Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Len Brown]

On Fri, 2004-05-07 at 15:41, Bob Gill wrote:
> OK, great!  Adding acpi=noirq to the kernel line made the lost
> interrupt problem go away.

Bob, Alex,
(or anybody else with a SIS-961 that now requires acpi=noirq),

I need some info to find out why your system recently broke.

Please open a bug here and attach the info,
http://bugzilla.kernel.org/enter_bug.cgi?product=ACPI
or just e-mail it to me and I'll open a bug for you.

Need the complete dmesg and /proc/interrupt from the most recent ACPI
enabled kernel that worked properly -- I guess -bk6 worked okay?

Any chance you can boot with "debug" and capture the console messages
from the failure?  If no, then the complete dmesg of the latest kernel
with "acpi=noirq" is the next best thing.

output from lspci -vv

output acpidmp available in /usr/sbin/, or in pmtools:
http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/utils/

thanks,
-Len

ps. would also be good to verify you're running an up-to-date BIOS.

pps. taking a wild guess, can you try backing out this patch?

#   ACPI: No IRQ known ... - using IRQ 255 (Bjarni Rúnar Einarsson)
#   http://bugzilla.kernel.org/show_bug.cgi?id=2148

http://linux.bkbits.net:8080/linux-2.5/gnupatch@408a06a6JHD43KPCLW3tDIYGowoxvg

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Alex Riesen]

Len Brown, Sun, May 09, 2004 04:56:07 +0200:
> On Fri, 2004-05-07 at 15:41, Bob Gill wrote:
> > OK, great!  Adding acpi=noirq to the kernel line made the lost
> > interrupt problem go away.
> 
> Bob, Alex,
> (or anybody else with a SIS-961 that now requires acpi=noirq),
> 
> I need some info to find out why your system recently broke.
> 

will do as soon as I get access to the box again. Monday, that is

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Len Brown]

[-- Attachment #1: Type: text/plain, Size: 1042 bytes --]

Bob,
thanks for the info.
The BIOS on this box has a bug where it is reporting a current
IRQ to be outside the list of possible IRQs:

ACPI: PCI Interrupt Link [LNKB] (IRQs 3 4 5 6 7 10 11 12 14 15) *9

It then references this with pinA of device 9:

Package (0x04) { 0x0009FFFF, 0x00, \_SB.PCI0.LNKB, 0x00 },

which is
00:09.0 Multimedia audio controller: Creative Labs SB Live! EMU10k1 (rev
07)

In the past, we'd enable this on IRQ9, even thought it is illegal.

ACPI: PCI Interrupt Link [LNKB] enabled at IRQ 9

But we found that broke some boxes.

So, now we choose an IRQ from the possible list, preferring
the highest interrupt number in the list -- 15.
Didn't see it in your .JPG dmesg, but I expect this was there:

ACPI: PCI Interrupt Link [LNKB] enabled at IRQ 15

and probably that set IRQ15 to level/low which killed IDE.

Please try the attached patch which disables the sanity
check above.

Also might be interesting to see what happens on this system
if it is booted (without the patch) with "acpi_irq_balance"

thanks,
-Len


[-- Attachment #2: sis-debug.patch --]
[-- Type: text/plain, Size: 435 bytes --]

===== drivers/acpi/pci_link.c 1.28 vs edited =====
--- 1.28/drivers/acpi/pci_link.c	Thu May  6 16:03:17 2004
+++ edited/drivers/acpi/pci_link.c	Sun May  9 21:57:39 2004
@@ -549,7 +549,7 @@
 	/*
 	 * if active found, use it; else pick entry from end of possible list.
 	 */
-	if (i != link->irq.possible_count) {
+	if (link->irq.active) {
 		irq = link->irq.active;
 	} else {
 		irq = link->irq.possible[link->irq.possible_count - 1];

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Alex Riesen]

Len Brown, Sun, May 09, 2004 04:56:07 +0200:
> Please open a bug here and attach the info,
> http://bugzilla.kernel.org/enter_bug.cgi?product=ACPI
> or just e-mail it to me and I'll open a bug for you.

