mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* RE: VIA KT400 + Kernel 2.6.12 + IO-APIC + uhci_hcd = IRQ trouble
@ 2005-07-26  2:13 Protasevich, Natalie
  2005-07-26 10:48 ` [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd " Michel Bouissou
  0 siblings, 1 reply; 7+ messages in thread
From: Protasevich, Natalie @ 2005-07-26  2:13 UTC (permalink / raw)
  To: Alan Stern, Michel Bouissou; +Cc: linux-kernel


> > Le Lundi 25 Juillet 2005 22:44, Alan Stern a écrit :
> > >
> > > Now that's strange.  When you plug the high-speed device into the 
> > > integrated ports, which IRQ counter changes?  Since 
> nothing is using 
> > > IRQ 21, it should be disabled and its counter should remain 
> > > constant.  Does this mean the interrupts show up on IRQ 
> 19 (used by 
> > > ehci-hcd), or do they not show up at all (i.e., is the 
> USB connection just being polled)?
> > 
> > I assume it's IRQ 19.
> > 
> > cat /proc/interrupts doesn't show IRQ21 at all when uhci 
> isn't loaded.
> 
> As it shouldn't, since nothing is supposed to be using that IRQ.
> 
> > IRQ 19 being shared with 4 IDE controllers that controls my hard 
> > drives, that's hard to isolate interrupts counts due to USB 
> activity 
> > from interrupts counts due to disks activity...
> 
> I was afraid you'd say that...
> 
> Natalie, that's all I can think of.  Now it's up to you to 
> invent a patch Michel can try out, to show just where the 
> IO-APIC code is going wrong.

I will sure try... I'm keeping an eye on your exchange don't worry :) 
just have to get done with urgent work piled up here while on my trip :< ...

--N
 
> Alan Stern
> 
> 

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd = IRQ trouble
  2005-07-26  2:13 VIA KT400 + Kernel 2.6.12 + IO-APIC + uhci_hcd = IRQ trouble Protasevich, Natalie
@ 2005-07-26 10:48 ` Michel Bouissou
  2005-07-26 18:54   ` Alan Stern
  0 siblings, 1 reply; 7+ messages in thread
From: Michel Bouissou @ 2005-07-26 10:48 UTC (permalink / raw)
  To: Protasevich, Natalie; +Cc: Alan Stern, dbrownell, linux-kernel

Le Mardi 26 Juillet 2005 04:13, Protasevich, Natalie a écrit :
> > > Le Lundi 25 Juillet 2005 22:44, Alan Stern a écrit :
> >
> > Natalie, that's all I can think of.  Now it's up to you to
> > invent a patch Michel can try out, to show just where the
> > IO-APIC code is going wrong.
>
> I will sure try... I'm keeping an eye on your exchange don't worry :)
> just have to get done with urgent work piled up here while on my trip :<
> ...

I'm afraid that I may have accidentally solved my problem ;-)

I've upgraded my Gigabyte GA7-VAXP motherboard's BIOS from release 7VAXP.F11 
to 7VAXP.F15, and it seems the problem is gone !

The strange thing is that now, cat /proc/interrupts shows that both uhci_hcd 
and ehci_hcd use the same IRQ (21) :

[root@totor etc]# cat /proc/interrupts
           CPU0
  0:     569004    IO-APIC-edge  timer
  1:       2107    IO-APIC-edge  i8042
  2:          0          XT-PIC  cascade
  4:       1589    IO-APIC-edge  serial
  7:          2    IO-APIC-edge  parport0
 14:       4615    IO-APIC-edge  ide4
 15:       4624    IO-APIC-edge  ide5
 16:      29033   IO-APIC-level  nvidia
 18:       7684   IO-APIC-level  eth0, eth1
 19:      28049   IO-APIC-level  ide0, ide1, ide2, ide3
 21:       7128   IO-APIC-level  ehci_hcd:usb1, uhci_hcd:usb2, uhci_hcd:usb3, 
uhci_hcd:usb4
 22:       5969   IO-APIC-level  VIA8233
NMI:          0
LOC:     568946
ERR:          0
MIS:          0

...and then, the system feels happy. I've played around with USB devices of 
all speeds in all sockets, and there are no "irq 21: nobody cared!" messages 
anymore...

