USB Crashing Problems Under Linux--Contact Me
Printed from: Neuros Forums
Topic URL: http://www.neurosaudio.com/community/forum/topic.asp?TOPIC_ID=1222
Printed on: 01/04/2007
Topic:
Topic author: adam(at)bostoncoop.net
Subject: USB Crashing Problems Under Linux--Contact Me I know several people have experienced USB-related crashes under Linux. I'm working with a USB developer to try to isolate the bug, whether it's in Linux or in the Neuros device.
Replies:
Posted on: 07/23/2003 4:37:39 PM
Message:
If you've experienced this problem (see., e.g., bugs #367), please contact me so we can gather more data.
Thanks!
Reply author: Sean Starkey
Replied on: 07/23/2003 5:29:58 PM
Message:
This happens to me all the time. 
Most of the time it just locks up the whole system for 30-60 secs, but recovers. Occasionally it locks up completely and I have to do a hardware reset.
Sean Starkey - Project Manager for Neuros Database Manipulator - http://neurosdbm.sourceforge.net
Reply author: Craig
Replied on: 07/24/2003 08:59:09 AM
Message:
I also get constant lockups but I didn't know anyone else was. This is on a 2.4.20 SMP kernel. I've also gotten kernel panics on a number of occasions but running mount in sync mode seems to have helped. Also, I've noticed that lockups for me only occur during long file transfers.
Reply author: slestak(at)cavtel.net
Replied on: 07/24/2003 12:16:54 PM
Message:
I use gentoo-sources 2.4.20r5 I think (from memory). Just for the record, I have not had any lockups with this machine (its the only machine I use to add software to my neuros, using Positron.
I got one large lockup on a family members computer, but that was under XP Home while transferring a large ISO to disk using windows explorer. I was on battery power, but I had plenty. Note to self, dont copy 650 MB files on battery power.
Edited:
Hopefully with USB2, we'll see less of this since technically, an ISO can be copied much faster, so the overral time the machine is stressing will be shorter.
Reply author: kgiverson
Replied on: 07/24/2003 9:13:23 PM
Message:
I've used my Neuros on two different Linux boxes with a range of results.
The first is an Athalon SMP machine that was initially running 2.4.18 and has OHCI USB. The Neuros would consistently lock up and cause the machine to lock up. I upgraded to 2.4.21 and after the upgrade the Neuros would still lock up, but it would no longer lock up the machine.
The second is a single processor pentium III with UHCI USB initially running 2.4.9 and using the usb-uhci kernel module. I had a couple of machine lock ups, but mostly just got "file system is read only" errors.
I upgraded to 2.4.21 which took care of the read only issues, but I still had some machine lock ups. I saw another post here about switching from usb-uhci to uhci, tried it and have not had an issue since.
Reply author: qvack_82
Replied on: 07/24/2003 10:22:28 PM
Message:
Here's My experiences with the Neuros under Linux.
linux-2.4.20, uhci. Neuros locks up, kernel has all sorts of errors that I don't understand.
linux-2.4.21, uhci. No worries at all, IF you wait for 10 seconds after turning off the neuros before plugging my USB cable in. If I don't wait, i get 'attempted access past end of device' errors.
linux-2.6.0-test1, uhci. Neuros goes through the Syncronization thing, with the animated icon, then stops, as normal. Linux hasn't recognised the neuros at all, no 'new usb device detected' or 'mass storage device found' etc. etc.
The most significant thing is to add a sync flag to your mount. ie
mount -t vfat -o sync /dev/sda1 /mnt/neuros
or
/etc/fstab
/dev/sda1 /mnt/neuros vfat user,noauto,sync 0 0
this means that there is no buffering, and writes will be immediatly committed to the neuros disk.
Hope this helps.
Cheers
Paul
ln -s /dev/null ~/.sig
Reply author: alecm
Replied on: 07/25/2003 12:25:41 PM
Message:
My experiences are similar. For me, switching from the usb-uhci driver to the uhci driver made a huge difference. The occasional loss of system interactive response (frozen pointer for some seconds) that remained was fixed by mounting with the sync option. I'm running 2.4.21 SMP enabled.
Reply author: enry
Replied on: 07/28/2003 08:07:46 AM
Message:
Please be sure your machine has the latest BIOS on it. There are cases where there will be USB firmware changes that can fix the problem.
