** Summary changed:
- Resume immediately after Suspend due to ACPI PM Timer event
+ Resume immediately after Suspend due to some WAKE event
** Description changed:
- When suspending the PC resumes immediately.
+ When suspending (S3) the PC resumes immediately.
- Using a custom-built debugging kernel reveals that the ACPI Power
- Management Timer has generated a wake-up event.
+ This occurs on a Sony PCG-SRC51P with a fresh install of Feisty. It also
+ affects Gutsy up to and beyond tribe-3 at least (all 32-bit).
- Investigation of this issue is ongoing.
+ On an identical model (SRX41P) the problem doesn't occur. Have tried
+ swapping the hard disks and memory modules between the two but the
+ problem persists on the SRX51P.
+
+ Both models have identical hardware but the SRX51P CPU is 50MHz faster,
+ and it has 128MB more RAM (SRX41P = Intel P-III 800Mhz 256MB RAM, SRX51P
+ = Intel P-III 850MHz 384MB RAM).
+
+ Using a custom-built debugging kernel reveals that some device has
+ generated a WAKE-UP event.
+
+ Running a diff against the hardware profiles from lshal shows both
+ models are identical. Running diffs on their dmesg confirms this.
+
+ A diff of the kern.log over a suspend/resume cycle is also identical.
+
+ The additional debug-message reporting I added into hwsleep() and
+ acpi_pm_enter() show identical activity in both. Unfortunately, even
+ with ACPI_FUTURE_USAGE enabled reading the WAKE event on resume always
+ reports it as PM_TIMER, even on the 'good' PC so the wake-event code
+ doesn't appear reliable.
+
+ When I thought PM_TIMER was correct I added code to disable the PM_TIMER
+ wake event just prior to suspend, and re-enabled it upon resume. That is
+ the reason for the debug messages below.
+
+ Here's the output to kern.log from my debug code starting in hwsleep()
+ and continuing in acpi_pm_enter() on resume:
+
+ agpgart-intel 0000:00:00.0: LATE suspend
+ hwgpe-0343 [05] hw_enable_wakeup_gpe_b: Enable Wake-up GPE for _GPE
+ hwgpe-0344 [05] hw_enable_wakeup_gpe_b: register_info->enable_for_wake 0x20
+ hwgpe-0345 [05] hw_enable_wakeup_gpe_b: event_info->register_bit 0x1
+ hwsleep-0284 [03] enter_sleep_state : PM Timer Event Enable value: 0x0
+ hwsleep-0285 [03] enter_sleep_state : PM1Enable Enable value: 0x20
+ hwsleep-0286 [03] enter_sleep_state : Disabling PM Timer
+ hwsleep-0290 [03] enter_sleep_state : PM Timer Event Enable value: 0x0
+ hwsleep-0301 [03] enter_sleep_state : Entering sleep state [S3]
+ hwsleep-0325 [03] enter_sleep_state : Writing PM1A (SLP_TYP data) 1401
+ hwregs-0698 [04] hw_register_write : Writing PM1A_CONTROL @ 1004
+ hwsleep-0336 [03] enter_sleep_state : Writing PM1B (SLP_TYP data) 1401
+ hwregs-0708 [04] hw_register_write : Writing PM1B_CONTROL @ 0
+ hwsleep-0354 [03] enter_sleep_state : Will write PM1A (SLP_TYP + SLP_EN)
3401
+ hwsleep-0355 [03] enter_sleep_state : Will write PM1B (SLP_TYP + SLP_EN)
3401
+ hwsleep-0358 [03] enter_sleep_state : About to flush CPU cache
+ Back to C!
+ Enabling PM Timer
+ Resume status=0
+ Resume ACPI state=3
+ Event 0=0x1
+ Event 1=0x0
+ Event 2=0x0
+ Event 3=0x0
+ agpgart-intel 0000:00:00.0: EARLY resume
** Description changed:
When suspending (S3) the PC resumes immediately.
- This occurs on a Sony PCG-SRC51P with a fresh install of Feisty. It also
+ This occurs on a Sony PCG-SRX51P with a fresh install of Feisty. It also
affects Gutsy up to and beyond tribe-3 at least (all 32-bit).
On an identical model (SRX41P) the problem doesn't occur. Have tried
swapping the hard disks and memory modules between the two but the
problem persists on the SRX51P.
Both models have identical hardware but the SRX51P CPU is 50MHz faster,
and it has 128MB more RAM (SRX41P = Intel P-III 800Mhz 256MB RAM, SRX51P
= Intel P-III 850MHz 384MB RAM).
Using a custom-built debugging kernel reveals that some device has
generated a WAKE-UP event.
Running a diff against the hardware profiles from lshal shows both
models are identical. Running diffs on their dmesg confirms this.
A diff of the kern.log over a suspend/resume cycle is also identical.
The additional debug-message reporting I added into hwsleep() and
acpi_pm_enter() show identical activity in both. Unfortunately, even
with ACPI_FUTURE_USAGE enabled reading the WAKE event on resume always
reports it as PM_TIMER, even on the 'good' PC so the wake-event code
doesn't appear reliable.
When I thought PM_TIMER was correct I added code to disable the PM_TIMER
wake event just prior to suspend, and re-enabled it upon resume. That is
the reason for the debug messages below.
+
+ If I disable *all* wake events just prior to suspend the PC doesn't
+ resume immediately, but of course a hard reset is required since the
+ PWRBTN has also been disabled!
