Re: r255873: ahcich7: Timeout on slot 0 port 0 FLOODING kernel messages

On Sep 27, 2013, at 7:38 AM, Boris Samorodov <bsam_at_passap.ru> wrote:

> 27.09.2013 16:59, Pietro Cerutti пишет:
>> On 2013-Sep-27, 05:57, Scott Long wrote:
>>> 
>>> On Sep 27, 2013, at 5:52 AM, Sergey V. Dyatko <sergey.dyatko_at_gmail.com> wrote:
>>> 
>>>>>> 
>>>>>> yes, these messages disappeared after revert 255870 and r255871
>>>>>> 
>>>>> 
>>>>> 
>>>>> Hi,
>>>>> 
>>>>> Nothing that I changes should have affected the ahci driver.  In
>>>>> fact, I tested this driver specifically during my development.  Can
>>>>> you confirm that you built both kernel and modules from a clean state?
>>>> 
>>>> [tiger_at_laptop]:/usr/src%svn st
>>>> M       sys/boot/i386/zfsboot/zfsboot.c
>>>> [tiger_at_laptop]:/usr/src%svn diff 
>>>> Index: sys/boot/i386/zfsboot/zfsboot.c
>>>> ===================================================================
>>>> --- sys/boot/i386/zfsboot/zfsboot.c     (revision 255873)
>>>> +++ sys/boot/i386/zfsboot/zfsboot.c     (working copy)
>>>> _at__at_ -302,6 +302,7 _at__at_
>>>>    * region in the SMAP, use the last 3MB of 'extended' memory as a
>>>>    * high heap candidate.
>>>>    */
>>>> +       high_heap_size = 0;
>>>>   if (bios_extmem >= HEAP_MIN && high_heap_size < HEAP_MIN) {
>>>>      high_heap_size = HEAP_MIN;
>>>>      high_heap_base = bios_extmem + 0x100000 - HEAP_MIN;
>>>> [tiger_at_laptop]:/usr/src%
>>>> 
>>>> nothing more. 
>>>> 
>>> 
>>> Yes, no local source changes, but are you loading the cam or ahci drivers
>>> a loadable modules that might be out of sync now, or did you use a short-
>>> cut for building your kernel, like NO_CLEAN?
>> 
>> I had the same problem with a clean src tree. After commenting out the
>> ahci line from my kernel conf file, I got these instead:
>> 
>> Sep 27 14:32:10 SN2000 kernel: ata3: setting up DMA failed
>> Sep 27 14:32:12 SN2000 kernel: ata3: FAILURE - zero length DMA transfer attempte
>> 
>> reverting those two commits solved the issue.
> 
> In my case just rebuilding and restarting of sysutils/hal helped.


Interesting.  I didn't test hal, probably should have.  The compat shims I put in
place should have made it work without a rebuild.  I wonder if it's somehow
corrupting kernel state.  Are others who are experiencing problems also running
the hal package?

Scott