Regression between 9-RELEASE and 9-STABLE [PCIB ?]

[Harry Newton]

Adrian - thanks !

John, if you write me a shopping list of what information you think
useful, assuming you have time and inclination, I'll do my best
provide it.


On 28 February 2012 23:04, Adrian Chadd <adrian at freebsd.org> wrote:
> A lot of the PCI bus enumeration and configuration code has changed.
>
> John's likely the best person to bring into this. He's been doing a
> lot of the PCI bus hacking and has found/fixed quite a few
> regressions.
>
> Good luck!
>
>
>
> Adrian
>
>
> On 28 February 2012 14:26, Harry Newton <hn at yewbarrow.net> wrote:
>> Not really. Verbose suggests it completes the scan of the PCI to LPC
>> Bridge and PCI to PCI bridge. Comparing with a working verbose boot,
>> the next stage, which isn't reached, is HyperTransport Technology
>> Configuration.
>>
>> Trying BUS_DEBUG, I got a lot of this:
>>
>> device_add_child_ordered: 1813: (null) at pci with orer 0 as unit -1
>> make_device: 1698: (null) at pci as unit -1
>>
>> I neglected to say before: the machine's been running FreeBSD 7-STABLE
>> quite happily with no problems.
>>
>> - Harry
>>
>>
>> On 28 February 2012 20:08, Ian Lepore <freebsd at damnhippie.dyndns.org> wrote:
>>> On Tue, 2012-02-28 at 19:15 +0000, Harry Newton wrote:
>>>> 9-RELEASE works fine. csup to 9-STABLE and make buildworld kernel etc
>>>> gives a system the hangs during the boot process. There are no error
>>>> messages during the part boot, and no panic. Last messages before hang
>>>> are:
>>>>
>>>> pcib0: <ACPI Host-PCI bridge> port 0xcf8-0xcff on acpi0
>>>> pci0: <ACPI PCI bus> on pcib0
>>>>
>>>> I am stumped about how to diagnose this, and would be _very_ grateful
>>>> for any suggestions.
>>>>
>>>> When it's hung I can't force a break to debugger (BREAK_TO_DEBUGGER)
>>>> in kern conf: I imagine I'm too early in the boot sequence.
>>>>
>>>> - Harry
>>>>
>>>
>>> I assume booting verbose doesn't provide any more clues?  I've sometimes
>>> gotten an extra clue from a hang during boot by building with option
>>> BUS_DEBUG.  Even if you don't suspect the newbus routines as directly
>>> being the problem, sometimes you get a bit more info about what was
>>> happening at the time of lockup.
>>>
>>> -- Ian
>>>
>>>
>>
>>
>>
>> --
>> Harry Newton
>> _______________________________________________
>> freebsd-stable at freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-stable
>> To unsubscribe, send any mail to "freebsd-stable-unsubscribe at freebsd.org"



-- 
Harry Newton