Reply author: qvack_82
Replied on: 07/29/2003 10:07:10 PM
Message:
some more luck with 2.6.0-test2
uhci-hcd, usb-storage, sd_mod.
dmesg says that the Neuros is recognised, and assigned to sda, and that's all, no /dev/sda1, no /dev/scsi/*. and then any modprobe/rmmod/lsmod execution locks up that terminal solid, a reboot then segfaults :o)
hooray for development kernels :o)
Cheers
Paul
ln -s /dev/null ~/.sig
Reply author: adam(at)bostoncoop.net
Replied on: 07/29/2003 10:16:22 PM
Message:
I've had no trouble getting the device recognize with the 2.5/2.6 line of kernels, it's just that the process gets stuck after a while and there's no way of recovering it. I'm not sure why you haven't been able to even mount the device in 2.5/2.6, as that's never been a problem for me.
Here's the kernel configuration I used, in case it is helpful:
http://bostoncoop.net/adam/temp/config-2.6.0
In other news, it seems to work perfectly, or nearly perfectly, in 2.4.22, even though it failed terribly in 2.4.21. The only problem is that sometimes it takes a while to flush the buffer, but I think this is actually normal (and goes away if I mount the device -o sync).
Reply author: MeatMyDemands
Replied on: 07/30/2003 12:35:09 PM
Message:
This is a big problem for me. I have tried 2.4.21 with both the regular and JE UHCI drivers. I have also tried the 2.6.0-test kernel. In all cases, the USB bus will hang when I try to do a positron rebuild or even just copy a large file (like, for instance, a firmware file). Using the sync option on the mount does not make the problem go away or reduce it notably.
I will mention here that my mainboard is a FIC SD11. I mention it because this mainboard has known chipset/implementation problems. One, for example, is the tendency to misbehave when trying to use UDMA with hard disks. I have not heard of USB problems with this mainboard, though.
My problem consists of a total hang of the USB subsystem. My USB mouse quits working, and does not come back. My system cannot even be shut down properly, because it hangs when trying to umount the neuros. I can reproduce this seemingly at will.
Reply author: DocWhat
Replied on: 08/02/2003 11:39:05 AM
Message:
I have a question: Does anyone with these locks use a new version ALSA? I discovered reciently that the newer versions of ALSA can cause corruptions of the PCI bus. That'll cause USB to lock up or even the whole machine.
If you do have alsa, try disabling it and preventing it from loading *any* modules. Then reboot and try using the neuros.
Ciao!
Neuros Order #2553
Reply author: alecm
Replied on: 08/03/2003 1:30:23 PM
Message:
I'm using ALSA 0.9.2 under 2.4.21 (SMP) with the JE drivers and the sync mount option. I have had no problems whatsoever with this setup.
Alec
Reply author: bitboy
Replied on: 09/09/2003 3:28:52 PM
Message:
Hi, all,
I don't know if you are still looking for victims^Wvolunteers, but I'm also having the same lockup problems (posted in another thread) with firmware 1.40 and Positron 1.0.
I'll be happy to assist in troubleshooting in any way I can.
Thanks again for all the hard work,
Brian
Reply author: adam(at)bostoncoop.net
Replied on: 09/09/2003 3:59:37 PM
Message:
quote:
Originally posted by bitboy
I don't know if you are still looking for victims^Wvolunteers,
You might want to try kernel 2.4.22 if you can. My problems were 95% fixed with that kernel, although some of them come back with 2.6.0test*.
Also, I assume you've seen the advice of mounting the device sync (mount -o sync or add sync to mount options in /etc/fstab).
Reply author: bitboy
Replied on: 09/16/2003 08:11:28 AM
Message:
Greetings, all.
Finally had the time to finish my upgrade to kernel 2.4.22 last night and it did fix the problems I was having with USB timeouts. Thanks for the suggestion.
Brian
Reply author: Toojays
Replied on: 10/14/2003 06:18:02 AM
Message:
My two cents on this.
I only got my Neuros today, and before that I hadn't tried using any USB devices on this computer. Anyway, it took me a while to get it working consistently. I wasn't getting full system lockups, but the Neuros would freeze, and it seemed as though there was some kind of lockup in the kernel module---I could not rmmod usb-storage or unmount the Neuros. When I tried to umount, the shell I was in would freeze up . . . even "killall -KILL umount" from another shell wouldn't work, nor would a "reboot", even though I could click around KDE and use other programs with no problems.
Anyway, it looks to be working now. The difference (I think) is with the options to the usb-storage module. In the kernel menuconfig, under "USB Mass Storage Support", it has a bunch of sub-options about debugging, CF readers, bridges, etc. On my first try to get things working, it had not selected any of these. After comparing it to my dad's computer, which did seem to work, I selected _all_ of the suboptions. Now it seems to work fine.