Done. It is #2665.

> Need the complete dmesg and /proc/interrupt from the most recent ACPI
> enabled kernel that worked properly -- I guess -bk6 worked okay?

yes. Also bk7 was ok.

> Any chance you can boot with "debug" and capture the console messages
> from the failure?  If no, then the complete dmesg of the latest kernel
> with "acpi=noirq" is the next best thing.

this is already attached to the bug report. I going to reboot now,
to check the bios, and set the "debug".

> #   ACPI: No IRQ known ... - using IRQ 255 (Bjarni Rúnar Einarsson)
> #   http://bugzilla.kernel.org/show_bug.cgi?id=2148
> http://linux.bkbits.net:8080/linux-2.5/gnupatch@408a06a6JHD43KPCLW3tDIYGowoxvg

compiling now.

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Alex Riesen]

Len Brown, Sun, May 09, 2004 04:56:07 +0200:
> I need some info to find out why your system recently broke.
...
> ps. would also be good to verify you're running an up-to-date BIOS.

I think I am.

> pps. taking a wild guess, can you try backing out this patch?

cannot be it. There is a message before the changed line (a warning
regarding no irq), which is never seen in the logs.

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Len Brown]

[-- Attachment #1: Type: text/plain, Size: 855 bytes --]

On Sun, 2004-05-09 at 22:16, Brown, Len wrote:
> ACPI: PCI Interrupt Link [LNKB] (IRQs 3 4 5 6 7 10 11 12 14 15) *9
...
> Didn't see it in your .JPG dmesg, but I expect this was there:
> 
> ACPI: PCI Interrupt Link [LNKB] enabled at IRQ 15
> 
> and probably that set IRQ15 to level/low which killed IDE.
> 
> Please try the attached patch which disables the sanity
> check above.
> 
> Also might be interesting to see what happens on this system
> if it is booted (without the patch) with "acpi_irq_balance"

Better yet, here's a proposd fix.  Please give it a whirl.

The curve ball was that "illegal" IRQ9.
Its "illegalness" caused us to grab a new IRQ off the possible list,
but since it wasn't zero, we didn't scrub the new IRQ with our
usual heuristics for selecting ISA range IRQs.

I expect this will assign sound to IRQ10 or IRQ11.

thanks,
-Len


[-- Attachment #2: pci_link.patch --]
[-- Type: text/plain, Size: 1301 bytes --]

===== drivers/acpi/pci_link.c 1.28 vs edited =====
--- 1.28/drivers/acpi/pci_link.c	Thu May  6 16:03:17 2004
+++ edited/drivers/acpi/pci_link.c	Sun May  9 23:16:48 2004
@@ -478,7 +478,7 @@
 	PIRQ_PENALTY_PCI_AVAILABLE,	/* IRQ9  PCI, often acpi */
 	PIRQ_PENALTY_PCI_AVAILABLE,	/* IRQ10 PCI */
 	PIRQ_PENALTY_PCI_AVAILABLE,	/* IRQ11 PCI */
-	PIRQ_PENALTY_ISA_TYPICAL,	/* IRQ12 mouse */
+	PIRQ_PENALTY_ISA_USED,	/* IRQ12 mouse */
 	PIRQ_PENALTY_ISA_USED,	/* IRQ13 fpe, sometimes */
 	PIRQ_PENALTY_ISA_USED,	/* IRQ14 ide0 */
 	PIRQ_PENALTY_ISA_USED,	/* IRQ15 ide1 */
@@ -545,17 +545,23 @@
 		if (link->irq.active == link->irq.possible[i])
 			break;
 	}
+	/*
+	 * forget active IRQ that is not in possible list
+	 */
+	if (i == link->irq.possible_count) {
+		if (acpi_strict)
+			printk(KERN_WARNING PREFIX "_CRS %d not found"
+				" in _PRS\n", link->irq.active);
+		link->irq.active = 0;
+	}
 
