Re: USB suspend problem with kernel 2.6.17
Tiago Salem Herrmann <[email protected]>
| Newsgroups | gmane.linux.usb.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi Alan, Alan Stern escreveu: > On Wed, 12 Sep 2007, Luiz Fernando N. Capitulino wrote: > > >> Hi guys, >> >> We have an embedded system running kernel 2.6.17.14 (plus Mandriva patches) >> which has the following problem: when we suspend it to ram with an external >> USB device (which requires external power) attached, the machine doesn't >> come back from suspend. But it does work fine with bus powered devices >> attached. >> >> Unfortunately, it's not so easy to upgrade the kernel because the machine >> needs some additional patches to work. We'll try it eventually, but we'd >> like to get 2.6.17 fixed. >> >> We're still debugging it, and now we're going to try it with the minimum >> amount of patches. >> >> Something important to notice is that Alan's 'usb persist' patch is >> also applied. >> >> Is this problem familiar to someone? Any debugging tips? >> >> I've uploaded some information about the machine's hardware and >> the usb persist patch we have applied to: >> >> http://users.mandriva.com.br/~lcapitulino/usb-bug-machine-info/ >> >> Thanks a lot. >> > > You need some way to find out what's happening during the resume. For > example, a serial or network console might be able to help (or then > again, it might not -- debugging resume problems is very difficult > since by definition you're trying to get information out of a mostly > non-functional system). > It seems that the USB devices were virtually unplugged and then reattached. It only happens if there is an external *and self-powered USB device plugged, and due to this situation, the mass storage devices are renamed. Considering the sda device contains the root partition, and after the resume it is renamed to sdc, the system crashes. > For testing purposes you can always try reverting the "usb persist" > patch. > We can't revert the usb_persist patch since the root partition is in the usb mass storage. Without usb_persist patch, we can't even resume the system. Searching for information we found this thread, and it seems to be related to our problem. http://lkml.org/lkml/2007/3/27/149 Do you still have that patch you made which manages "virtual replugging"? > You should post a dmesg log showing what happens during a resume with > only bus-powered devices attached; maybe it will contain some clues. > I've uploaded the dmesg to http://users.mandriva.com.br/~salem/dmesg_suspend.txt The only way to get it was using netconsole. We'd appreciate any information you have about this problem. Thank you for your reply. Tiago Salem Herrmann Mandriva Conectiva ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ [email protected] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel