Bug 9134

Summary:	2.6.23 regression: do_nanosleep will not return
Product:	Timers	Reporter:	Rafael J. Wysocki (rjwysocki)
Component:	Other	Assignee:	john stultz (john.stultz)
Status:	CLOSED INVALID
Severity:	normal	CC:	bs, tglx
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	2.6.23-rc	Subsystem:
Regression:	Yes	Bisected commit-id:
Bug Depends on:
Bug Blocks:	9056

Description Rafael J. Wysocki 2007-10-08 12:02:02 UTC

References      : http://lkml.org/lkml/2007/10/8/59
Submitter       : Bernd Schubert <bs@q-leap.de>

Comment 1 Rafael J. Wysocki 2007-10-08 12:05:34 UTC

According to Bernd, boot option clocksource={other than acpi_pm} helps.

Comment 2 john stultz 2007-10-08 12:33:33 UTC

Any chance we can get working and not working dmesg logs?

Comment 3 Bernd Schubert 2007-10-26 03:04:59 UTC

Sorry for the long delay, actually I had/have problems to reproduce it.
After I enabled hpet in the bios, it didn't matter which clocksource I did choose, sleep/do_nanosleep() always did work as it should.
In order to create the requested dmesg log I disabled hpet again, but now sleep still worked as it should. No matter what I tried, I simply couldn't reproduce it - until I had to go back to 2.6.22. Suddenly the problem came up when I booted this version. As soon as I enabled hpet in the bios again, the problem disappeared.

Seems to be a weird hardware bug, so I suggest to close this bugzilla entry.

Comment 4 Rafael J. Wysocki 2007-10-26 15:35:04 UTC

OK, closing.

Please reopen if necessary.