Re: Panic: EHCI and umass

On Sat, 3 Jul 2004, Lukas Ertl wrote:

> On Mon, 28 Jun 2004, Bryan Liesner wrote:
>
>> Fatal trap 12: page fault while in kernel mode
>> fault virtual address	= 0x53425355
>> fault code		= supervisor read, page not present
>> instruction pointer	= 0x8:0xc051c117
>> stack pointer	        = 0x10:0xd4294ab0
>> frame pointer	        = 0x10:0xd4294ad0
>> code segment		= base 0x0, limit 0xfffff, type 0x1b
>> 			= DPL 0, pres 1, def32 1, gran 1
>> processor eflags	= interrupt enabled, resume, IOPL = 0
>> current process		= 20 (irq10: pcm0 ehci0)
>> kernel: type 12 trap, code=0
>
> Since I've now gotten hold of a large USB2 disk, I tried to reproduce this 
> but successfully dumped 4GB of data, until I hit the msdosfs filesize limit.
>

Thank you for your efforts. I was getting around to sending this info 
back to the list, but I was up late last night and am only starting to wake up.
It seems to actually be an ATA problem of sorts.

Here's what I found:

I have (had) in my box a Maxtor UltraDMA 133 controller, which is 
actually a relabeled Promise PDC20269 UDMA133 controller. Attached to 
that is a Maxtor 6L060J3 60GB disk.

I was playing around and found that if I used atacontrol to drop it 
down to WDMA2, the dump completed without a problem, but slowly.

Doing simple speed tests on my Maxtor HD, like dd if=/dev/zero 
of=blahblah, I found that when set up as UDMA133, the throughput was 
only about 20MB/s.  I dropped it down to UDMA100, and the throughput 
increased.  At UDMA66, the throughput was even better and came out to 
be about 40MB/s  !!!!!

I removed the Maxtor/Promise card and attached my hard drive to the 
VIA 8233 UDMA100 controller on my motherboard, and retried my simple 
throughput test, which levels off at about 40MB/s.

The bottom line is, EHCI is and was just fine and dumps are now 
working fine.  ATA was screwed up for me until I changed the controller.

But the question remains - Is the Promise card I have damaged, 
braindead, not well supported, in conflict with what's on the 
motherboard, or is there another issue going on here that is difficult 
to find or detect?

During the course of this trouble, and for that matter, before the 
trouble started, I saw no complaints from the ATA driver about 
anything going wrong.  I had poor ATA performance going on here for 
quite a while.  The EHCI commit seems to have brought out the worst of 
it.

Thanks again, and I hope you didn't pull too much of your hair out :)

I have to search through the archives, but another guy had a similar 
issue after the EHCI update.  I wonder what his ATA configuration was. 
And what is yours, if you're using ATA?