Description of problem: When plugging in my USB stick, the kernel recognizes it, device nodes get created by udev, scsi.agent recognizes it and pam_console.dev restores console perms for the device files, but it doesn't show up in lshal or hal-device-manager. Needless to say that no entries show up in fstab. Version-Release number of selected component (if applicable): hal-0.2.98.cvs20040929-3 How reproducible: Easy. Steps to Reproduce: 1. Plug in the USB stick Actual results: As described. Expected results: HAL should recognize the device and run fstab-sync accordingly to update /etc/fstab so that in the end Nautilus shows me the device and the console user can mount it. Additional info: Logs following.
Created attachment 104936 [details] syslog messages when plugging in the USB stick.
nils@gibraltar:~> cat /proc/scsi/scsi Attached devices: Host: scsi2 Channel: 00 Id: 00 Lun: 00 Vendor: Generic Model: Traveling Disk Rev: 1.11 Type: Direct-Access ANSI SCSI revision: 02
Created attachment 104937 [details] Output from "hald --daemon=no --verbose=yes" run as root
Created attachment 104938 [details] syslog messages when plugging in the USB stick.
Created attachment 104939 [details] Output from "hald --daemon=no --verbose=yes" run as root What I did is here: - start hald as described - plug in USB stick, doesn't get recognized - plug it out and plug it in again, it get's recognized
Created attachment 104940 [details] lshal output when it doesn't recognize the device
Created attachment 104941 [details] lshal output when it recognizes the device
Funny: In the course of this, it mounted the stick a number of times. It only doesn't cause the mount points (in the GUI) to vanish. I'll attach a screenshot and the resulting fstab.
Created attachment 104942 [details] screenshot of many funny icons that are all my one and only usb stick
Created attachment 104943 [details] /etc/fstab
Only the mountpoint in /media/usbdisk3 has contents, the others are empty -- but they should be umounted and deleted from fstab. Another data point is that in the course of trying things, the device got first /dev/sda,m then sdb, sdc, now sdd.
As we discussed on IRC last week, I've not fixed the issue with the very first hotplug event. Please try hal-0.4.0-2. Please see if it now works for you and if it does please close this bug. Thanks, David
It works if I always remember to umount it before unplugging ;-). IN fact I was a little surprised, when it umounted it automatically -- which worked for the first 2 times. The third time it didn't, I got a lot of SCSI errors in the log and when trying to umount it manually I got a pile of stuck processes: root@gibraltar:~> ps auxw|awk '{if ($8~"D") print}' root 3322 0.1 0.5 9048 6156 ? Ds 16:57 0:01 hald root 6054 0.0 0.0 0 0 ? D 17:04 0:00 [scsi_eh_4] root 6813 0.0 0.0 3640 648 pts/3 D+ 17:08 0:00 umount /media/409A-06C3/ root@gibraltar:~> BTW: Is there a way to give the device a more obvious mount point, e.g. the old "/media/usbstick"? Versions I use: root@gibraltar:~> uname -a Linux gibraltar 2.6.8-1.610 #1 Tue Oct 12 17:32:34 EDT 2004 i686 i686 i386 GNU/Linux root@gibraltar:~> rpm -q hal udev hal-0.4.0-2 udev-038-2
> It works if I always remember to umount it before unplugging ;-). IN > fact I was a little surprised, when it umounted it automatically -- > which worked for the first 2 times. The third time it didn't, I got a > lot of SCSI errors in the log and when trying to umount it manually I > got a pile of stuck processes: Sounds like a kernel bug to me - it would be nice if the kernel would do the right thing even if we don't cleanly unmount the device but it is my understanding this needs upstream kernel work. If you can reproduce this with cleanly unmounting the device every time then please reopen this bug because I'm going to close it as RAWHIDE. > /media/409A-06C3/ .. > BTW: Is there a way to give the device a more obvious mount point, > e.g. the old "/media/usbstick"? Certainly. See the fstab-sync man page for details. Btw, I've changed the defaults to not use the UUID so it will be /media/usbdisk%d or something. If you put a label on the filesystem and it meets certain criteria this will be used instead. See bug 135907 for details. Thanks, David