Testing CAM wrapper for ata(4) controller drivers

Tom Uffner tom at uffner.com
Tue Dec 15 16:51:15 PST 2009

Alexander Motin wrote:
> Tom Uffner wrote:
>> this change breaks my Promise SATA card
> Are you sure that it was this change?
>> atapci0: <Promise PDC20771 SATA300 controller> port
>> 0xdc00-0xdc7f,0xd800-0xd8ff mem
>> 0xfeaff000-0xfeafffff,0xfeac0000-0xfeadffff irq 23 at device 11.0 on pci2
>> kernels from after the commit on 12/6 are unable to mount my drives.
>> i get (from memory, since i can't even get to single user mode)
>> followed by lots of SETFEATURE foo errors then a READ_DMA timeout
>> followed by a bus reset and the cycle repeats...
>> before this commit it worked fine.
> How long ago?

kernel built from cvs update to D2009. boots & works
fine. kernels from cvs updates after the dev/ata commit on 12/6 can
not mount my ufs partitions in either legacy or w/ options ATA_CAM

>> this is in the legacy mode. i am building a kernel with "options ATA_CAM"
>> now
> OK. Tell me if there any difference.
> Send me please full _verbose_ dmesg with original kernel and if possible
> with new. If you can't get full with new, enable verbose messages and
> make screenshots of drives detection and first errors.

here is verbose dmesg from kernel built w/ sources updated to 12/5/2009 at
midnight EST:  http://xiombarg.uffner.com/dmesg_20091205.txt

as soon as i can set up another PC & serial console i'll get you a verbose
boot from new kernel.


More information about the freebsd-current mailing list