Topic: Firewire transfer challenges.... Vista bad?

[it-goes-to-eleven]

TS member manitouman recently reported some problems with noisy wavefiles from his new 7xx recorder.  After some testing he found the problem was a bad firewire cable.  Trouble is, that isn't supposed to happen with firewire because the transfers are validated via checksum.  I had always assumed/hoped that any errors during transfer would be detected but I really wasn't sure...  Silent data corruption is Always a possiblity...

He offered to send me the cable (thanks manitouman!) and I ran some tests with my 722.  Transferring under Linux, I almost immediately encountered serious errors at the command line and also logged to the console.  A very bad cable indeed.

Under XP, repeated transfers resulted in no errors being reported. The files were transferred and had the correct final md5.  Nice to see XP retrying but I wish there was an easy way to see the retry stats after every transfer.

I believe the Linux driver aborts on error.  Apparently *my* XP driver retries when it receives bad data. I emphasize *my* XP driver because your driver may be different.

It turns out the original silent corruption happened under Windows Vista Ultimate Edition(!!)... So take that fwiw. I'd also be concerned that system might not detect usb transfer errors.

I wish we had an easy way for everyone to verify that their firewire and usb drivers detect bad data... But short of sending the cable around, I can't think of one. I wouldn't advise wiggling the cable in the socket.


Errors observed under Linux during test:

# cp -v * /m4/tmp/fwtest/
`t1168.wav' -> `/m4/tmp/fwtest/t1168.wav'

# time cp -v * /m4/tmp/fwtest/
`t1168.wav' -> `/m4/tmp/fwtest/t1168.wav'
cp: reading `t1168.wav': Input/output error
`t1169.wav' -> `/m4/tmp/fwtest/t1169.wav'
cp: reading `t1169.wav': Input/output error
`t1170.wav' -> `/m4/tmp/fwtest/t1170.wav'
cp: reading `t1170.wav': Input/output error
0.008u 0.784s 1:40.40 0.7%      0+0k 0+0io 0pf+0w

Console errors:

Feb 27 18:05:17  kernel: sda: Current: sense key=0x0
Feb 27 18:05:17  kernel:     ASC=0x0 ASCQ=0x0
Feb 27 18:05:47  kernel: ieee1394: sbp2: aborting sbp2 command
Feb 27 18:05:47  kernel: sd 11:0:0:0:
Feb 27 18:05:47  kernel:         command: cdb[0]=0x28: 28 00 01 ca 1b 93 00 00 08 00
Feb 27 18:05:57  kernel: ieee1394: sbp2: aborting sbp2 command
Feb 27 18:05:57  kernel: sd 11:0:0:0:
Feb 27 18:05:57  kernel:         command: cdb[0]=0x0: 00 00 00 00 00 00
Feb 27 18:05:57  kernel: ieee1394: sbp2: reset requested
Feb 27 18:05:57  kernel: ieee1394: sbp2: Generating sbp2 fetch agent reset
Feb 27 18:06:07  kernel: ieee1394: sbp2: aborting sbp2 command
Feb 27 18:06:07  kernel: sd 11:0:0:0:
Feb 27 18:06:07  kernel:         command: cdb[0]=0x0: 00 00 00 00 00 00
Feb 27 18:06:07  kernel: sd 11:0:0:0: scsi: Device offlined - not ready after error recovery
Feb 27 18:06:07  kernel: sd 11:0:0:0: SCSI error: return code = 0x50000
Feb 27 18:06:07  kernel: end_request: I/O error, dev sda, sector 30022547
Feb 27 18:06:07  kernel: sd 11:0:0:0: rejecting I/O to offline device
Feb 27 18:06:07  last message repeated 6 times
Feb 27 18:06:07  kernel: sd 11:0:0:0: SCSI error: return code = 0x10000
Feb 27 18:06:07  kernel: end_request: I/O error, dev sda, sector 30022555
Feb 27 18:06:07  kernel: sd 11:0:0:0: rejecting I/O to offline device

[boojum]

MS takes a year or so to beta test their operating systems in a world-wide beta test, AKA a release.  Operating systems are pretty complex but there is no excuse for what MS gets away with.  I think it is sloppy work.  I am sticking with my XP box for another year or so.

[manitouman]

Thanks for the hard work, Freelunch.

I must have gotten lucky (knock on wood) with my system running Vista. I built the machine myself and installed a full version of it. Really never had problems with it. I still have another machine running XP but failed to try and transfer with that machine due to I'd have to install editing software to see the wave files. It's the wife's computer and I didn't want all that clutter on her machine.

I didn't know that file transfers are supposed to be checked before transferring a bad/corrupt file. I just thought they just transferred. When I used the USB to transfer with the MT, using XP, the files just copied over. Same thing with Vista. Every now and then the file won't transfer and I'll get an error message. Then I try again and it will either error again or it will just transfer. I don't know?

[it-goes-to-eleven]

A bit more info on my firewire hardware and kernel:

0000:02:0c.0 FireWire (IEEE 1394): VIA Technologies, Inc. IEEE 1394 Host Controller (rev 46) (prog-if 10 [OHCI])
        Subsystem: Unknown device 0574:086c
        Flags: bus master, medium devsel, latency 32, IRQ 5
        Memory at fc004000 (32-bit, non-prefetchable) [size=2K]
        I/O ports at 9800 [size=128]
        Capabilities: [50] Power Management version 2

Kernel version:

2.6.17.1 #3 PREEMPT Wed Jun 28 13:04:45 EDT 2006 x86_64 AMD Athlon(tm) 64 Processor 3000+ AuthenticAMD GNU/Linux