>-----Original Message-----
>From: [email protected]
>[mailto:[email protected]] On Behalf Of
>Borislav Petkov
>Sent: Sunday, October 30, 2005 12:36 AM
>To: Linus Torvalds
>Cc: Linux Kernel Mailing List;
>[email protected]; [email protected]
>Subject: Re: Linux 2.6.14 ehci-hcd hangs machine
>
>On Thu, Oct 27, 2005 at 05:28:50PM -0700, Linus Torvalds wrote:
>>
>> Ok, it's finally there.
>... and it still won't boot on my machine. It hangs while initializing
>the ehci usb host controller saying:
>
><snip>
>...
>[4294691.834000] usb usb3: Product: UHCI Host Controller
>[4294691.840000] usb usb3: Manufacturer: Linux 2.6.14 uhci_hcd
>[4294691.847000] usb usb3: SerialNumber: 0000:00:1d.2
>[4294691.880000] hub 3-0:1.0: USB hub found
>[4294691.885000] hub 3-0:1.0: 2 ports detected
>[4294694.855000] ACPI: PCI Interrupt 0000:00:1d.7[D] -> GSI 23 (level,
> low) -> IRQ 20
>[4294694.864000] ehci_hcd 0000:00:1d.7: EHCI Host Controller
>[4294694.870000] ehci_hcd 0000:00:1d.7: debug port 1
></snip>
>
>and dies. This bug is actually in there since 2.6.14-rc4 (see:
>http://bugzilla.kernel.org/show_bug.cgi?id=5428) and David Brownell
>supplied a patch which turned out to be useless eventually
>since _rebooting_
>the kernel with the 'usb-handoff' (and without the patch)
>solved the problem.
>As it turns out, it actually solves the problem only for the
>reboot case.
>My machine still hangs on an initial boot with and without
>'usb-handoff'.
>.config attached.
Boris,
While running with 'usb-handoff' turned on, do you see something like
"EHCI early BIOS handoff failed" (in power on or reboot cases) ?
Aleks.
>
>Regards,
> Boris.
>
On Mon, Oct 31, 2005 at 09:53:15AM -0800, Aleksey Gorelov wrote:
> >-----Original Message-----
> >From: [email protected]
> >[mailto:[email protected]] On Behalf Of
> >Borislav Petkov
> >Sent: Sunday, October 30, 2005 12:36 AM
> >To: Linus Torvalds
> >Cc: Linux Kernel Mailing List;
> >[email protected]; [email protected]
> >Subject: Re: Linux 2.6.14 ehci-hcd hangs machine
> >
> >On Thu, Oct 27, 2005 at 05:28:50PM -0700, Linus Torvalds wrote:
> >>
> >> Ok, it's finally there.
> >... and it still won't boot on my machine. It hangs while initializing
> >the ehci usb host controller saying:
> >
> ><snip>
> >...
> >[4294691.834000] usb usb3: Product: UHCI Host Controller
> >[4294691.840000] usb usb3: Manufacturer: Linux 2.6.14 uhci_hcd
> >[4294691.847000] usb usb3: SerialNumber: 0000:00:1d.2
> >[4294691.880000] hub 3-0:1.0: USB hub found
> >[4294691.885000] hub 3-0:1.0: 2 ports detected
> >[4294694.855000] ACPI: PCI Interrupt 0000:00:1d.7[D] -> GSI 23 (level,
> > low) -> IRQ 20
> >[4294694.864000] ehci_hcd 0000:00:1d.7: EHCI Host Controller
> >[4294694.870000] ehci_hcd 0000:00:1d.7: debug port 1
> ></snip>
> >
> >and dies. This bug is actually in there since 2.6.14-rc4 (see:
> >http://bugzilla.kernel.org/show_bug.cgi?id=5428) and David Brownell
> >supplied a patch which turned out to be useless eventually
> >since _rebooting_
> >the kernel with the 'usb-handoff' (and without the patch)
> >solved the problem.
> >As it turns out, it actually solves the problem only for the
> >reboot case.
> >My machine still hangs on an initial boot with and without
> >'usb-handoff'.
> >.config attached.
>
> Boris,
>
> While running with 'usb-handoff' turned on, do you see something like
> "EHCI early BIOS handoff failed" (in power on or reboot cases) ?
Nope,
nothing of the like in the serial console log.
Regards,
Boris.
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
On Monday 31 October 2005 2:15 pm, Borislav Petkov wrote:
> > >and dies. This bug is actually in there since 2.6.14-rc4 (see:
> > >http://bugzilla.kernel.org/show_bug.cgi?id=5428) and David Brownell
> > >supplied a patch which turned out to be useless eventually
> > >since _rebooting_
> > >the kernel with the 'usb-handoff' (and without the patch)
> > >solved the problem.
In current GIT kernels (2.6.14+) that patch is now merged, and
also usb-handoff is the default.
> > >As it turns out, it actually solves the problem only for the
> > >reboot case.
> > >My machine still hangs on an initial boot with and without
> > >'usb-handoff'.