This is with a K7S5A motherboard, kernel 2.4.20-gentoo-r7, usb-ohci.
Reply author: Toojays
Replied on: 10/15/2003 05:39:34 AM
Message:
Grrr,
Scratch that last post . . . it just did it again, so I have not fixed the problem (but perhaps made it less obvious).
Reply author: ExecutorElassus
Replied on: 02/10/2004 12:56:57 PM
Message:
So, i'm running 2.4.21-166, on SuSE 9.0, with uhci and ehci-hcd running as modules. neuros firmware is 1.40. i have three VIA usb2 hubs on the motherboard, and SuSE reports that they're working properly (i have a zip drive and a joystick on one hub, and both work perfectly).
when i plug the neuros in, the "synchronization" window pops up, but the animation freezes almost immediately. sometimes the computer beeps (when it detects a new device) and sometimes not. in all cases, once the neuros is frozen, the kernel doesn't see it as a device, and i can't mount it.
any idea why this would freeze? should i install the 2.6 kernel? i could use the neuros perfectly before i upgraded the motherboard (the old one had built-in usb1.1), so i'm guessing this is a kernel problem.
hope someone can help! i haven't been able to use the neuros since november or so.
thanks,
EE
Reply author: kronin
Replied on: 02/10/2004 1:05:12 PM
Message:
quote:
Originally posted by ExecutorElassus
So, i'm running 2.4.21-166, on SuSE 9.0, with uhci and ehci-hcd running as modules. neuros firmware is 1.40. i have three VIA usb2 hubs on the motherboard, and SuSE reports that they're working properly (i have a zip drive and a joystick on one hub, and both work perfectly).
Did you try booting with no other USB devices attached, then try plugging in the Neuros? Also, are you getting any output at all in /var/log/messages?
Reply author: BuddhaTigger
Replied on: 02/12/2004 10:17:16 AM
Message:
I'm also experiencing problems with the Neuros and Linux (Fedora 2.4.22-1.2149.nptl). It usually locks up when unmounting. But I think it (the Neuros) locks up after syncing the new songs onto it using NDBM. When it locks up I see something like lost interrupt in dmesg. To make matters worse that USB is sharing the same IRQ with the ethernet (just found that out, btw it also gets a lost interrupt)! I'll remove all the USB devices to limit what I'm seeing. Any suggestions on what to look for or what other info I should get?
[Edited March 8, 2004]
I'm now able to mount and write to the disk with ndbm. Unfortunately I still had problems with my system after I umounted the neuros. I tried to mount it again the neuros and the mount command hung.
Thanks
Reply author: Ashkelon
Replied on: 02/18/2004 9:24:21 PM
Message:
I'm running SuSE 9.0 default kernel 2.4.21-166. I've expierenced one complete lock-up from plugging the nerous in to the usb port. However, I've also had my system hang for no apparent reason (not realated to the neuros). On one machine (HP Kayak) I have severe problems with hangs on almost a daily basis but havn't been able to figure out what's causing it...even just sitting there doing nothing it will croak. For my (relativily) stable system I have usb drives I use for backups and have no problems. I've been able to mount the neuros only twice out of a bazillion try's. Normally it will start into its sync for a second or two and just hang. It's never assigned a device (error messages run the spectrum). When I finally got the thing mounted I was able to upgrade firmware and load some songs - no problem from positron. It all proved futile becuase the neuros was hung and reuilt the db on me. It always hangs no matter if I can mount it or not. I'm thinking there is some interaction with the kernel, hardware settings, and neuros that makes it useless. Well, it would be nice if I can get this thing to work. For now its just collecting dust until I can find the time to fiddle with it somemore (and I JUST bought it!!!)
Reply author: ExecutorElassus
Replied on: 02/19/2004 12:09:01 AM
Message:
[/quote]Did you try booting with no other USB devices attached, then try plugging in the Neuros? Also, are you getting any output at all in /var/log/messages?
[/quote]
i haven't tried booting with no usb attached yet (friday's my day off, so i'll try then) but /var/log/messages says these things:
when i first plug in the neuros, it looks like i get this:
Feb 10 13:42:46 Ukiyo-e kernel: hub.c: new USB device 00:10.2-2, assigned address 5
Feb 10 13:42:46 Ukiyo-e kernel: WARNING: USB Mass Storage data integrity not assured
Feb 10 13:42:46 Ukiyo-e kernel: USB Mass Storage device found at 5
Feb 10 13:42:49 Ukiyo-e kernel: usb_control/bulk_msg: timeout
Feb 10 13:42:49 Ukiyo-e kernel: usbdevfs: USBDEVFS_CONTROL failed dev 5 rqt 128 rq 6 len 18 ret -110
the last two lines repeat several hundred times, and then i get:
Feb 10 13:43:15 Ukiyo-e /etc/hotplug/usb.agent[21210]: cannot get config descriptor: Connection timed out
Feb 10 13:43:15 Ukiyo-e kernel: sdc : READ CAPACITY failed.
