recipients, please trim the Cc when answering.
> >
> > Subject : cpufreq not working on AMD K8
> > References : http://lkml.org/lkml/2006/10/10/114
> > Submitter : Christian <[email protected]>
> > Status : unknown
>
> As Mark mentioned in his followup, powernow-k8 didn't change in .19 at all.
> I'm suspecting an ACPI change meant that we no longer find the PST tables
> correctly.
>
> Christian, can you post the full dmesg's from the working/broken kernels.
> It may be useful to enable CONFIG_ACPI_DEBUG too.
>
> Dave
>
On Fri, Nov 03, 2006 at 11:25:37AM +0300, Alexey Starikovskiy wrote:
> Could this be a problem?
> --------------------
> ...
> CONFIG_ACPI_PROCESSOR=m
> ...
> CONFIG_X86_POWERNOW_K8=y
Hmm, possibly. Christian, does it work again if you set them both to =y ?
Dave
--
http://www.codemonkey.org.uk
Am Freitag, 3. November 2006 16:56 schrieb Dave Jones:
> On Fri, Nov 03, 2006 at 11:25:37AM +0300, Alexey Starikovskiy wrote:
> > Could this be a problem?
> > --------------------
> > ...
> > CONFIG_ACPI_PROCESSOR=m
> > ...
> > CONFIG_X86_POWERNOW_K8=y
>
> Hmm, possibly. Christian, does it work again if you set them both to =y ?
Yes, it works now! Only the change to CONFIG_ACPI_PROCESSOR=y made it work
again!
Nice catch ;-)
Thank you very much!
-Christian
On Sun, Nov 05, 2006 at 06:32:12PM +0100, Christian wrote:
> Am Freitag, 3. November 2006 16:56 schrieb Dave Jones:
> > On Fri, Nov 03, 2006 at 11:25:37AM +0300, Alexey Starikovskiy wrote:
> > > Could this be a problem?
> > > --------------------
> > > ...
> > > CONFIG_ACPI_PROCESSOR=m
> > > ...
> > > CONFIG_X86_POWERNOW_K8=y
> >
> > Hmm, possibly. Christian, does it work again if you set them both to =y ?
>
> Yes, it works now! Only the change to CONFIG_ACPI_PROCESSOR=y made it work
> again!
So, the reasoning behind this, is that we have this construct..
config X86_POWERNOW_K8_ACPI
bool
depends on X86_POWERNOW_K8 && ACPI_PROCESSOR
depends on !(X86_POWERNOW_K8 = y && ACPI_PROCESSOR = m)
default y
Which makes us use the ACPI stuff if it's there, otherwise not,
and in your case, it seems your system _needs_ this enabled
to make powernow work.
Thing is, this was there in 2.6.18 too, so strictly speaking,
we haven't regressed here, and you're getting exactly what you asked for.
The problem is that it's completely silent as to why it then fails.
I'm open to improvements, but I'm not sure what the right thing to do
here is.. opinions ?
Dave
--
http://www.codemonkey.org.uk
On Sun, Nov 05, 2006 at 06:32:12PM +0100, Christian wrote:
> Am Freitag, 3. November 2006 16:56 schrieb Dave Jones:
> > On Fri, Nov 03, 2006 at 11:25:37AM +0300, Alexey Starikovskiy wrote:
> > > Could this be a problem?
> > > --------------------
> > > ...
> > > CONFIG_ACPI_PROCESSOR=m
> > > ...
> > > CONFIG_X86_POWERNOW_K8=y
> >
> > Hmm, possibly. Christian, does it work again if you set them both to =y ?
>
> Yes, it works now! Only the change to CONFIG_ACPI_PROCESSOR=y made it work
> again!
You said 2.6.18 worked for you.
Did you have CONFIG_ACPI_PROCESSOR=y in 2.6.18, or did
CONFIG_ACPI_PROCESSOR=m, CONFIG_X86_POWERNOW_K8=y work for you in 2.6.18?
> Nice catch ;-)
>
> Thank you very much!
> -Christian
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
Am Montag, 6. November 2006 07:00 schrieb Adrian Bunk:
> On Sun, Nov 05, 2006 at 06:32:12PM +0100, Christian wrote:
> > Am Freitag, 3. November 2006 16:56 schrieb Dave Jones:
> > > On Fri, Nov 03, 2006 at 11:25:37AM +0300, Alexey Starikovskiy wrote:
> > > > Could this be a problem?
> > > > --------------------
> > > > ...