Huh? That's not what you'd reported before. What changed?
You're saying that _with_ that patch, and usb-handoff, it fails?
> > While running with 'usb-handoff' turned on, do you see something like
> > "EHCI early BIOS handoff failed" (in power on or reboot cases) ?
>
> Nope, nothing of the like in the serial console log.
If the problem was IRQs getting somehow enabled after the handoff,
that patch would probably be a very useful thing. If that does
help, I'll suggest that be merged into 2.6.14.x ... it'd have been
good to merge for 2.6.14.0 but quick merges seem mostly to be a
thing of the past any more.
- Dave
On Mon, Oct 31, 2005 at 04:24:31PM -0800, David Brownell wrote:
> On Monday 31 October 2005 2:15 pm, Borislav Petkov wrote:
> > > >and dies. This bug is actually in there since 2.6.14-rc4 (see:
> > > >http://bugzilla.kernel.org/show_bug.cgi?id=5428) and David Brownell
> > > >supplied a patch which turned out to be useless eventually
> > > >since _rebooting_
> > > >the kernel with the 'usb-handoff' (and without the patch)
> > > >solved the problem.
>
> In current GIT kernels (2.6.14+) that patch is now merged, and
> also usb-handoff is the default.
Just tested 2.6.14 from the git repo of today and the machine hangs at
the same point in the boot process.
>
>
> > > >As it turns out, it actually solves the problem only for the
> > > >reboot case.
> > > >My machine still hangs on an initial boot with and without
> > > >'usb-handoff'.
>
> Huh? That's not what you'd reported before. What changed?
> You're saying that _with_ that patch, and usb-handoff, it fails?
Yeah, the symptoms are really weird. Let me rehash the whole history:
First, we did some testing with 2.6.14-rc4, _with_ the patch and the
'handoff' cmd option and it worked. Then, several boots later, I noticed
that it started hanging itself again at the same position not while rebooting
but at _initial_ boot in the morning. Then, you told me on bugzilla that
in my case the patch shouldn't be needed since i'm using 'usb-handoff'
but, unfortunately, without the patch and with usb-handoff it hangs itself too.
2.6.14 shows the same behavior and the git kernel i compiled this morning hangs
itself too, as i said above.
To sum up, IMHO, the patch helps partially only while rebooting. Initial
boots still remain hanging so there's got to be something else that
kills the machine. maybe some IRQs or something? The thing is that not
even sysrq is functional, I don't get any BUG() traces or anything -
the hang is pretty tough. 2.6.13, in contrast, boots just fine. Hope
that helps,
Boris.
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
On Tuesday 01 November 2005 3:23 am, Borislav Petkov wrote:
> Yeah, the symptoms are really weird. Let me rehash the whole history:
> First, we did some testing with 2.6.14-rc4, _with_ the patch and the
> 'handoff' cmd option and it worked. Then, several boots later, I noticed
> that it started hanging itself again at the same position not while rebooting
> but at _initial_ boot in the morning. ...
So you're saying your hardware doesn't act consistently.
Is there a BIOS update you can try? The failure sure seems
to be board-specific.
> ... 2.6.13, in contrast, boots just fine. Hope
> that helps,
Well that's news, and not mentioned in the bug report; in fact, you
said explicitly that it _never_ worked on earlier kernels (see your
comment #2).
This means you could use "git bisect" to point a finger at a patch that,
if it doesn't actually cause the problem, at least surfaces a latent bug
in other code. Could you please try that?
I'm starting to suspect some IRQ setup problem here; those are classically
issues in ACPI code, even when the breakage shows only with USB.
- Dave
On Tue, Nov 01, 2005 at 10:18:04AM -0800, David Brownell wrote:
> On Tuesday 01 November 2005 3:23 am, Borislav Petkov wrote:
> > Yeah, the symptoms are really weird. Let me rehash the whole history:
> > First, we did some testing with 2.6.14-rc4, _with_ the patch and the
> > 'handoff' cmd option and it worked. Then, several boots later, I noticed
> > that it started hanging itself again at the same position not while rebooting
> > but at _initial_ boot in the morning. ...
>
> So you're saying your hardware doesn't act consistently.
> Is there a BIOS update you can try? The failure sure seems
> to be board-specific.
Just did a BIOS update - the 2.6.14 git version from yesterday booted
just fine.
>
>
> > ... 2.6.13, in contrast, boots just fine. Hope
> > that helps,
>
> Well that's news, and not mentioned in the bug report; in fact, you
> said explicitly that it _never_ worked on earlier kernels (see your
> comment #2).
Ups, my bad, i got kinda confused by the formulation "most recent kernel
where..", sorry.
> This means you could use "git bisect" to point a finger at a patch that,
> if it doesn't actually cause the problem, at least surfaces a latent bug
> in other code. Could you please try that?
>
> I'm starting to suspect some IRQ setup problem here; those are classically
> issues in ACPI code, even when the breakage shows only with USB.
This is superfluous, I guess, now that 2.6.14 boots.
Thanks for your help.
Regards,
Boris.
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de