On Fri, Jul 23, 2004 at 02:59:20PM -0400, Alan Stern wrote:
> On Fri, 23 Jul 2004, Marc Haber wrote:
> > On Fri, Jul 23, 2004 at 11:55:54AM -0400, Alan Stern wrote:
> > > Yes, and the subsequent errors looked exactly the same. It's very
> > > consistent. Apparently your drive doesn't like to transfer data from
> > > sector 18743651. (I got that number by taking the start sector address
> > > 0x011e0120 from the second line in the log and adding 0x43 = 34304/512,
> > > the number of sectors successfully transferred when the error occurred.
> > > Subsequent attempts encountered the same error at the same sector number.
> > > This is just beyond the 9 GB mark.) Instead it's sending an invalid
> > > signal on the USB bus.
> >
> > It's a different sector number every time, and this is the third drive
> > that I have tried in that enclosure (the first two being rather recent
> > models made by Seagate, the one that you have seen the log from an
> > older IBM DTTA from 1999).
>
> Do you mean that it's a different sector number every time you run your
> test, or a different sector number every time you put a different disk in
> the enclosure, or something else?
The number reported in the "I/O error" line (with USB debugging turned
of) is a different one every time the test is run. Once the error is
shown, the USB subsystem seems to get into a bad state and all
subsequent requests fail. When I reboot and repeat the test, the error
repeats after different time spans (sometimes after a minute,
sometimes after multiple hours) with the same sector number. Once the
first error has appeared, the sector numbers of the subsequent errors
are all near the first one.
Subsequent tries with the two other disks show the first error at
different sector numbers, but always the same number for a given disk.
However, trying the same on another system, gives a different "same
number" for a given disk, and all three disks are fine if connected
directly to the local IDE bus. This makes me think that I do _not_
have a disk problem.
> In the log you sent there were multiple
> errors and they _all_ referred to the _same_ sector.
I cut the log after the first I/O error message. The next read command
is "28 00 01 1e 01 28 00 00 78 00" which fails again, this time after
30208.
> > > It's not clear whether the real source of the error is in the drive or in
> > > the enclosure.
> >
> > I am pretty sure it's not the drive (I have tried three), but can the
> > USB adapter in the system be a problem as well?
>
> It's possible. The USB adapter just might be sensitive to certain
> patterns of data, and the sector in question might contain one of the bad
> patterns.
This is the lspci output of my EHCI adapter:
0000:00:0c.2 USB Controller: VIA Technologies, Inc. USB 2.0 (rev 63) (prog-if 20
[EHCI])
Subsystem: VIA Technologies, Inc. USB 2.0
Flags: bus master, medium devsel, latency 32, IRQ 9
Memory at f4000000 (32-bit, non-prefetchable) [size=256]
Capabilities: [80] Power Management version 2
0000:00:0c.2 Class 0c03: 1106:3104 (rev 63) (prog-if 20)
Subsystem: 1106:3104
Flags: bus master, medium devsel, latency 32, IRQ 9
Memory at f4000000 (32-bit, non-prefetchable) [size=256]
Capabilities: [80] Power Management version 2
> There was a discussion a few months ago from someone who found
> that his USB-IDE adapter would fail every time he tried to transfer
> sectors containing a certain pattern of 3 or 4 bytes right at the sector
> boundary. See these email threads:
>
> http://marc.theaimsgroup.com/?l=linux-usb-devel&m=107962180123336&w=2
>
> and
>
> http://marc.theaimsgroup.com/?l=linux-usb-devel&m=108045000409630&w=2
That's a Genesys issue. We all know the Genesys adapters are
fundamentally broken. I have VIA in the host, and Cypress in the
enclosure.
Greetings
Marc
--
-----------------------------------------------------------------------------
Marc Haber | "I don't trust Computers. They | Mailadresse im Header
Karlsruhe, Germany | lose things." Winona Ryder | Fon: *49 721 966 32 15
Nordisch by Nature | How to make an American Quilt | Fax: *49 721 966 31 29
-------------------------------------------------------
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-users