> > > > CONFIG_ACPI_PROCESSOR=m
> > > > ...
> > > > CONFIG_X86_POWERNOW_K8=y
> > >
> > > Hmm, possibly. Christian, does it work again if you set them both to
> > > =y ?
> >
> > Yes, it works now! Only the change to CONFIG_ACPI_PROCESSOR=y made it
> > work again!
>
> You said 2.6.18 worked for you.
>
> Did you have CONFIG_ACPI_PROCESSOR=y in 2.6.18, or did
> CONFIG_ACPI_PROCESSOR=m, CONFIG_X86_POWERNOW_K8=y work for you in 2.6.18?
It worked with CONFIG_ACPI_PROCESSOR=m in 2.6.18-rc7. Since 2.6.19-rc1 it
doesn't work anymore with CONFIG_ACPI_PROCESSOR=m.
user@ubuntu:~/Projekte/linux-2.6.18-rc7$ uname -a
Linux ubuntu.localnet 2.6.18-rc7 #2 SMP Wed Sep 13 11:28:41 CEST 2006 x86_64
GNU/Linux
user@ubuntu:~/Projekte/linux-2.6.18-rc7$ lsmod | grep -Ei "processor|acpi|
power"
powernow_k8 16096 1
freq_table 6848 2 powernow_k8,cpufreq_stats
cpufreq_powersave 3584 0
asus_acpi 20644 0
processor 36872 2 powernow_k8,thermal
user@ubuntu:~/Projekte/linux-2.6.18-rc7$ grep -i
ACPI_PROCESSOR /boot/config-2.6.18-rc7
CONFIG_ACPI_PROCESSOR=m
user@ubuntu:~/Projekte/linux-2.6.18-rc7$
grep -Ei "POWERNOW_K8" /boot/config-2.6.18-rc7
CONFIG_X86_POWERNOW_K8=m
CONFIG_X86_POWERNOW_K8_ACPI=y
+++ There's a difference in 2.6.19! CONFIG_X86_POWERNOW_K8_ACPI is gone +++
user@ubuntu:~/Projekte/linux-2.6.18-rc7$
grep -Ei "POWERNOW_K8" /boot/config-2.6.19-rc1
CONFIG_X86_POWERNOW_K8=y
user@ubuntu:~/Projekte/linux-2.6.18-rc7$ grep -Ei "CPUFREQ|
CPU_FREQ" /boot/config-2.6.18-rc7
CONFIG_CPU_FREQ=y
CONFIG_CPU_FREQ_TABLE=m
# CONFIG_CPU_FREQ_DEBUG is not set
CONFIG_CPU_FREQ_STAT=m
CONFIG_CPU_FREQ_STAT_DETAILS=y
CONFIG_CPU_FREQ_DEFAULT_GOV_PERFORMANCE=y
# CONFIG_CPU_FREQ_DEFAULT_GOV_USERSPACE is not set
CONFIG_CPU_FREQ_GOV_PERFORMANCE=y
CONFIG_CPU_FREQ_GOV_POWERSAVE=m
CONFIG_CPU_FREQ_GOV_USERSPACE=m
CONFIG_CPU_FREQ_GOV_ONDEMAND=m
CONFIG_CPU_FREQ_GOV_CONSERVATIVE=m
# CPUFreq processor drivers
CONFIG_X86_ACPI_CPUFREQ=m
# CONFIG_X86_ACPI_CPUFREQ_PROC_INTF is not set
-Christian
On Mon, Nov 06, 2006 at 04:43:13PM +0100, Christian wrote:
> > Did you have CONFIG_ACPI_PROCESSOR=y in 2.6.18, or did
> > CONFIG_ACPI_PROCESSOR=m, CONFIG_X86_POWERNOW_K8=y work for you in 2.6.18?
>
> It worked with CONFIG_ACPI_PROCESSOR=m in 2.6.18-rc7. Since 2.6.19-rc1 it
> doesn't work anymore with CONFIG_ACPI_PROCESSOR=m.
