[PlanetCCRMA] updates: new kernel for fc9/10

Fri Jan 9 21:29:06 PST 2009

well, I spoke to soon... I just got another hard lock, so the problem
is still present even with the drive plugged through a sata port.

Hector

On Sat, Jan 10, 2009 at 12:21 AM, Hector Centeno <hcengar at gmail.com> wrote:
> Thanks Fernando,
>
> I connected the drive to a SATA port and although I still get the
> kernel failure warnings I haven't gotten any more kernel locks! So I
> guess it is good news but also there seems to be something broken in
> the kernel. My log for the kernel failure is:
>
> BUG: sleeping function called from invalid context IRQ-19(443) at
> arch/x86/mm/highmem_32.c:8
> in_atomic():0 [00000000], irqs_disabled():1
> Pid: 443, comm: IRQ-19 Tainted: P
> 2.6.26.8-1.rt12.2.fc10.ccrma.i686.rt #1
>  [<c041ff44>] __might_sleep+0xe8/0xed
>  [<c041cd5d>] kmap+0x42/0x55
>  [<f8913db5>] ata_sff_hsm_move+0x3d3/0x66b [libata]
>  [<f89142ac>] ata_sff_interrupt+0x153/0x1eb [libata]
>  [<c0461000>] handle_IRQ_event+0x49/0xcd
>  [<c04612e5>] thread_simple_irq+0x45/0x7c
>  [<c046131c>] ? do_irqd+0x0/0x229
>  [<c04613cb>] do_irqd+0xaf/0x229
>  [<c046131c>] ? do_irqd+0x0/0x229
>  [<c043b10f>] kthread+0x3b/0x61
>  [<c043b0d4>] ? kthread+0x0/0x61
>  [<c040581f>] kernel_thread_helper+0x7/0x10
>
> Hope this helps.
>
> Cheers,
>
> Hector
>
>
> On Fri, Jan 9, 2009 at 8:25 PM, Fernando Lopez-Lezcano
> <nando at ccrma.stanford.edu> wrote:
>> On Fri, 2009-01-09 at 20:10 -0500, Hector Centeno wrote:
>>> Thanks Fernando for the quick reply. Something I forgot to mention is
>>> that I'm booting from an external USB drive that seems to be plugged
>>> right into the USB with IRQ19, which is the one being reported in the
>>> kernel failure warnings.
>>
>> Ah, yes, that sound suspicious... I tried once using an external usb
>> drive but was not very successful.
>>
>>> Maybe If I install in the internal SATA drive I won't get the locks.
>>
>> If you can do that probably it is worth a try. Or trying to use a
>> different usb interface if there is another that is not using that
>> interrupt (it could be that the problem is _because_ the drive is in
>> that interface, of course).
>>
>> -- Fernando
>>
>>
>>> On Fri, Jan 9, 2009 at 7:58 PM, Fernando Lopez-Lezcano
>>> <nando at ccrma.stanford.edu> wrote:
>>> > On Fri, 2009-01-09 at 19:49 -0500, Hector Centeno wrote:
>>> >> Hello,
>>> >>
>>> >> I just installed FC10, the recent 2.6.26.8-1.rt12.2 kernel and I'm
>>> >> also experiencing the same kind of warnings (Your system had a kernel
>>> >> failure) similar to the ones John Dey reported but in my case related
>>> >> to IRQ19 (which happens to be assigned to one of the USB controllers
>>> >> uhci_hcd:usb6 and to ata_piix. Beside that I'm experiencig occasional
>>> >> random system hard locks, sometimes at boot and other several minutes
>>> >> after. How can I debug this locks?
>>> >
>>> > Sigh, well, so much for a good kernel :-(
>>> > I have not seen a single hang yet (on fc9, same kernel)
>>> >
>>> > I don't really know how you can easily debug the hard locks. The only
>>> > method I know involves two computers with serial interfaces and a serial