Feb 10 13:43:15 Ukiyo-e kernel: sdc : status = 1, message = 00, host = 0, driver = 08
Feb 10 13:43:15 Ukiyo-e kernel: Current sd00:00: sense key Not Ready
Feb 10 13:43:15 Ukiyo-e kernel: Additional sense indicates Medium not present
Feb 10 13:43:15 Ukiyo-e kernel: sdc : block size assumed to be 512 bytes, disk size 1GB.
Feb 10 13:43:15 Ukiyo-e kernel: sdc: Write Protect is off
Feb 10 13:43:15 Ukiyo-e kernel: sdc: I/O error: dev 08:20, sector 0
Feb 10 13:43:15 Ukiyo-e kernel: I/O error: dev 08:20, sector 0
Feb 10 13:43:15 Ukiyo-e kernel: I/O error: dev 08:20, sector 2097144
Feb 10 13:43:15 Ukiyo-e kernel: I/O error: dev 08:20, sector 2097144
Feb 10 13:43:15 Ukiyo-e kernel: I/O error: dev 08:20, sector 0
Feb 10 13:43:15 Ukiyo-e kernel: I/O error: dev 08:20, sector 0
Feb 10 13:43:15 Ukiyo-e kernel: unable to read partition table
Feb 10 13:43:15 Ukiyo-e kernel: Device not ready. Make sure there is a disc in the drive.
Feb 10 13:43:15 Ukiyo-e kernel: I/O error: dev 08:20, sector 0
Feb 10 13:43:20 Ukiyo-e kernel: usb_control/bulk_msg: timeout
Feb 10 13:43:22 Ukiyo-e kernel: usb.c: USB disconnect on device 00:10.2-2 address 5
the weird thing is this: sdc is the ZIP drive. the neuros, when it was connecting, got sdd. for a while, i sometimes got a new fstab entry assigning /dev/sdd4 to /media/sdd4. but even that would only last a few seconds, and then the hotplugger would disconnect it and delete the entry. i tried entering it manually, but that got deleted, too.
does this help at all? it seems like the neuros is just wigging out: that the box just isn't getting a response, and it's timing out. but maybe not?
thanks,
EE
Reply author: ExecutorElassus
Replied on: 02/23/2004 2:39:42 PM
Message:
quote:
Originally posted by kronin
Did you try booting with no other USB devices attached, then try plugging in the Neuros? Also, are you getting any output at all in /var/log/messages?
okay, so i tried this, and the neuros still froze. also, now my eth0 is giving me a "no IPv4 routers found" message in /var/log/messages (even after rebooting when the two usb devices i removed were re-attached, and the neuros not connected), and i can't connect to the internet. i can logon/use/connect to my router's admin page, so i KNOW eth0 is working. the router says it's connected to the gateway, and i'm connected to the router, so i don't know what's hanging there, either. but this only started after i tried attaching the neuros with no other usb attached.
is this a kernel problem? i'm a little apprehensive about trying the 2.6 kernel, but perhaps i should?
also, i should mention that the neuros has lately started switching into sync mode (with the animation and the "do not disconnect" message) when i plug a 1/8-in jack into the line-in port, with nothing in the usb port. i don't know how to knock it out of sync mode without doing the [UP]-[PAUSE] force shutdown, which i'd like to avoid.
any help on these? it seems like my neuros is having serious problems...
thanks,
EE
Reply author: kronin
Replied on: 02/23/2004 3:07:21 PM
Message:
quote:
Originally posted by ExecutorElassus
also, i should mention that the neuros has lately started switching into sync mode (with the animation and the "do not disconnect" message) when i plug a 1/8-in jack into the line-in port, with nothing in the usb port. i don't know how to knock it out of sync mode without doing the [UP]-[PAUSE] force shutdown, which i'd like to avoid.
From an earlier post, I'm assuming you're still running firmware 1.40. There was a bug that went randomly into sync mode that was fixed somewhere along the line, so I would definitely upgrade to 1.45.
Are there any BIOS updates for your motherboard? Were you running the same kernel version on your computer when it worked, before you upgraded your motherboard? Did you select any different options when you compiled your kernel (or is it just a stock SUSE 9 kernel)? Have you tried the usb-uhci module instead of just the uhci module?