Still, high-speed USB devices seem to work only in the motherboard integrated 
USB sockets, otherwise I get some "usb 1-4: device not accepting address 26, 
error -71" messages, but this isn't an issue for me, and might be a cable 
problem as you said...

So it seems that the BIOS upgrade fixed the IRQ issue (or masked it ?).

What is weird is that, from the Gigabyte's BIOS list, I wouldn't have expected 
this upgrade to fix this, as their changelog from F11 to F15 only says:

F11: AMD Barton CPU support (2500+/2800+/3000+) 

F13: Fixed BIOS flash utility(flash8xx) can't be used issue

F14: Support new AMD Duron model8 CPU (64K L2 Cache FSB 266)

F15: Added AMD Sempron CPU support

So this BIOS upgrade was just a desperate move on my part ;-)

Well, anyway, things look much better here now...

Natalie, if you would like me to try some of your VIA patches to check them 
out on my system, I would be glad to help in my turn...

Cheers, and thanks to all of you again for all your precious help.

-- 
Michel Bouissou <michel@bouissou.net> OpenPGP ID 0xDDE8AC6E

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd = IRQ trouble
  2005-07-26 10:48 ` [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd " Michel Bouissou
@ 2005-07-26 18:54   ` Alan Stern
  2005-07-26 19:39     ` Michel Bouissou
  0 siblings, 1 reply; 7+ messages in thread
From: Alan Stern @ 2005-07-26 18:54 UTC (permalink / raw)
  To: Michel Bouissou; +Cc: Protasevich, Natalie, dbrownell, linux-kernel

On Tue, 26 Jul 2005, Michel Bouissou wrote:

> I'm afraid that I may have accidentally solved my problem ;-)
> 
> I've upgraded my Gigabyte GA7-VAXP motherboard's BIOS from release 7VAXP.F11 
> to 7VAXP.F15, and it seems the problem is gone !
> 
> The strange thing is that now, cat /proc/interrupts shows that both uhci_hcd 
> and ehci_hcd use the same IRQ (21) :
> 
> [root@totor etc]# cat /proc/interrupts
>            CPU0
>   0:     569004    IO-APIC-edge  timer
>   1:       2107    IO-APIC-edge  i8042
>   2:          0          XT-PIC  cascade
>   4:       1589    IO-APIC-edge  serial
>   7:          2    IO-APIC-edge  parport0
>  14:       4615    IO-APIC-edge  ide4
>  15:       4624    IO-APIC-edge  ide5
>  16:      29033   IO-APIC-level  nvidia
>  18:       7684   IO-APIC-level  eth0, eth1
>  19:      28049   IO-APIC-level  ide0, ide1, ide2, ide3
>  21:       7128   IO-APIC-level  ehci_hcd:usb1, uhci_hcd:usb2, uhci_hcd:usb3, 
> uhci_hcd:usb4
>  22:       5969   IO-APIC-level  VIA8233
> NMI:          0
> LOC:     568946
> ERR:          0
> MIS:          0
> 
> ...and then, the system feels happy. I've played around with USB devices of 
> all speeds in all sockets, and there are no "irq 21: nobody cared!" messages 
> anymore...

Not strange at all, since the EHCI controller actually _was_ using IRQ 21!  
The difference is that now Linux knows this.

> Still, high-speed USB devices seem to work only in the motherboard integrated 
> USB sockets, otherwise I get some "usb 1-4: device not accepting address 26, 
> error -71" messages, but this isn't an issue for me, and might be a cable 
> problem as you said...
> 
> So it seems that the BIOS upgrade fixed the IRQ issue (or masked it ?).
> 
> What is weird is that, from the Gigabyte's BIOS list, I wouldn't have expected 
> this upgrade to fix this, as their changelog from F11 to F15 only says:
> 
> F11: AMD Barton CPU support (2500+/2800+/3000+) 
> 
> F13: Fixed BIOS flash utility(flash8xx) can't be used issue
> 
> F14: Support new AMD Duron model8 CPU (64K L2 Cache FSB 266)
> 
> F15: Added AMD Sempron CPU support
> 
> So this BIOS upgrade was just a desperate move on my part ;-)

Sometimes people don't add things into the changelog...

So long as it's working now, that's the important thing.  I'll have to 
remember this BIOS update trick the next time someone has an equivalent 
IRQ-mapping problem.