>
> user@ubuntu:~/Projekte/linux-2.6.18-rc7$ grep -i ACPI_PROCESSOR /boot/config-2.6.18-rc7
> CONFIG_ACPI_PROCESSOR=m
>
> user@ubuntu:~/Projekte/linux-2.6.18-rc7$
> grep -Ei "POWERNOW_K8" /boot/config-2.6.18-rc7
> CONFIG_X86_POWERNOW_K8=m
> CONFIG_X86_POWERNOW_K8_ACPI=y
>
> +++ There's a difference in 2.6.19! CONFIG_X86_POWERNOW_K8_ACPI is gone +++
I don't understand how this was allowed. Because when I try this
with a 2.6.18 tree.. (nothing changed between -rc7 and final for cpufreq)
<editted a .config to match your config>
$ grep ACPI_PROCESSOR .config
CONFIG_ACPI_PROCESSOR=m
$ grep POWERNOW_K8 .config
CONFIG_X86_POWERNOW_K8=y
CONFIG_X86_POWERNOW_K8_ACPI=y
and then after a make oldconfig the CONFIG_X86_POWERNOW_K8_ACPI is removed
as it isn't valid.
Did you edit your .config by hand ?
Dave
--
http://www.codemonkey.org.uk
On Mon, Nov 06, 2006 at 12:20:49PM -0500, Dave Jones wrote:
> On Mon, Nov 06, 2006 at 04:43:13PM +0100, Christian wrote:
>
> > > Did you have CONFIG_ACPI_PROCESSOR=y in 2.6.18, or did
> > > CONFIG_ACPI_PROCESSOR=m, CONFIG_X86_POWERNOW_K8=y work for you in 2.6.18?
> >
> > It worked with CONFIG_ACPI_PROCESSOR=m in 2.6.18-rc7. Since 2.6.19-rc1 it
> > doesn't work anymore with CONFIG_ACPI_PROCESSOR=m.
> >
> > user@ubuntu:~/Projekte/linux-2.6.18-rc7$ grep -i ACPI_PROCESSOR /boot/config-2.6.18-rc7
> > CONFIG_ACPI_PROCESSOR=m
> >
> > user@ubuntu:~/Projekte/linux-2.6.18-rc7$
> > grep -Ei "POWERNOW_K8" /boot/config-2.6.18-rc7
> > CONFIG_X86_POWERNOW_K8=m
> > CONFIG_X86_POWERNOW_K8_ACPI=y
> >
> > +++ There's a difference in 2.6.19! CONFIG_X86_POWERNOW_K8_ACPI is gone +++
>
> I don't understand how this was allowed. Because when I try this
> with a 2.6.18 tree.. (nothing changed between -rc7 and final for cpufreq)
>
> <editted a .config to match your config>
>
> $ grep ACPI_PROCESSOR .config
> CONFIG_ACPI_PROCESSOR=m
> $ grep POWERNOW_K8 .config
> CONFIG_X86_POWERNOW_K8=y
> CONFIG_X86_POWERNOW_K8_ACPI=y
>
> and then after a make oldconfig the CONFIG_X86_POWERNOW_K8_ACPI is removed
> as it isn't valid.
>
> Did you edit your .config by hand ?
Look closer, his linux-2.6.18-rc7 .config contains
CONFIG_ACPI_PROCESSOR=m, CONFIG_X86_POWERNOW_K8=m.
> Dave
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
On Sun, Nov 05, 2006 at 03:04:48PM -0500, Dave Jones wrote:
> On Sun, Nov 05, 2006 at 06:32:12PM +0100, Christian wrote:
> > Am Freitag, 3. November 2006 16:56 schrieb Dave Jones:
> > > On Fri, Nov 03, 2006 at 11:25:37AM +0300, Alexey Starikovskiy wrote:
> > > > Could this be a problem?
> > > > --------------------
> > > > ...
> > > > CONFIG_ACPI_PROCESSOR=m
> > > > ...
> > > > CONFIG_X86_POWERNOW_K8=y
> > >
> > > Hmm, possibly. Christian, does it work again if you set them both to =y ?
> >
> > Yes, it works now! Only the change to CONFIG_ACPI_PROCESSOR=y made it work
> > again!
>
> So, the reasoning behind this, is that we have this construct..
>
> config X86_POWERNOW_K8_ACPI
> bool
> depends on X86_POWERNOW_K8 && ACPI_PROCESSOR
> depends on !(X86_POWERNOW_K8 = y && ACPI_PROCESSOR = m)
> default y
>
>
> Which makes us use the ACPI stuff if it's there, otherwise not,
> and in your case, it seems your system _needs_ this enabled
> to make powernow work.
>
> Thing is, this was there in 2.6.18 too, so strictly speaking,
> we haven't regressed here, and you're getting exactly what you asked for.
> The problem is that it's completely silent as to why it then fails.
>
> I'm open to improvements, but I'm not sure what the right thing to do
> here is.. opinions ?