As to why you're having network issues, that's very odd. Maybe your default gateway on your linux box somehow got unset, but simply rebooting with some USB devices unplugged shouldn't affect this at all.
Reply author: ExecutorElassus
Replied on: 02/23/2004 3:33:35 PM
Message:
quote:
Originally posted by kronin
Are there any BIOS updates for your motherboard? Were you running the same kernel version on your computer when it worked, before you upgraded your motherboard? Did you select any different options when you compiled your kernel (or is it just a stock SUSE 9 kernel)? Have you tried the usb-uhci module instead of just the uhci module?
this is the stock SuSE9 kernel. i've tried using usb-uhci, with the same result. i can check for BIOS updates, but i won't be able to do anything with them until i can get internet access at home (or make the boot-disk elsewhere). i'll give that a shot.
the problems occurred first when i upgraded the mobo. the first mobo was an EpoX 8KHA+ with builtin 2xUSB1.1 (one controller); the new one is a soyo KT400 platinum with 6xUSB2 (three controllers). same cpu chip, same RAM, etc. i'm guessing the kernel i had at the time (SuSE8.2 stock) couldn't do usb2, as i was having problems with the ZIP drive, too. but upgrading to 9, while it fixed the ZIP drive, still didn't fix the neuros. maybe i should think to bring my neuros to work someday, and use one of their windows boxes to update the firmware.
quote:
As to why you're having network issues, that's very odd. Maybe your default gateway on your linux box somehow got unset, but simply rebooting with some USB devices unplugged shouldn't affect this at all.
i'm thinking this was going wrong for a few minutes before i tried rebooting this morning, but i have no idea why. it was working perfectly last night. someone else on the forum mentioned that the neuros was getting the same irq as eth0, so maybe that's what's going on here. it still hasn't cleared up (if it did, it would be autofetching my email, which it isn't) so i'm not sure what to fix.
thanks for getting back so quickly!
EE
Reply author: kronin
Replied on: 02/23/2004 7:57:46 PM
Message:
quote:
Originally posted by ExecutorElassusquote:
As to why you're having network issues, that's very odd. Maybe your default gateway on your linux box somehow got unset, but simply rebooting with some USB devices unplugged shouldn't affect this at all.
i'm thinking this was going wrong for a few minutes before i tried rebooting this morning, but i have no idea why. it was working perfectly last night. someone else on the forum mentioned that the neuros was getting the same irq as eth0, so maybe that's what's going on here. it still hasn't cleared up (if it did, it would be autofetching my email, which it isn't) so i'm not sure what to fix.
thanks for getting back so quickly!
EE
If you can browse to the router's admin page, then my guess is it's either a routing table issue or a default gateway issue. The router lives in the same network segment as your computer, whereas if you try to browse anywhere else it needs to go to a different network segment. This is where the default gateway comes into play. Did you change any settings on your router? Can you hook a different box up behind the router and get through? What's the output of ifconfig -a and route -n?
Reply author: ExecutorElassus
Replied on: 02/23/2004 10:01:25 PM
Message:
quote:
If you can browse to the router's admin page, then my guess is it's either a routing table issue or a default gateway issue. The router lives in the same network segment as your computer, whereas if you try to browse anywhere else it needs to go to a different network segment. This is where the default gateway comes into play. Did you change any settings on your router? Can you hook a different box up behind the router and get through? What's the output of ifconfig -a and route -n?
my bad: i released/renewed the dsl connection, and now it's totally fine. should have thought to do that, since the phone people were working on the line yesterday.
anyway, now about that usb... i'll try to update the firmware from a windows box tomorrow. otherwise, if that doesn't work, i don't know how to get it syncing. sigh...
thanks!
EE
Reply author: ExecutorElassus
Replied on: 02/24/2004 10:30:30 PM
Message:
more!
so, i updated to 1.45 on a windows box, and tried connecting it. this time, the neuros didn't freeze until the computer beeped. but when i tried to mount it, the computer totally froze.
so, i'm guessing that it's something with linux, since i could connect it to an OSX box and a windows box.
grr...
EE
Reply author: nxg125
Replied on: 02/25/2004 08:42:13 AM
Message:
quote:
so, i updated to 1.45 on a windows box, and tried connecting it. this time, the neuros didn't freeze until the computer beeped. but when i tried to mount it, the computer totally froze.
Did you try mounting with the sync option (mount -o sync)? That has helped a number of people.
