irq nobody cared issue on 2.6.2x
hi all,
i just moved my win tv dvb-s card (PCI) from my old to my actual pc.
its an ASUS M2N32 WS Professional AMD64 X2 Board equiped with the nvidia nForce 590 SLI MCP chipset.
in the past, i had to use the noapic kernel cmdline param to get linux booting and working properly.
iirc versions >=2.6.18 of the kernel fixed all of my previous problems, thus i am not using noapic since then.
if you would be so kind to provide me some infos,
how i would be able to track the problem down _and_ maybe how to fix it.
please find two snippets of dmesg after booting 2.6.21-rc3-git (2.6.20-1 has got the same results)
best regards
marcel
snip of logs
_normal_ boot on 2.6.21-rc3 git
Freeing unused kernel memory: 264k freed
saa7146: register extension 'dvb'.
ACPI: PCI Interrupt Link [APC1] enabled at IRQ 16
ACPI: PCI Interrupt 0000:02:06.0[A] -> Link [APC1] -> GSI 16 (level, low) -> IRQ 16
saa7146: found saa7146 @ mem ffffc20000070000 (revision 1, irq 16) (0x13c2,0x0003).
DVB: registering new adapter (Technotrend/Hauppauge WinTV Nexus-S rev2.X). adapter has MAC addr = 00:d0:5c:1f:6d:3f
irq 16: nobody cared (try booting with the "irqpoll" option)
Call Trace: <IRQ> [<ffffffff801473d7>] [<ffffffff801475e2>] [<ffffffff80147e97>] [<ffffffff8010a62c>] [<ffffffff8010c0e2>] [<ffffffff8010882e>] [<ffffffff80109a41>] <EOI> [<ffffffff80108857>] [<ffffffff801088bc>]
handlers:
[<ffffffff88015619>]
Disabling IRQ #16
dvb-ttpci: info @ card 0: firm f0240009, rtsl b0250018, vid 71010068, app 80f12623
dvb-ttpci: firmware @ card 0 supports CI link layer interface
boot on 2.6.21-rc3 git irqpoll option
Freeing unused kernel memory: 260k freed
saa7146: register extension 'dvb'.
ACPI: PCI Interrupt Link [APC1] enabled at IRQ 16
ACPI: PCI Interrupt 0000:02:06.0[A] -> Link [APC1] -> GSI 16 (level, low) -> IRQ 16
saa7146: found saa7146 @ mem ffffc20000026000 (revision 1, irq 16) (0x13c2,0x0003).
DVB: registering new adapter (Technotrend/Hauppauge WinTV Nexus-S rev2.X). adapter has MAC addr = 00:d0:5c:1f:6d:3f
dvb-ttpci: info @ card 0: firm f0240009, rtsl b0250018, vid 71010068, app 80f12623
dvb-ttpci: firmware @ card 0 supports CI link layer interface
dvb-ttpci: adac type set to 0 @ card 0saa7146_vv: saa7146 (0): registered device video0 [v4l2]saa7146_vv: saa7146 (0): registered device vbi0 [v4l2]
DVB: registering frontend 0 (ST STV0299 DVB-S)...
input: DVB on-card IR receiver as /class/input/input3
dvb-ttpci: found av7110-0.
XFS mounting filesystem sda1
later on....
Adding 1004052k swap on /dev/hda2. Priority:-1 extents:1 across:1004052k irq 16: nobody cared (try booting with the "irqpoll" option)
Call Trace: <IRQ> [<ffffffff80147763>] [<ffffffff80147982>] [<ffffffff801481a6>] [<ffffffff8010a7cc>] [<ffffffff8010c291>] [<ffffffff80108a3e>] [<ffffffff80109c51>] <EOI>
handlers:
[<ffffffff88015612>]
Disabling IRQ #16
Quoted 4 lines · Mws
if you would be so kind to provide me some infos,
how i would be able to track the problem down _and_ maybe how to fix it.
The first step is to figure out as exactly as possible _when_ it started happening.
Quoted 2 lines
please find two snippets of dmesg after booting 2.6.21-rc3-git (2.6.20-1 has got the same results)
So is 2.6.19 good? 2.6.20-rc1? -rc2?
If you're a git user (or even just marginally interested in learning), the best thing to do is to find some good kernel (the more recent, the better, but the keyword is *some* kernel that works well), and a bad kernel (the older the better, just to avoid unnecessary work, but again, you can just take the most recent too), and then just use "git bisect".
If you bisect it down to one particular commit, that will help us a lot, but even if you only bisect it down by booting a handful of kernels (say, 5-6), it will already have cut things down a lot, and probably more easily and efficiently than if you just downloaded daily snapshots and tried to narrow it down that way.
Linus
Quoted 7 lines · Linus Torvalds
Quoted 4 lines · Mws
if you would be so kind to provide me some infos,
how i would be able to track the problem down _and_ maybe how to fix it.
Quoted 7 lines
The first step is to figure out as exactly as possible _when_ it started happening.
Quoted 2 lines
please find two snippets of dmesg after booting 2.6.21-rc3-git (2.6.20-1 has got the same results)
So is 2.6.19 good? 2.6.20-rc1? -rc2?
went down to 2.6.17 - earlier is lacking due to correctly booting my machine.
Quoted 6 lines
If you're a git user (or even just marginally interested in learning), the best thing to do is to find some good kernel (the more recent, the better, but the keyword is *some* kernel that works well), and a bad kernel (the older the better, just to avoid unnecessary work, but again, you can just take the most recent too), and then just use "git bisect".
Quoted 7 lines
If you bisect it down to one particular commit, that will help us a lot, but even if you only bisect it down by booting a handful of kernels (say, 5-6), it will already have cut things down a lot, and probably more easily and efficiently than if you just downloaded daily snapshots and tried to narrow it down that way.
Linus
i started to bring in a load of printk's into v4l dvb repository to find out what exactly is causing the interrupt to appear and why it is not handled.
the bug is reproducible by insmod/rmmod dvb_ttpci over and over. yesterday late night i tried a bit too much and my machine freezed as a whole.
even kernel sysrq didn't work anymore.
i found some point, that after fireware load the interrupt is fired again.
will check if there is a solution to create a workaround and try to fix it afterwards.
it does NOT appear on a different computer same card. i think it is a problem in combination chipset - card - bios.... dunno by now, but will find out.
thanks for your information, linus.
marcel
Am Mittwoch, 7. März 2007 20:48 schrieb Mws:
Quoted 4 lines
hi all,
i just moved my win tv dvb-s card (PCI) from my old to my actual pc.
Quoted 8 lines
its an ASUS M2N32 WS Professional AMD64 X2 Board equiped with the nvidia nForce 590 SLI MCP chipset.
in the past, i had to use the noapic kernel cmdline param to get linux booting and working properly.
iirc versions >=2.6.18 of the kernel fixed all of my previous problems, thus i am not using noapic since then.
Hmm, I'm using the same board here since two weeks. openSUSE 10.2 (i586, kernel 2.6.18.{2,8}) didn't installed properly even with noapic (problems with eth device while installing from nfs server :-(, temporarily work arounded with an e1000). I found a BIOS update necessary for proper apic operation. No problems since then, but not using a dvb card in there either, since my vdr is a seperate system ;-)..
Cheers,
Pete