Hey! Sorry for the late reply but I didn't get notifications from the
list :(
Isn't such patch enough to fix the issue?
nox.
On 17/09/13 08:55, Ritesh Raj Sarraf wrote:
Okay!!! So looks like what has happened is a lot of changes. First,
there's unofficial comments [1] stating that ondemand should not be used.
Recently, the kernel seems to have changed the CPU Scaling driver for
Intel CPUs to intel_pstate. In this driver, there's no ondemand governor.
Now, we could make the change, but I'd like to know how all this looks
on a box with an AMD CPU? Can somebody with an AMD Cpu share what
their interfaces are like?
[1] https://plus.google.com/117091380454742934025/posts/2vEekAsG2QT
On Monday 16 September 2013 07:08 PM, Ritesh Raj Sarraf wrote:
On Monday 16 September 2013 07:20 AM, Gianluigi Tiesi wrote:
[....] Laptop mode disabled, not active SETTING OF KERNEL PARAMETER FAILED: echo ondemand \>
/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor SETTING OF KERNEL PARAMETER FAILED: echo ondemand
\> /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor SETTING OF KERNEL PARAMETER FAILED: echo
ondemand \> /sys/devices/system/cpu/cpu2/cpufreq/scaling_governor SETTING OF KERNEL PARAMETER
FAILED: echo ondemand \> /sys/devices/system/cpu/cpu3/cpufreq/scaling_governor SETTING OF KERNEL
PARAMETER FAILED: echo ondemand \> /sys/devices/system/cpu/cpu4/cpufreq/scaling_governor SETTING OF
KERNEL PARAMETER FAILED: echo ondemand \> /sys/devices/system/cpu/cpu5/cpufreq/scaling_governor
SETTING OF KERNEL PARAMETER FAILED: echo ondemand \>
/sys/devices/system/cpu/cpu6/cpufreq/scaling_governor SETTING OF KERNEL PARAMETER FAILED: echo ondemand
\> /sys/devices/system/cpu/cpu7/cpufreq/sca[ ok governor.
while:
cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors
performance powersave
echo ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
-bash: echo: write error: Invalid argument
Yes. I've noticed that too. Thank you for reporting the bug report. I
need to investigate what the new expected behavior is going to be...
On Tuesday 10 September 2013 02:51 PM, Ritesh Raj Sarraf wrote:
I couldn't cover for this issue in 1.64 release. If people have
suggestions on how to fix this, please drop a note on this list.
On Thursday 08 August 2013 10:05 AM, Ritesh Raj Sarraf wrote:
I plan on a release next week and would like to resolve this issue.
Have you seen the interface (or a different interface) for the 3.10
kernel on your setup?
On Saturday 20 July 2013 03:52 PM, Ritesh Raj Sarraf wrote:
I see what changed.
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=4d5dcc4211f9def4281eafb54b8ed483862e8135
But I'm still not clear why my box doesn't have the ignore_nice
interface
On Saturday 20 July 2013 03:41 PM, Ritesh Raj Sarraf wrote:
Have things changed for others??
This is what I see on my kernel:
rrs@zan:/lib/modules/3.10-1-amd64/kernel/drivers/cpufreq$ ls
/sys/devices/system/cpu/cpufreq/
rrs@zan:/lib/modules/3.10-1-amd64/kernel/drivers/cpufreq$ ls
/sys/devices/system/cpu/cpu0
cache/ cpuidle/ crash_notes_size node0@ subsystem@
topology/
cpufreq/ crash_notes microcode/ power/ thermal_throttle/
uevent
rrs@zan:/lib/modules/3.10-1-amd64/kernel/drivers/cpufreq$ ls
/sys/devices/system/cpu/cpu0/cpufreq/
affected_cpus cpuinfo_min_freq scaling_available_governors
scaling_max_freq
cpuinfo_cur_freq cpuinfo_transition_latency
scaling_driver scaling_min_freq
cpuinfo_max_freq related_cpus scaling_governor
scaling_setspeed
On my kernel, ondemand is set to in-kernel, but that should have
broken the sysfs interface?????
# CONFIG_CPU_FREQ_STAT_DETAILS is not set
# CONFIG_CPU_FREQ_DEFAULT_GOV_PERFORMANCE is not set
# CONFIG_CPU_FREQ_DEFAULT_GOV_USERSPACE is not set
CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
# CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE 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=y
CONFIG_CPU_FREQ_GOV_CONSERVATIVE=m
--
Ritesh Raj Sarraf
RESEARCHUT -http://www.researchut.com
"Necessity is the mother of invention."
--- /usr/share/laptop-mode-tools/modules/cpufreq 2013-09-15 20:05:11.000000000 +0300
+++ /usr/share/laptop-mode-tools/modules/cpufreq 2013-10-04 21:22:42.521002802 +0300
@@ -99,6 +99,10 @@
if [ -f "/sys/devices/system/cpu/cpufreq/$THIS_CPU_GOVERNOR/ignore_nice_load" ]; then
log "VERBOSE" "Setting CPU ignore_nice_load for all cpus"
set_sysctl /sys/devices/system/cpu/cpufreq/$THIS_CPU_GOVERNOR/ignore_nice_load $CPU_IGNORE_NICE_LOAD
+ # For kernels 3.10 and above the sysfs has yet another interface
+ elif [ -f "/sys/devices/system/cpu/cpufreq/$THIS_CPU_GOVERNOR/ignore_nice" ]; then
+ log "VERBOSE" "Setting CPU ignore_nice_load for all cpus"
+ set_sysctl /sys/devices/system/cpu/cpufreq/$THIS_CPU_GOVERNOR/ignore_nice $CPU_IGNORE_NICE_LOAD
else
log "VERBOSE" "No generic CPU ignore_nice_load interface available"
fi