--Nick
Reply author: ExecutorElassus
Replied on: 02/25/2004 09:39:18 AM
Message:
quote:
Did you try mounting with the sync option (mount -o sync)? That has helped a number of people.
i tried, but the neuros freezes before the kernel even sees the device, so i get "/dev/sdd4 is not a valid block device." (i do have sda through sdc, so sdd would be the normal device. i'm not sure why it's sdd4, but that's what it's always called it.) it seems like the neuros locks up before i get to that point. /var/log/messages now also says this:
hub.c: new USB device 00:10.2-2, assigned address 3
Feb 25 10:32:10 Ukiyo-e kernel: usb_control/bulk_msg: timeout
Feb 25 10:32:30 Ukiyo-e last message repeated 4 times
Feb 25 10:32:30 Ukiyo-e kernel: usb.c: unable to get device descriptor (error=-110)
Feb 25 10:32:30 Ukiyo-e kernel: hub.c: new USB device 00:10.2-2, assigned address 4
Feb 25 10:32:35 Ukiyo-e kernel: usb_control/bulk_msg: timeout
Feb 25 10:32:35 Ukiyo-e kernel: usb.c: USB device not accepting new address=4 (error=-110)
huh?
EE
Reply author: bitboy
Replied on: 02/25/2004 10:14:07 AM
Message:
quote:
Originally posted by ExecutorElassus
i tried, but the neuros freezes before the kernel even sees the device, so i get "/dev/sdd4 is not a valid block device." (i do have sda through sdc, so sdd would be the normal device. i'm not sure why it's sdd4, but that's what it's always called it.) it seems like the neuros locks up before i get to that point. /var/log/messages now also says this:
Something funky this way comes... Unless you've repartitioned your Neuros (I rather doubt it would work at all if you had), it should be showing up as /dev/sd[abcd...]1. Something else is going on here. I'll have to go back and look at previous posts when I get the chance, but when glancing at this, there was something else that struck me as fishy.
I think you mentioned that the hardware itself is USB 2.0 High Speed? If so, the EHCI driver should be loading as well. I've done a very little bit of research on USB hardware, and it is my understanding that the USB 2.0 hardware doesn't actually provide the backwards compatibility required to talk to a USB 1.0/1.1 device. There is generally a second UHCI or OHCI controller on the USB 2.0 card that talks to the older spec stuff. So, it would be interesting to note whether or not there is a problem trying to get the drivers bound to the proper hardware.
I know on my home system (Athlon 1.2 ghz/ASUS A7V/Slackware 9.1 with kernel 2.6.3), the stock USB 1.1 controller on the motherboard works fine with my Neuros under Linux. I've also added a USB 2.0 PCI card (in anticipation of the Glorious Day , and my Neuros really doesn't like that USB 2.0 card.
Brian
Reply author: nxg125
Replied on: 02/25/2004 11:05:59 AM
Message:
quote:
i'm not sure why it's sdd4, but that's what it's always called it.) it seems like the neuros locks up before i get to that point. /var/log/messages now also says this:
I agree, that is very strange. Have you tried reformatting and/or running dosfsck on the Neuros? I've never seen mine show up as anything other than partition 1.
--Nick
Reply author: Azraak
Replied on: 02/25/2004 11:09:52 AM
Message:
I posted in the "mounting on Linux" thread elsewhere about issues I was having with 2.4.22 and earlier kernels. My Neuros was freezing after several mb (there did not seem to be any exact figure, but usually a good 30 mb would definitely bring about a freeze, if not far less) and the kernel was reporting USB timeouts. I suspect the USB timeouts were in fact caused by the Neuros freezing. This problem appears to be a combination of several factors: hardware, the kernel and the Neuros. While one would think the Neuros was not the issue given it works for some and not for others, I have heard of cases where other USB mass storage devices worked fine. It seems like it is quite a difficult option to track down and is probably a result of the kernel/hardware not quite behaving itself (I will wager that some USB devices might be a little more 'tolerant' or something, while the Neuros just bales out altogether, so it probably isn't actually a 'problem' with the way the Neuros implements USB. However, this is all juts conjecture and I really have no clue.)
Anyway, I have been running 2.6 for a while now and finally got around to testing my Neuros (currently using 2.6.3). I am pleased to announce that it now works without issues; I haven't tried copying any files greater than about 100 mb, but I suspect that should not be a problem, as I do not notice any errors, lag, etc. that might signify potential problems. I am mounting with the sync option. Please note that I have not tried 2.4 kernels after 2.4.22 on this machine; I am quite happy with 2.6 and require some of its features. I am not quite sure what board this is, but it is an Intel Celeron 300 with a Via chipset; /proc/pci tells me:
USB Controller: VIA Technologies, Inc. USB (rev 8)
Being Via, I use the uhci module (2.6 uses the alternate uhci driver and as far as I can tell, the other one has been removed completely).