>>> > null-modem cable :-) You can run the console over serial and then you
>>> > would be able to see the kernel panic or whatever is happening on the
>>> > hung computer (and use sysrq to get a list of all processes and their
>>> > state). I did that once but it is a lot of work...
>>> >
>>> > Maybe you should just run the Fedora kernel...
>>> >
>>> > You could see what warnings are you getting by looking at the output of
>>> > dmesg. Probably of the type "invalid context called from sleeping
>>> > function" or something similar (can't remember).
>>> >
>>> > -- Fernando
>>> >
>>> >
>>> >> Anything I should be trying? I
>>> >> haven't installed the nvidia drivers. I have an Asus P5B-VM
>>> >> motherboard, a Core 2 Duo CPU, 2 gigs of ram and a Nvidia 8600 video
>>> >> interface. Also I have a Edirol FA-101 but the locks happen without
>>> >> running jack. This system has been quite stable with previous RT
>>> >> kernels (pre 2.6.25).
>>> >>
>>> >> Thanks!
>>> >>
>>> >> Hector
>>> >>
>>> >>
>>> >> On Mon, Jan 5, 2009 at 3:04 PM, Fernando Lopez-Lezcano
>>> >> <nando at ccrma.stanford.edu> wrote:
>>> >> > On Mon, 2009-01-05 at 15:00 -0500, John Dey wrote:
>>> >> >> Fernando,
>>> >> >>
>>> >> >> I tried install testing rt kernel on F10 but on login got report
>>> >> >> saying:  Your system had a kernel failure.  See attachment for details.
>>> >> >
>>> >> > Yes, I have seen those as well. I posted to lkml but never got a fix for
>>> >> > the problem (the kernel seems to work fine otherwise). Maybe I should
>>> >> > try again....
>>> >> >
>>> >> > -- Fernando
>>> >> >
>>> >> >>
>>> >> >> John
>>> >> >>
>>> >> >> On Jan 5, 2009, at 1:58 PM, Fernando Lopez-Lezcano wrote:
>>> >> >>
>>> >> >> > Hi all, I just pushed new realtime kernel packages for fc9 (testing
>>> >> >> > repository) and fc10. This is 2.6.26.8-1.rt12.2 and includes both the
>>> >> >> > latest rt patch _and_ Adam Sampson's tasklet_hi_action patch that
>>> >> >> > fixes
>>> >> >> > MIDI! (kudos to Adam for his bisecting work and patch!).
>>> >> >> >
>>> >> >> > I tested this briefly on Fedora 9 and it appears to work fine.
>>> >> >> >
>>> >> >> > As usual, both i386 and x86_64 versions are available and also PAE
>>> >> >> > versions for i386. Updated planetccrma-core packages are also
>>> >> >> > available.
>>> >> >> >
>>> >> >> > I have been running 2.6.26.8-rt12 for days on end without a hung
>>> >> >> > machine, but that may only indicate I have been lucky. Anyway, maybe
>>> >> >> > this is, after all, a good realtime kernel spin... feedback welcome as
>>> >> >> > usual.
>>> >> >> >
>>> >> >> > [Also: Arnaud updated the nova ladspa plugins on fc9/fc10...]
>>> >> >> >
>>> >> >> > Enjoy! Happy New Year, Happy New Kernel...
>>> >> >
>>> >> >
>>> >> > _______________________________________________
>>> >> > PlanetCCRMA mailing list
>>> >> > PlanetCCRMA at ccrma.stanford.edu
>>> >> > http://ccrma-mail.stanford.edu/mailman/listinfo/planetccrma
>>> >> >
>>> >
>>> >
>>>
>>> _______________________________________________
>>> PlanetCCRMA mailing list
>>> PlanetCCRMA at ccrma.stanford.edu
>>> http://ccrma-mail.stanford.edu/mailman/listinfo/planetccrma
>>
>>
>