Re: system lockup - ata spurious interrupts - somehow updated

From: Putinas Piliponis <putinas.piliponis_at_icnspot.net>
Date: Sun, 21 Mar 2004 14:37:00 +0200
I hope what sooner or later, you will have such hw. I wish I could provide
it, but I am not that rich.
One more question, with today kernel, vmstat -i doesn't list my second
sata controller as using irq at all.
I see what ata0 and ata1 are reporting using irq, but ata2 and ata3 is using 
only
port. Does it mean, what my hardware generates interrupt even if it's not 
asked
for ?

interrupt                          total       rate
irq1: atkbd0                         412          3
irq6: fdc0                             5          0
irq8: rtc                          13331        125
irq13: npx0                            1          0
irq15: ata1                           35          0
irq16: uhci0 uhci3                     2          0
irq18: uhci2+                       3353         31
irq20: pcm0                            1          0
irq23: fxp0                           34          0
irq0: clk                          10415         98
Total                              27589        260

and the dmesg part with ata:
atapci0: <Intel ICH5 UDMA100 controller> port 
0xfc00-0xfc0f,0-0x3,0-0x7,0-0x3,0-0x7 at device 31.1 on pci0
ata0: at 0x1f0 irq 14 on atapci0
ata0: [MPSAFE]
ata1: at 0x170 irq 15 on atapci0
ata1: [MPSAFE]
atapci1: <Intel ICH5 SATA150 controller> port 
0xef90-0xef9f,0xefa8-0xefab,0xefa0-0xefa7,0xefac-0xefaf,0xefe0-0xefe7 irq 18 
at device 31.2 on pci0
atapci1: [MPSAFE]
ata2: at 0xefe0 on atapci1
ata2: [MPSAFE]
ata3: at 0xefa0 on atapci1
ata3: [MPSAFE]
acd0: CDRW <HL-DT-ST GCE-8481B> at ata1-master PIO4
ad4: 245416MB <HDS722525VLSA80> [498625/16/63] at ata2-master UDMA100
ad6: 245416MB <HDS722525VLSA80> [498625/16/63] at ata3-master UDMA100

----- Original Message ----- 
From: "Søren Schmidt" <sos_at_DeepCore.dk>
To: "Putinas Piliponis" <putinas.piliponis_at_icnspot.net>
Cc: <current_at_freebsd.org>
Sent: Sunday, March 21, 2004 1:33 PM
Subject: Re: system lockup - ata spurious interrupts - somehow updated


> Putinas Piliponis wrote:
>> As for me, it looks what ata code is not correctly handling more then one
>> ata controller. If I set in bios compatible mode, and I have only P-ATA+
>> S-ATA, everything works fine, I can boot in verbose mode, no spurious
>> interrupts at all. If I set in bios as bios default Enchanced mode for 
>> S-ATA
>> then I still can boot in verbose mode, I see plenty of spurious 
>> interrupts
>> on ata2 and ata3 controllers, but system is still continues booting. If I 
>> set
>> enchanced
>> mode for S-ATA + P-ATA I cannot boot either in verbose mode or either in
>> normal mode. System gets stuck forever ( or at least for long time ) 
>> after I
>> see:
>> "Mounting root from ufs:ad8s3a".
>> and if I boot then in verbose mode, I see nonstopable spurious interrupts 
>> on
>> ata2, ata3, ata4 and ata5.
>>
>> This is also could explain, why 5.2.1 is still stable with 4 ide drives,
>> and continuesly freezing or panicing with 6 ( or more I guess ) ide 
>> drives.
>> and as well why for some people configuration is working, either they use
>> 4 drives not on all different channels, but some of them maybe on same
>> channel ( again my guess ).
>
> ATA works just fine with any number of controllers (however your HW may 
> not), my main test box has 16 channels in it and that still works :)
>
> Now, the above is more likely an interrupt setup/routing problem of sorts, 
> if you get spurious interrupts *something* is yanking the interrupt line, 
> but ATA knows that it hasn't requested anything and spits out the 
> "spurious" warning.
>
> That said, the ICH5 support is written "blindfolded" as I do not have any 
> such HW here in the lab. So if there is a problem with ATA on those, 
> someone with the HW and enough kernel clue should look into it (or someone 
> could land the needed HW in my lab, which could also bring support for the 
> SW RAID on intel's)...
>
> -- 
> -Søren
>
>
> 
Received on Sun Mar 21 2004 - 03:37:06 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:48 UTC