Alan Stern


^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd = IRQ trouble
  2005-07-26 18:54   ` Alan Stern
@ 2005-07-26 19:39     ` Michel Bouissou
  2005-07-26 19:44       ` Michel Bouissou
  2005-07-26 23:37       ` Grant Coady
  0 siblings, 2 replies; 7+ messages in thread
From: Michel Bouissou @ 2005-07-26 19:39 UTC (permalink / raw)
  To: Alan Stern; +Cc: Protasevich, Natalie, dbrownell, linux-kernel

Le Mardi 26 Juillet 2005 20:54, Alan Stern a écrit :
> >
> > ...and then, the system feels happy. I've played around with USB devices
> > of all speeds in all sockets, and there are no "irq 21: nobody cared!"
> > messages anymore...
>
> Not strange at all, since the EHCI controller actually _was_ using IRQ 21!
> The difference is that now Linux knows this.
[...]
> So long as it's working now, that's the important thing.

Yes, but it doesn't tell us why kernels 2.4x felt perfectly happy with the old 
BIOS...

Furthermore, I'm afraid I may have exchanged a problem for a worse one :-(
Besides this USB/IRQ issue, my system was completely stable and had been for 2 
years. Never experienced a system hang (except when I tried early 2.6 kernels 
up to 2.6.8...)

Alas, today it hanged completely 3 times :-(

When I say completely, I mean complete system hang, display frozen, keyboard 
dead (even the Magic SysRQ key doesn't respond), network dead (doen't ping 
back), no visible activity of any kind :-((

...and of course nothing got logged.

This happened 4 times. The 3 first times, I was running with IO-APIC enabled 
and the system was under high disk activity:

1/ 1st time, I was copying the complete contents of a loopback-mounted ISO 
image to a directory. The system hanged.

2/ 2nd time, I tried the same. The system hanged again.

3/ 3rd time, I was generating an ISO fs (mkisofs). The system hanged.

The 4th time, I was running the kernel booted with "noapic". The system was 
doing nothing and had gone to screen saver mode, then to screen off. When I 
moved the mouse nothing happened. I noticed the system had silently died :-(

Now I'm running with IO-APIC enabled, bus USB 2.0 adn ehci completely disabled 
(both in BIOS and modprobe.conf).

The system hasn't hanged again, but I haven't tried to play with on-disk ISO 
filesystems since...

Sigh. I never had any problem with this system for 2 years running kernel 2.4. 
I've had nothing but problems since running kernel 2.6 :-((

(Not to mention that EVMS snapshots don't work anymore in kernel 2.6 and badly 
corrupt filesystems [snapshots origins] spitting tons of :
totor kernel: attempt to access beyond end of device
totor kernel: dm-5: rw=0, want=990604112, limit=2031614
totor kernel: attempt to access beyond end of device
totor kernel: dm-5: rw=0, want=990604112, limit=2031614
...but this is yet another story...
I'm afraid I will end hating kernel 2.6 as much as I hate M$ things... )

Cheers.

-- 
Michel Bouissou <michel@bouissou.net> OpenPGP ID 0xDDE8AC6E

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd = IRQ trouble
  2005-07-26 19:39     ` Michel Bouissou
@ 2005-07-26 19:44       ` Michel Bouissou
  2005-07-27 17:59         ` Michel Bouissou
  2005-07-26 23:37       ` Grant Coady
  1 sibling, 1 reply; 7+ messages in thread
From: Michel Bouissou @ 2005-07-26 19:44 UTC (permalink / raw)
  To: Alan Stern; +Cc: Protasevich, Natalie, dbrownell, linux-kernel

Le Mardi 26 Juillet 2005 21:39, Michel Bouissou a écrit :
> Alas, today it hanged completely 3 times :-(
>
> When I say completely, I mean complete system hang, display frozen,
> keyboard dead (even the Magic SysRQ key doesn't respond), network dead
> (doen't ping back), no visible activity of any kind :-((
>
> ...and of course nothing got logged.
>
> This happened 4 times. The 3 first times, I was running with IO-APIC
> enabled and the system was under high disk activity:
>
> 1/ 1st time, I was copying the complete contents of a loopback-mounted ISO
> image to a directory. The system hanged.
>
> 2/ 2nd time, I tried the same. The system hanged again.
[...]
> Now I'm running with IO-APIC enabled, bus USB 2.0 adn ehci completely
> disabled (both in BIOS and modprobe.conf).
>
> The system hasn't hanged again, but I haven't tried to play with on-disk
> ISO filesystems since...