 	/*
 	 * if active found, use it; else pick entry from end of possible list.
 	 */
-	if (i != link->irq.possible_count) {
+	if (link->irq.active) {
 		irq = link->irq.active;
 	} else {
 		irq = link->irq.possible[link->irq.possible_count - 1];
-		if (acpi_strict)
-			printk(KERN_WARNING PREFIX "_CRS %d not found"
-				" in _PRS\n", link->irq.active);
 	}
 
 	if (acpi_irq_balance || !link->irq.active) {

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Alex Riesen]

> Better yet, here's a proposd fix.  Please give it a whirl.

worked for me (with 2.6.6).

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out with 2.6.6-rc3-bk8

[Bob Gill]

>zero IDE changes in bk6 -> bk8 but a lot of ACPI / IRQ related

OK.  My APIC is a SiS961.  
I can't send data from 2.6.6-rc3-bk8, but from 2.6.6-rc3-bk6,
/proc/acpi gives:
(/proc/acpi/processor/CPU0)
processor id:            0
acpi id:                 0
bus mastering control:   yes
power management:        no
throttling control:      no
limit interface:         no
<not supported>
active state:            C1
default state:           C1
bus master activity:     00000000
states:
   *C1:                  promotion[--] demotion[--] latency[000]
usage[00000000]
    C2:                  <not supported>
    C3:                  <not supported>
<not supported>

cat /proc/acpi/sleep shows:
S0 S1 S4 S5
cat /proc/ide/sis shows:
SiS 5513 Ultra 100 chipset
--------------- Primary Channel ---------------- Secondary Channel
-------------
Channel Status: On                               On
Operation Mode: Compatible                       Compatible
Cable Type:     80 pins                          80 pins
Prefetch Count: 512                              512
Drive 0:        Postwrite Enabled                Postwrite Disabled
                Prefetch  Enabled                Prefetch  Disabled
                UDMA Enabled                     UDMA Enabled
                UDMA Cycle Time    3 CLK         UDMA Cycle Time    6
CLK
                Data Active Time   3 PCICLK      Data Active Time   3
PCICLK
                Data Recovery Time 1 PCICLK      Data Recovery Time 1
PCICLK
Drive 1:        Postwrite Enabled                Postwrite Disabled
                Prefetch  Enabled                Prefetch  Disabled
                UDMA Enabled                     UDMA Disabled
                UDMA Cycle Time    2 CLK         UDMA Cycle Time   
Reserved
                Data Active Time   3 PCICLK      Data Active Time   3
PCICLK
                Data Recovery Time 1 PCICLK      Data Recovery Time 3
PCICLK


... I don't know if any of this is helpful.  The IRQ and IO address
assignments (sent with my first email) are from 2.6.6-rc3-bk6.  

Thanks,
Bob

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Alex Riesen]

>> zero IDE changes in bk6 -> bk8 but a lot of ACPI / IRQ related

it's bk8 which broke sis961 (or revealed it native brokenness)

> OK.  My APIC is a SiS961.

I have the same problem (and the same chipset).
Passing acpi=noirq helps to work it around.

Re: hdc: lost interrupt ide-cd: cmd 0x3 timed out ...

[Bob Gill]

OK, great!  Adding acpi=noirq to the kernel line made the lost interrupt
problem go away, and getting rid of rhgb on the same line made the
problems I had with nvidia drivers killing the system at the login
prompt (white screen, no kbd response) go away.
One goes in, one goes out! ;)

Thanks,
Bob  

On Fri, 2004-05-07 at 05:07, Alex Riesen wrote:
> >> zero IDE changes in bk6 -> bk8 but a lot of ACPI / IRQ related
> 
> it's bk8 which broke sis961 (or revealed it native brokenness)
> 
> > OK.  My APIC is a SiS961.
> 
> I have the same problem (and the same chipset).
> Passing acpi=noirq helps to work it around.
>