Here's the output to kern.log from my debug code starting in hwsleep()
and continuing in acpi_pm_enter() on resume:
agpgart-intel 0000:00:00.0: LATE suspend
hwgpe-0343 [05] hw_enable_wakeup_gpe_b: Enable Wake-up GPE for _GPE
hwgpe-0344 [05] hw_enable_wakeup_gpe_b: register_info->enable_for_wake 0x20
hwgpe-0345 [05] hw_enable_wakeup_gpe_b: event_info->register_bit 0x1
hwsleep-0284 [03] enter_sleep_state : PM Timer Event Enable value: 0x0
hwsleep-0285 [03] enter_sleep_state : PM1Enable Enable value: 0x20
hwsleep-0286 [03] enter_sleep_state : Disabling PM Timer
hwsleep-0290 [03] enter_sleep_state : PM Timer Event Enable value: 0x0
hwsleep-0301 [03] enter_sleep_state : Entering sleep state [S3]
hwsleep-0325 [03] enter_sleep_state : Writing PM1A (SLP_TYP data) 1401
hwregs-0698 [04] hw_register_write : Writing PM1A_CONTROL @ 1004
hwsleep-0336 [03] enter_sleep_state : Writing PM1B (SLP_TYP data) 1401
hwregs-0708 [04] hw_register_write : Writing PM1B_CONTROL @ 0
hwsleep-0354 [03] enter_sleep_state : Will write PM1A (SLP_TYP + SLP_EN)
3401
hwsleep-0355 [03] enter_sleep_state : Will write PM1B (SLP_TYP + SLP_EN)
3401
hwsleep-0358 [03] enter_sleep_state : About to flush CPU cache
Back to C!
Enabling PM Timer
Resume status=0
Resume ACPI state=3
Event 0=0x1
Event 1=0x0
Event 2=0x0
Event 3=0x0
agpgart-intel 0000:00:00.0: EARLY resume
** Description changed:
When suspending (S3) the PC resumes immediately.
This occurs on a Sony PCG-SRX51P with a fresh install of Feisty. It also
affects Gutsy up to and beyond tribe-3 at least (all 32-bit).
On an identical model (SRX41P) the problem doesn't occur. Have tried
swapping the hard disks and memory modules between the two but the
problem persists on the SRX51P.
Both models have identical hardware but the SRX51P CPU is 50MHz faster,
and it has 128MB more RAM (SRX41P = Intel P-III 800Mhz 256MB RAM, SRX51P
= Intel P-III 850MHz 384MB RAM).
Using a custom-built debugging kernel reveals that some device has
generated a WAKE-UP event.
Running a diff against the hardware profiles from lshal shows both
models are identical. Running diffs on their dmesg confirms this.
A diff of the kern.log over a suspend/resume cycle is also identical.
The additional debug-message reporting I added into hwsleep() and
acpi_pm_enter() show identical activity in both. Unfortunately, even
with ACPI_FUTURE_USAGE enabled reading the WAKE event on resume always
reports it as PM_TIMER, even on the 'good' PC so the wake-event code
doesn't appear reliable.
When I thought PM_TIMER was correct I added code to disable the PM_TIMER
wake event just prior to suspend, and re-enabled it upon resume. That is
the reason for the debug messages below.
If I disable *all* wake events just prior to suspend the PC doesn't
resume immediately, but of course a hard reset is required since the
- PWRBTN has also been disabled!
+ PWRBTN has also been disabled! This would seem to confirm that an
+ unexpected WAKE event is the cause.
Here's the output to kern.log from my debug code starting in hwsleep()
and continuing in acpi_pm_enter() on resume:
agpgart-intel 0000:00:00.0: LATE suspend
hwgpe-0343 [05] hw_enable_wakeup_gpe_b: Enable Wake-up GPE for _GPE
hwgpe-0344 [05] hw_enable_wakeup_gpe_b: register_info->enable_for_wake 0x20
hwgpe-0345 [05] hw_enable_wakeup_gpe_b: event_info->register_bit 0x1
hwsleep-0284 [03] enter_sleep_state : PM Timer Event Enable value: 0x0
hwsleep-0285 [03] enter_sleep_state : PM1Enable Enable value: 0x20
hwsleep-0286 [03] enter_sleep_state : Disabling PM Timer
hwsleep-0290 [03] enter_sleep_state : PM Timer Event Enable value: 0x0
hwsleep-0301 [03] enter_sleep_state : Entering sleep state [S3]
hwsleep-0325 [03] enter_sleep_state : Writing PM1A (SLP_TYP data) 1401
hwregs-0698 [04] hw_register_write : Writing PM1A_CONTROL @ 1004
hwsleep-0336 [03] enter_sleep_state : Writing PM1B (SLP_TYP data) 1401
hwregs-0708 [04] hw_register_write : Writing PM1B_CONTROL @ 0
hwsleep-0354 [03] enter_sleep_state : Will write PM1A (SLP_TYP + SLP_EN)
3401
hwsleep-0355 [03] enter_sleep_state : Will write PM1B (SLP_TYP + SLP_EN)
3401
hwsleep-0358 [03] enter_sleep_state : About to flush CPU cache
Back to C!
Enabling PM Timer
Resume status=0
Resume ACPI state=3
Event 0=0x1
Event 1=0x0
Event 2=0x0
Event 3=0x0
agpgart-intel 0000:00:00.0: EARLY resume
** Attachment added: "side-by-side diff (SRX41P, SRX51P) of lshal"
http://launchpadlibrarian.net/8732201/lshal.diff.txt
--
Resume immediately after Suspend due to some WAKE event
https://bugs.launchpad.net/bugs/128315
You received this bug notification because you are a member of Ubuntu
Bugs, which is the bug contact for Ubuntu.
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