I just did the same as 1/ and 2/ again, with USB 2.0 disabled, to make my 
mind. It didn't hang (otherwise I would need to reboot before writing this 
message ;-)

-- 
Michel Bouissou <michel@bouissou.net> OpenPGP ID 0xDDE8AC6E

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd = IRQ trouble
  2005-07-26 19:39     ` Michel Bouissou
  2005-07-26 19:44       ` Michel Bouissou
@ 2005-07-26 23:37       ` Grant Coady
  1 sibling, 0 replies; 7+ messages in thread
From: Grant Coady @ 2005-07-26 23:37 UTC (permalink / raw)
  To: Michel Bouissou; +Cc: Alan Stern, Protasevich, Natalie, dbrownell, linux-kernel

On Tue, 26 Jul 2005 21:39:26 +0200, Michel Bouissou <michel@bouissou.net> wrote:
>
>Yes, but it doesn't tell us why kernels 2.4x felt perfectly happy with the old 
>BIOS...
You turned off 4k stacks?  I have a Via KM400 chipset box locked 
up a few times, once under 2.4.31-hf2 after 4.5 hours compiling 
kernels, fixed with some .config tuning.  Done 48+ hours continuous 
compiling kernels with 2.6.12.3 + 8k stacks, load ~2.2.  Just adding 
a datapoint...  Dunno if this relevant.  Simpler, older box with 
Intel 440BX chipset is happy with 4k stacks, only 2 USB ports.

Grant.


^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd = IRQ trouble
  2005-07-26 19:44       ` Michel Bouissou
@ 2005-07-27 17:59         ` Michel Bouissou
  0 siblings, 0 replies; 7+ messages in thread
From: Michel Bouissou @ 2005-07-27 17:59 UTC (permalink / raw)
  To: Alan Stern; +Cc: Protasevich, Natalie, dbrownell, linux-kernel

Le Mardi 26 Juillet 2005 21:44, Michel Bouissou a écrit :
>
> > Now I'm running with IO-APIC enabled, bus USB 2.0 and ehci completely
> > disabled (both in BIOS and modprobe.conf).
> >
> > The system hasn't hanged again, but I haven't tried to play with on-disk
> > ISO filesystems since...
>
> I just did the same as 1/ and 2/ again, with USB 2.0 disabled, to make my
> mind. It didn't hang (otherwise I would need to reboot before writing this
> message ;-)

Update: The system hasn't hanged anymore and has shown stable and fast since 
I've disabled USB 2.0.

Today I've stressed it quite a lot, played with ISOFS, loopback FS, performed 
disk-to-disk backups, burned CDs, used wine and dosemu etc. and it worked 
perfectly. None of the things that caused it to hang yesterday when USB 2.0 
was enabled caused any problem today.

So now it summarizes as :

- Old BIOS, kernel 2.4.x : System perfectly stable (with IO-APIC and USB 2.0)

- Old BIOS, kernel 2.6.12 : With IO-APIC and USB 2.0, "IRQ 21 - nobody cared" 
problems. Disabling either IO-APIC or USB 2.0 causes the problem to 
disappear.

- New BIOS, kernel 2.6.12 : With IO-APIC and USB 2.0, uhci and ehci are now on 
the same IRQ. System doesn't complain anymore about USB or IRQ lost, BUT it 
may completely *HANG* (is unstable) especially under high activity. Nothing 
gets logged or displayed as the system purely and simply hangs.
=> With USB 2.0 completely disabled, the problem seems to disappear and the 
system looks stable again.

Cheers.

-- 
Michel Bouissou <michel@bouissou.net> OpenPGP ID 0xDDE8AC6E

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2005-07-27 17:59 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2005-07-26  2:13 VIA KT400 + Kernel 2.6.12 + IO-APIC + uhci_hcd = IRQ trouble Protasevich, Natalie
2005-07-26 10:48 ` [SOLVED ?] VIA KT400 + Kernel 2.6.12 + IO-APIC + ehci_hcd " Michel Bouissou
2005-07-26 18:54   ` Alan Stern
2005-07-26 19:39     ` Michel Bouissou
2005-07-26 19:44       ` Michel Bouissou
2005-07-27 17:59         ` Michel Bouissou
2005-07-26 23:37       ` Grant Coady

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®