Re: yep, umass still broken

2003-09-29 Thread Doug White
What motherboards are you using with the OHCI controller?

John-Mark Gurney (jmg@) and I kicked this around at bsdcon but couldn't
reproduce it.  If its a common board or chipset we can probably get it and
work on the issue.  The zeroed block showed up in 8KB boundaries in the
instance we saw, which is asupicious since most OHCI controllers have a
DMA limitation of 1 page crossing.

I have a ppro with Intel OHCI which I haven't tested yet.

On Sat, 27 Sep 2003, Dave Truesdell wrote:

 -- Your message was:   (from Wesley Morgan)
 On Fri, 26 Sep 2003, Brian Fundakowski Feldman wrote:

  I can get fdisk to read the MBR, but when I try mdir, I get this trace back
  (of course, no crash dump because those haven't worked for me in a year):
  trap 0xc
  memcpy()
  ohci_softintr()
  usb_schedsoftintr()
  ohci_intr1()
  ohci_intr()
  ithread_loop()
 
  Anyone have any clued?  I'll include my dmesg, of course.

 It was unbroken for a while, but has been broken for at least a month
 (seem my earlier post about it). The umass driver has been a constant
 source of frustation for me and suffers from constant breakage and
 neglect.

 -- End of Message

 You may not want to blame the umass driver.  I've been doing a little
 experimenting trying to get a handle on what's going on.  Luckily I have two
 machines sitting side-by-side, one with OHCI and one with UHCI.  Many of the
 UMASS devices I have fail with 5.1-CURRENT on the OHCI machine but work just
 fine on the UHCI system.

 Here's the note I sent as a followup to kern/54982:
   I am encountering this problem as well.  What I've seen so far is this:

   1. The corruption does not occur with all UMASS devices.  For example, I
   see data corruption with a Creative Labs MUVO (128M) and NEXDISK (256M)
   devices, but not with an Easydisc (128M) device.

   2. I've only seen the corruption with OHCI based controllers.  When I
   connect the same device to a UHCI based machine, built from an identical
   copy of the source tree, I see no corruption.

   3. The pattern of corruption is decidely non-random.  If you view the
   file as a series of 4K blocks numbered 0 to N, the corruption I've seen
   follows the following pattern:
  (B == a zero filled 4k block)

   Original: 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 ...
   Corrupt:  0  3  2  B  4  7  6  B  8 11 10  B 12 15 14  B ...

   I can provide logs of a file copy done on both the OHCI and UHCI based
   systems done with hw.usb.debug=3 hw.usb.uhci.debug=6 hw.usb.ohci.debug=6
   and hw.usb.umass.debug=4294901760 if you wish.  They are far too long to
   attach here.

   This test was last run on 5.1-CURRENT cvsup'd on Sep 17th 2003.

 I'm presently updating both machines to -CURRENT cvsup'd this afternoon.

 I haven't gotten to the point where I understand the interactions between
 umass, ohci/uhci and cam well enough to even hazzard a guess about where the
 corruption is occuring.




 ___
 [EMAIL PROTECTED] mailing list
 http://lists.freebsd.org/mailman/listinfo/freebsd-current
 To unsubscribe, send any mail to [EMAIL PROTECTED]


-- 
Doug White|  FreeBSD: The Power to Serve
[EMAIL PROTECTED]  |  www.FreeBSD.org
___
[EMAIL PROTECTED] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: yep, umass still broken

2003-09-27 Thread Dave Truesdell
-- Your message was:   (from Wesley Morgan)
On Fri, 26 Sep 2003, Brian Fundakowski Feldman wrote:

 I can get fdisk to read the MBR, but when I try mdir, I get this trace back
 (of course, no crash dump because those haven't worked for me in a year):
 trap 0xc
 memcpy()
 ohci_softintr()
 usb_schedsoftintr()
 ohci_intr1()
 ohci_intr()
 ithread_loop()

 Anyone have any clued?  I'll include my dmesg, of course.

It was unbroken for a while, but has been broken for at least a month
(seem my earlier post about it). The umass driver has been a constant
source of frustation for me and suffers from constant breakage and
neglect.

-- End of Message

You may not want to blame the umass driver.  I've been doing a little 
experimenting trying to get a handle on what's going on.  Luckily I have two 
machines sitting side-by-side, one with OHCI and one with UHCI.  Many of the 
UMASS devices I have fail with 5.1-CURRENT on the OHCI machine but work just 
fine on the UHCI system.

Here's the note I sent as a followup to kern/54982:
  I am encountering this problem as well.  What I've seen so far is this:
 
  1. The corruption does not occur with all UMASS devices.  For example, I 
  see data corruption with a Creative Labs MUVO (128M) and NEXDISK (256M) 
  devices, but not with an Easydisc (128M) device.
 
  2. I've only seen the corruption with OHCI based controllers.  When I 
  connect the same device to a UHCI based machine, built from an identical 
  copy of the source tree, I see no corruption.
 
  3. The pattern of corruption is decidely non-random.  If you view the 
  file as a series of 4K blocks numbered 0 to N, the corruption I've seen 
  follows the following pattern:
 (B == a zero filled 4k block)
 
  Original: 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 ...
  Corrupt:  0  3  2  B  4  7  6  B  8 11 10  B 12 15 14  B ...
 
  I can provide logs of a file copy done on both the OHCI and UHCI based 
  systems done with hw.usb.debug=3 hw.usb.uhci.debug=6 hw.usb.ohci.debug=6 
  and hw.usb.umass.debug=4294901760 if you wish.  They are far too long to 
  attach here.
 
  This test was last run on 5.1-CURRENT cvsup'd on Sep 17th 2003.

I'm presently updating both machines to -CURRENT cvsup'd this afternoon.

I haven't gotten to the point where I understand the interactions between 
umass, ohci/uhci and cam well enough to even hazzard a guess about where the 
corruption is occuring.




___
[EMAIL PROTECTED] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: yep, umass still broken

2003-09-26 Thread tokza
On Friday 26 September 2003 09:18, Brian Fundakowski Feldman wrote:

 ohci_intr()
 ithread_loop()

 Anyone have any clued?  I'll include my dmesg, of course.

Yes, the same bad thing with umass on my sony vaio pcg-v505bx laptop. 
This is the only reason I have to keep slackware linux within to burn cd's and 
mount usb-drives on this laptop :) 

___
[EMAIL PROTECTED] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: yep, umass still broken

2003-09-26 Thread Wesley Morgan
On Fri, 26 Sep 2003, Brian Fundakowski Feldman wrote:

 I can get fdisk to read the MBR, but when I try mdir, I get this trace back
 (of course, no crash dump because those haven't worked for me in a year):
 trap 0xc
 memcpy()
 ohci_softintr()
 usb_schedsoftintr()
 ohci_intr1()
 ohci_intr()
 ithread_loop()

 Anyone have any clued?  I'll include my dmesg, of course.

It was unbroken for a while, but has been broken for at least a month
(seem my earlier post about it). The umass driver has been a constant
source of frustation for me and suffers from constant breakage and
neglect.

-- 
Hi! I'm a .signature virus! Copy me into your ~/.signature to help me spread!
___
[EMAIL PROTECTED] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to [EMAIL PROTECTED]