Interestingly enough, 2.4.22 worked fine with the Neuros on another machine using an Intel USB controller (uhci driver). Actually, I haven't heard of any USB/Neuros problems with Intel USB controllers (but I haven't read these topics too closely, so I might have missed something), which is interesting. Any USB issues I have ever had seem to be with Via chipsets. Arrrg!
Jamie
Reply author: ExecutorElassus
Replied on: 02/25/2004 5:30:12 PM
Message:
quote:
Originally posted by bitboy
I think you mentioned that the hardware itself is USB 2.0 High Speed? If so, the EHCI driver should be loading as well. I've done a very little bit of research on USB hardware, and it is my understanding that the USB 2.0 hardware doesn't actually provide the backwards compatibility required to talk to a USB 1.0/1.1 device. There is generally a second UHCI or OHCI controller on the USB 2.0 card that talks to the older spec stuff. So, it would be interesting to note whether or not there is a problem trying to get the drivers bound to the proper hardware.
lsmod says:
input 3488 0 [keybdev hid adi mousedev joydev evdev]
usb-uhci 23824 0 (unused)
ehci-hcd 18028 0 (unused)
usbcore 64364 1 [usb-storage usbserial hid usb-uhci ehci-hcd]
i'm not exactly sure what this means. anybody have an idea how to track which module is getting loaded when i plug the neuros in? my hardware info thingie lists several usb resources, as follows:
----- manually configured hardware -----
UniqueID=B3Fu.gwWiykcvg64
ParentID=wRyD.1U83dZBAcO6
HWClass=hub
Model=Linux 2.4.21-166-athlon ehci_hcd VIA Technologies, Inc. USB 2.0
Configured=no
Available=yes
Needed=no
Active=unknown
Bus=0x86
Slot=0x101
BaseClass=0x10a
DeviceName=VIA Technologies, Inc. USB 2.0
VendorName=Linux 2.4.21-166-athlon ehci_hcd
RevisionName=2.04
Serial=00:10.3
USBGUID=000000000000000fffffffe3
Hotplug=4
HWClassList=000000080100
Res.Baud=1500000,0,0,0x00,0x00
UniqueID=B3Fu.U0mJCy94H96
HWClass=hub
Model=VIA Root Hub
Configured=no
Available=no
Needed=no
Active=unknown
Bus=0x86
Slot=0x201
BaseClass=0x10a
VendorID=1106
DeviceID=0000
DeviceName=Root Hub
VendorName=VIA Technologies, Inc.
USBGUID=000000000000000000000000
Hotplug=4
HWClassList=000000080100
Res.Baud=0,0,0,0x00,0x00
UniqueID=B3Fu.UMIIpMR0DK5
HWClass=hub
Model=USB UHCI Root Hub
Configured=new
Available=yes
Needed=no
Active=unknown
Bus=0x86
Slot=0x301
BaseClass=0x10a
DeviceName=USB UHCI Root Hub
Serial=e000
USBGUID=00000000000000000000e000
Hotplug=4
HWClassList=000000080100
Res.Baud=1500000,0,0,0x00,0x00
UniqueID=B3Fu.Ec5hf_oLR7C
ParentID=f5xu.D4EZuGTknpB
HWClass=hub
Model=VIA USB UHCI Root Hub
Configured=no
Available=yes
Needed=no
Active=unknown
Bus=0x86
Slot=0x301
BaseClass=0x10a
VendorID=1106
DeviceID=0000
DeviceName=USB UHCI Root Hub
VendorName=VIA Technologies, Inc.
Serial=e000
USBGUID=00000000000000000000e000
Hotplug=4
HWClassList=000000080100
Res.Baud=1500000,0,0,0x00,0x00
UniqueID=RPlM.mJUDT7x0wp7
HWClass=disk
Model=NEUROS dig. audio comp.
Configured=no
Available=no
Needed=no
Active=unknown
Bus=0x84
Slot=0x300
BaseClass=0x106
DeviceName=dig. audio comp.
VendorName=NEUROS
RevisionName=1.00
Serial= UnixDevice=/dev/sdd
USBGUID=045154090000000000000000
Hotplug=4
HWClassList=000020080200
Res.Size=3,39070048,512
Res.DiskGeometry=19077,64,32,1
i'm sorry for the length, but i don't know how to interpret these. there are several more, but i'm not sure which are important. any ideas?
thanks!
