Re: PCI IRQ Pins

[Sergei Shtylyov <sshtylyov@ru.mvista.com>]

Hello.

Karl Hiramoto wrote:

>>>>>>Normally you get a backtrace when a "nobody cared" message is
>>>>>>issued -
>>>>>>this should tell you which driver is probably the cause.

>>>>>Right - or that the other device is the cause (stuck IRQ line). So,
>>>>>Karl, please just post the backtrace.

>>>>Krzysztof, you mentioned clearing the IRQ in the platform code, is
>>>>there an example of this somewhere?

>>>>There is a Compact flash on hda connected the the HPT371N, looking
>>>>at the IDE code it looks like the drive my not be ready, or the
>>>>drive may raise the IRQ..

>>>>As soon as request_irq is called, the IRQ happens.

>>>>CCing linux-IDE now, as it may be an issue with this driver.

>>>>Backtrace below, sorry about some of the lines being wrapped.

>>>I think i see the problem:

>>>In the platform code, i should save the frequency of 33 Mhz in the
>>>correct register.

>>You mean the PCI frequency?

>>>>hpt366: HPT371N chipset detected
>>>>hpt366 0000:00:01.0: IDE controller (0x1103:0x0007 rev 0x02)
>>>>PCI: enabling device 0000:00:01.0 (0140 -> 0141)
>>>>hpt366 0000:00:01.0: IDE port disabled
>>>>hpt366 0000:00:01.0: no clock data saved by BIOS
>>>>hpt366 0000:00:01.0: DPLL base: 77 MHz, f_CNT: 120, assuming 50 MHz PCI

>>Hum, interesting... is your PCI indeed running at a frequency close to
>>50 MHz?

    To be precise, the formula yields 48 MHz PCI clock for f_CNT of 120. It 
then is clamped to 50 MHz (which I'm not quite sure is a good idea).

> No, it's being miscalculated by the hpt366 driver.

    Hum... it's hard to miscalculate something when you average f_CNT value 
over 128 register reads. :-)

> I think it should be 33 Mhz.

    Probably. And there's a chance that your PCI is clocked incorrectly.

> At least i have not yet seen this IRQ "nobody cared" message
> when i set it to 33Mhz in the hpt366 driver.

    Hum, interesting, interesting...

> Still doing more tests to see if it comes up again, but it's hard to
> reproduce.

    Is it readily reproducible with unmodified driver?

MBR, Sergei