USB HDD trouble

**demian** · 08-18-2003, 06:20 PM

Hey everyone.

I got myself an external HDD case that supports USB 2.0 and Firewire. Currently I'm using the USB option and I've got some trouble. The drive is jumpered as master and all the usb modules and the sd_mod load without problem. I can use the drive for a few minutes as you would expect and then it starts to hang and my syslog records all sorts of I/O errors.

This is definitely not because of a bad drive. I tried two different ones and both work fine when they are connected to the IDE bus. I experience the same bahaviour when I use my system at home running linux 2.4.21-ck3 and at work (SuSE 8.2 linux-2.4.20-4GB-athlon). The filesystem I currently use on the external drive is ext3.

Here's a bit of my /var/log/messages:

Code:

Aug 18 20:13:50 dirac kernel:  I/O error: dev 08:01, sector 3936320
Aug 18 20:13:50 dirac last message repeated 37 times
Aug 18 20:13:50 dirac kernel:  I/O error: dev 08:01, sector 7602184
Aug 18 20:13:50 dirac kernel: EXT3-fs error (device sd(8,1)): read_inode_bitmap: Cannot read inode bitmap - block_group = 29, inode_bitmap = 950273
Aug 18 20:13:50 dirac kernel:  I/O error: dev 08:01, sector 0
Aug 18 20:13:50 dirac kernel: EXT3-fs error (device sd(8,1)) in ext3_new_inode:
IO failure

Any success stories out there for external usb harddrives? What fs do you use? What mount options? What kernel?

Thanks,
demian

**demian** · 08-18-2003, 07:49 PM

I did some more trouble shooting and turned on verbose debugging for the usb-storage part. Turns out something times out and then the reset of the scsi disk doesn't work. (No, I did _not_ disconnect the drive.)

Code:

Aug 18 21:23:02 dirac kernel: usb_control/bulk_msg: timeout
Aug 18 21:23:12 dirac last message repeated 2 times
Aug 18 21:23:13 dirac kernel: usb.c: USB disconnect on device 00:09.2-2 address 2
Aug 18 21:23:14 dirac kernel: hub.c: new USB device 00:09.2-2, assigned address 3
Aug 18 21:23:14 dirac kernel: Product: USB TO IDE
Aug 18 21:23:28 dirac kernel: scsi: device set offline - command error recover failed: host 1 channel 0 id 0 lun 0
Aug 18 21:23:28 dirac kernel: SCSI disk error : host 1 channel 0 id 0 lun 0 return code = 6070000

This happens when I use the ehci-hcd driver (USB 2.0) btw. The uhci (USB 1.1) seems to work fine it's running stable for almost an hour now. Of course, it's no fun to try to fill up a 120GB drive when throughput rarely exceeds 1MB/s. ;D

Oh well, I guess it's time to check out the 2.6.0-test series.

**Schotty** · 08-18-2003, 08:13 PM

When I get done fixing my Slack setup (no gui, and I didnt install any text mail readers) I can look thru my Red Hat mail list archive. I saw some people run into issues around 7.3/8.0 but had most of it resolved around RHL 9. I do recall that some had the issue of the USB2 not being supported, and that is the ONLY mode that the HDD would talk.

I bought a Maxtor card for my box and my external devices work fine. But then agian, my mobo is 4 years old, and doesn't have anything but USB 1.1 ports.

But USB HDD units do work, and just fine. I will get the spiel and see if any of it assists you.

**demian** · 08-19-2003, 01:11 PM

I did it

I grabbed Greg Kroah-Hartman's usb patch for 2.4.21 (most of which is included in 2.4.22-rc2, I believe). And it works!! Yay

Ok, that's the good news. Now this thing made me realize something: I am aware of the fact that USB 2.0 support started in the 2.5.x kernel tree and that everything we have in 2.4 is a backport. Also I am totally aware that my running debian unstable with a custom kernel that's patched from various sources pretty much leaves me alone when it comes to trouble shooting and I can get down with that no problem.

BUT SUSE?!?! 8.2?!?!? Released, what? a few months ago? Doesn't handle usb 2.0 disks? They have their own kernel developers who backport the stuff a home user would need from the develeopment branch and include it in their own kernel version, right. The controller I have, btw. is a VIA make which is about as mainstream as it gets. And I also tried it with a SuSE install on a machine with a different mainboard with an SiS controller -> same thing. This is just sad. I always liked SuSE and recommended it to anybody who was new to and interested in Linux. Well, I guess my SuSE pimpin' days are over.

Schotty, I am still interested in the stuff from the mailing list. If you could send me a link I'd most appreciate it. After all, I will have to find another mainstream distro that I can shove on my friends to convince them that the Penguin actually _is_ your friend.

**Schotty** · 08-20-2003, 06:13 PM

This is teh parent of the thread. The others were dumb shit like, not having usb in the kernel and soforth.

I have a Maxtor 3000LE USB hard disk that I'm trying to use on a Redhat
9 system. I created a partition on the disk using fdisk, then tried to
make a new file system with mke2fs - and it runs for a while, then
eventually hoses up the machine enough that it stops responding to
everything except ping, and I have to reboot the box.

I've tried several different parameters on mkfs, etc, to get it to work
- all failed.

When I take this drive and move it over to an older (RH 7.1) system, it
works fine.

Any ideas or experience anyone can share would be great.

Thanks.

--
redhat-list mailing list