found 652475 3.2.4-1 thanks Hi there! On Fri, 30 Dec 2011 13:48:59 +0100, Ben Hutchings wrote: > On Sat, 2011-12-17 at 17:17 +0100, Luca Capello wrote: >> Package: linux-2.6 >> Version: 3.1.1-1 The bug is still present in newer kernels, version updated. >> There is a problem somewhere with USB autosuspending on my ThinkPad X60. >> When enabled through powertop, the kernel splits out messages different >> times every seconds, thus filling the log files: > [...] > > I don't know that there is anything wrong with the auto-suspend > behaviour. Are your USB devices working correctly? Yes, there is no problem with them except with an external 100 GB S-ATA HD, but this happens when copying >5 GB files and also with other machines/OSs, thus I guess the fault is the double S-ATA/IDE converter which does not give enough power to the HD (everything is fine with HDs smaller in size). However, I should note that sometime there is a general problem I have already reported in the past, IIRC directly to the linux-acpi mailing list (sorry, I do not have direct link right now): ===== Message from syslogd@gismo at Feb 20 14:12:52 ... kernel:[184232.065089] Uhhuh. NMI received for unknown reason 31 on CPU 0. Message from syslogd@gismo at Feb 20 14:12:52 ... kernel:[184232.065141] Do you have a strange power saving mode enabled? Message from syslogd@gismo at Feb 20 14:12:52 ... kernel:[184232.065182] Dazed and confused, but trying to continue ===== > Clearly it should not generate so many log messages. Can you test > whether the attached patch fixes that problem for you? See > <http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-official> for instructions on how to do this. I am sorry for the delay, but I tried to build a kernel and it stopped because of disk space (I had something like 2.5 GB available). Then I freed 8.1 GB and still this was not enough. I would like some of this information was already available on the kernel-handbook documentation, at least I would have not tried beforehand. Never mind, I managed to freed 18 GB and compile 3.2.4-1 with the suggested patch, the resulting .deb are available upon request (I sort of hated that the compilation was not done on a clean environment à la pbuilder, but this is another story). With this kernel I still get the same messages as without the patch :-( I found something strange with plain Debian kernels: after the first pm-hibernate cycle, the bug is not reproducible, i.e. activating autosuspend is OK. It seems that de- and re-activating autosuspend is enough to trigger the bug. But please take this as a pure comment. Please also note that the culprit seems to be the activation of both powertop's "Runtime PM for PCI Device Intel Corporation N10/ICH 7 Family USB UHCI Controller #[...]" and "Autosuspend for USB device [...]HCI Host Controller [usb[...]]" voices: activating only one of the two seems to be OK. Thx, bye, Gismo / Luca
Attachment:
pgpPWY59rv5bI.pgp
Description: PGP signature