EE
Reply author: nxg125
Replied on: 02/25/2004 6:23:14 PM
Message:
quote:
lsmod says:
input 3488 0 [keybdev hid adi mousedev joydev evdev]
usb-uhci 23824 0 (unused)
ehci-hcd 18028 0 (unused)
usbcore 64364 1 [usb-storage usbserial hid usb-uhci ehci-hcd]
OK, one thing you can try is to use another USB driver. There are two of them, one called usb-uhci (which you are currently using) and one just called uhci. For some people, one works better than the other.
--Nick
Reply author: Azraak
Replied on: 02/25/2004 6:48:13 PM
Message:
The driver called simply "uhci" is I believe generally considered to be the better driver anyway. Also, EE, you might want to upgrade your kernel to a later 2.4.x kernel; 2.4.25 is the latest at time of writing. I believe 2.4.21 had some USB problems that were fixed in later versions.
Jamie
Reply author: adam(at)bostoncoop.net
Replied on: 02/25/2004 7:46:57 PM
Message:
I started this thread long ago when I was having all sorts of problems. The problems for the most part disappeared around 2.4.23 kernel for me, and since I've been using the most recent 2.6 kernels I've had no crashing whatsoever. 2.6.2 and 2.6.3 have been particularly good. If you are in a position to try a 2.6 kernel and you are experiencing Neuros-USB issues, I highly recommend it.
There is one remaining problem which is a little weird but I haven't quick tracked down: positron frequently gets "hung up" when copying a file to the device. It might pause for a minute or two, or sometimes even as long as ten minutes, and then continue. Once in a very long while I have to remount the device. When I use cp rather than positron, and then have positron rebuild the database, I don't get any of these problems.
I have the device mounted -o sync, so I don't think the issue is buffering. I've tried reducing the blocksize in positron to as small as 1K and it still happens. I can't guess what might make positron different from cp, but that's the result I get. In any case, it's not a big deal, I don't mind using cp and rebuild, although if I had the 20G model I suppose it would be a bigger problem.
Reply author: ExecutorElassus
Replied on: 02/26/2004 12:32:04 AM
Message:
when i switched uhci for usb-uhci, i still got the timeouts. i do have ehci-hcd running, but not ehci.
sigh... i'm almost desperate enough to attempt installing a 2.6 kernel. i've never done it before (the only things i compile myself right now are apps) so i'm apprehensive. is there a good howto out there? can you recommend a good, stable, 2.6, or should i stay on the 2.4 branch?
thanks, y'all. it's encouraging to have help.
EE
Reply author: nxg125
Replied on: 02/26/2004 07:25:25 AM
Message:
quote:
sigh... i'm almost desperate enough to attempt installing a 2.6 kernel. i've never done it before (the only things i compile myself right now are apps) so i'm apprehensive. is there a good howto out there? can you recommend a good, stable, 2.6, or should i stay on the 2.4 branch?
Bummer... I don't blame you for being frustrated. There was a discussion yesterday on Slashdot about upgrading to 2.6. Hopefully that has some good info for you. I would say just go for the latest (2.6.3?) version from kernel.org. I haven't used the 2.6 branch too much so I can't speak to its stability and all that, but several people here have said it cleared up their USB issues.
Good luck...
--Nick
Reply author: eulachon
Replied on: 05/12/2004 10:40:27 PM
Message:
I'm running fine under Mandrake 10 community (2.6 kernel) using supermount to automount it when I plug it in. works great!
but... it does have a buffering issue as it runs really fast and then freezes, really fast and freezes. I've added sync to the supermount options in fstab, see what happens when i next reboot.
Reply author: ExecutorElassus
Replied on: 05/12/2004 10:49:11 PM
Message:
quote:
Originally posted by ExecutorElassus
is there a good howto out there? can you recommend a good, stable, 2.6, or should i stay on the 2.4 branch?
thanks, y'all. it's encouraging to have help.
EE
how about this: is ANYBODY out there running any of the gentoo 2.4 branch sources on THIS mobo? my problems seem to be with SuSE's kernel, so i'm going to start by replacing that. i don't want to use the 2.6 yet, because i use the winex engine (love playing windoze games without buying windoze), and it seems to flip out under 2.6.
to refresh, my mobo is a soyo KT400 Dragon (platinum edition). anybody get gentoo 2.4 kernels to run a neuros on that?
thanks,
EE
Neuros Forums : http://www.neurosaudio.com/community/forum/
© Copyright ©2002-05 Neuros Technology International, LLC all rights reserved.