Bug 3558 - CPU frequency drops when using Standby
Summary: CPU frequency drops when using Standby
Status: CLOSED INVALID
Alias: None
Product: ACPI
Classification: Unclassified
Component: Power-Sleep-Wake (show other bugs)
Hardware: i386 Linux
: P2 normal
Assignee: Venkatesh Pallipadi
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2004-10-12 17:10 UTC by William
Modified: 2007-07-25 21:38 UTC (History)
1 user (show)

See Also:
Kernel Version: 2.6.8.1
Subsystem:
Regression: ---
Bisected commit-id:


Attachments
output of dmesg -s40000 (15.05 KB, text/plain)
2004-10-12 17:12 UTC, William
Details
output of dmidecode (10.44 KB, text/plain)
2004-10-12 17:13 UTC, William
Details
output of acpidmp (86.70 KB, text/plain)
2004-10-12 17:13 UTC, William
Details
output of lspci -vv (5.36 KB, text/plain)
2004-10-12 17:14 UTC, William
Details
output of /proc/interrupts (527 bytes, text/plain)
2004-10-12 17:15 UTC, William
Details
output of /proc/acpi/processor/CPU0/power AFTER stanby (301 bytes, text/plain)
2004-10-12 17:15 UTC, William
Details

Description William 2004-10-12 17:10:52 UTC
Distribution: 
Gentoo

Hardware Environment:
Toshiba 2250CDT (laptop)
Celeron Coppermine 600Mhz
192MB RAM
(output of lspci below)

Software Environment:
ACPI 1.6 (according to BIOS, latest BIOS upgrade)
acpid 1.0.2

Problem Description:
Very strange bug;when I switch to "stanby" sleep state (S1), the processor
speed, probably to save power, drops significantly (from 600mhz to 80mhz
according to x86info -mhz). The problem is that the frequency never goes back to
normal when the computer return from his sleep. /proc/cpuinfo shows no change,
but the decrease of speed is evident. 
Also, the processor power state switched from C2 to C1 once it gets back from
standby. I don't know if this is normal but the processor never uses C1
otherwise; cat /proc/acpi/processor/CPU0/power (see below) always shows me the
processor is in "C2" state unless I try to standby. C1 state however always has
some "counts" (but something like .0001% of C2). There seems to be no way to
bring back processor into C2 unless I reboot.
Please note this laptop doesn't support CPU frequency scaling (sadly). Also, the
ACPI version is very primitive, with the latest BIOS, it uses ACPI 1.6. I have
to force the use of ACPI (acpi=force flag while booting) or the kernel doesn't
recognize this computer as ACPI capable (APM has some bad bugs too so I prefer
to use ACPI).

Steps to reproduce:
echo "1" > /proc/acpi/sleep

(feel free to ask if you need any more info)
Comment 1 William 2004-10-12 17:12:29 UTC
Created attachment 3814 [details]
output of dmesg -s40000
Comment 2 William 2004-10-12 17:13:06 UTC
Created attachment 3815 [details]
output of dmidecode
Comment 3 William 2004-10-12 17:13:45 UTC
Created attachment 3816 [details]
output of acpidmp
Comment 4 William 2004-10-12 17:14:22 UTC
Created attachment 3817 [details]
output of lspci -vv
Comment 5 William 2004-10-12 17:15:02 UTC
Created attachment 3818 [details]
output of /proc/interrupts
Comment 6 William 2004-10-12 17:15:50 UTC
Created attachment 3819 [details]
output of /proc/acpi/processor/CPU0/power AFTER stanby
Comment 7 Len Brown 2004-12-01 19:01:23 UTC
still a problem in 2.6.9 or later?
Comment 8 William 2004-12-03 06:55:57 UTC
Yes. 
2.6.9 does the same bug. 
It's worst with 2.6.10-rc2, the computer doesn't even come back from standby, 
the "state" lid turn to green but nothing else happen and the computer doesn't 
even answer to ping. 
Comment 9 Venkatesh Pallipadi 2005-12-14 17:23:02 UTC
What driver (if any) are you using to change the frequency. 

Can you give me the output of all the files under 
/sys/devices/system/cpu/cpu0/cpufreq (if one exists)
Comment 10 William 2006-01-15 18:43:04 UTC
No driver for frequency and there is nothing in
/sys/devices/system/cpu/cpu0/cpufreq, the laptop does not support cpu throlling.
The bug seems to be related to power state ( /proc/acpi/processor/CPU0/power ),
as the CPU frequency seems to drop when it use "C2" power state. Somehow, it
stick to C2 when it come back from standby. 

Note that the problem seems to be fixed in earlier kernel. Kernel earlier than
2.6.12 ( ? ) don't get this bug, or at least, I haven't seen it so far. 
Comment 11 Venkatesh Pallipadi 2007-07-25 19:26:38 UTC
Is this C2 issue still present with latest kernel.
Please reopen the bug if the issue is still there.
Thanks.

Note You need to log in before you can comment on or make changes to this bug.