The extreme solution would be
config X86_POWERNOW_K8
tristate "AMD Opteron/Athlon64 PowerNow!"
select CPU_FREQ_TABLE
depends ACPI_PROCESSOR
A medium solution might be
config X86_POWERNOW_K8
tristate "AMD Opteron/Athlon64 PowerNow!"
select CPU_FREQ_TABLE
depends (ACPI_PROCESSOR || ACPI_PROCESSOR=n)
But in the end, the best solution depends on how many percent of the
X86_POWERNOW_K8 users have Christian's problem of requiring
ACPI_PROCESSOR. If there are only very few people with this problem, I'd
say leave it as it is.
> Dave
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
On Mon, Nov 06, 2006 at 04:43:13PM +0100, Christian wrote:
> Am Montag, 6. November 2006 07:00 schrieb Adrian Bunk:
> > On Sun, Nov 05, 2006 at 06:32:12PM +0100, Christian wrote:
> > > Am Freitag, 3. November 2006 16:56 schrieb Dave Jones:
> > > > On Fri, Nov 03, 2006 at 11:25:37AM +0300, Alexey Starikovskiy wrote:
> > > > > Could this be a problem?
> > > > > --------------------
> > > > > ...
> > > > > CONFIG_ACPI_PROCESSOR=m
> > > > > ...
> > > > > CONFIG_X86_POWERNOW_K8=y
> > > >
> > > > Hmm, possibly. Christian, does it work again if you set them both to
> > > > =y ?
> > >
> > > Yes, it works now! Only the change to CONFIG_ACPI_PROCESSOR=y made it
> > > work again!
> >
> > You said 2.6.18 worked for you.
> >
> > Did you have CONFIG_ACPI_PROCESSOR=y in 2.6.18, or did
> > CONFIG_ACPI_PROCESSOR=m, CONFIG_X86_POWERNOW_K8=y work for you in 2.6.18?
>
> It worked with CONFIG_ACPI_PROCESSOR=m in 2.6.18-rc7. Since 2.6.19-rc1 it
> doesn't work anymore with CONFIG_ACPI_PROCESSOR=m.
>
> user@ubuntu:~/Projekte/linux-2.6.18-rc7$ uname -a
> Linux ubuntu.localnet 2.6.18-rc7 #2 SMP Wed Sep 13 11:28:41 CEST 2006 x86_64
> GNU/Linux
>
> user@ubuntu:~/Projekte/linux-2.6.18-rc7$ lsmod | grep -Ei "processor|acpi|
> power"
> powernow_k8 16096 1
> freq_table 6848 2 powernow_k8,cpufreq_stats
> cpufreq_powersave 3584 0
> asus_acpi 20644 0
> processor 36872 2 powernow_k8,thermal
>
>
> user@ubuntu:~/Projekte/linux-2.6.18-rc7$ grep -i
> ACPI_PROCESSOR /boot/config-2.6.18-rc7
> CONFIG_ACPI_PROCESSOR=m
>
> user@ubuntu:~/Projekte/linux-2.6.18-rc7$
> grep -Ei "POWERNOW_K8" /boot/config-2.6.18-rc7
> CONFIG_X86_POWERNOW_K8=m
> CONFIG_X86_POWERNOW_K8_ACPI=y
>
> +++ There's a difference in 2.6.19! CONFIG_X86_POWERNOW_K8_ACPI is gone +++
>
> user@ubuntu:~/Projekte/linux-2.6.18-rc7$
> grep -Ei "POWERNOW_K8" /boot/config-2.6.19-rc1
> CONFIG_X86_POWERNOW_K8=y
>....
It's gone because you changed CONFIG_X86_POWERNOW_K8 from m to y.
> -Christian
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
On Mon, Nov 06, 2006 at 06:35:28PM +0100, Adrian Bunk wrote:
> config X86_POWERNOW_K8
> tristate "AMD Opteron/Athlon64 PowerNow!"
> select CPU_FREQ_TABLE
> depends (ACPI_PROCESSOR || ACPI_PROCESSOR=n)
>
> But in the end, the best solution depends on how many percent of the
> X86_POWERNOW_K8 users have Christian's problem of requiring
> ACPI_PROCESSOR. If there are only very few people with this problem, I'd
> say leave it as it is.
Well, it's been this way for a while, and only recently this has come up.
There was a similar report for powernow-k7, which has a similar construct.
Dave
--
http://www.codemonkey.org.uk