Description of problem: Accessing file on CDROM give "Input/Outut error" at prompt and "attempt to access beyond end of device 03:00: rw=0, want=393400, limit=10560" in dmesg. . Version-Release number of selected component (if applicable): it is a fresh install of Red hat enterprise AS v3. uname -a: Linux xxxxxx.xxxxx.xxxx 2.4.21-15.ELsmp #1 SMP Thu Apr 22 00:09:01 EDT 2004 x86_64 x86_64 x86_64 GNU/Linux How reproducible: I have this strange problem on all my new opteron boxes. They all have the same cdrom player (samsung ide cdrom sn-124), and when cp nedit-5.3-4.x86_64.rpm /tmp cp: reading `nedit-5.3-4.x86_64.rpm': Input/output error I installed the system from cdrom, it worked fine, but after the reboot I get strange errors. Fedora core 1 does not have the problem. Steps to Reproduce: 1. (auto)mount cdrom (for example 2nd installation cdrom) 2. cd /mnt/cdrom/RedHat/RPMS 3. ls (works fine) 4. cp nedit-5.3-4.x86_64.rpm /tmp cp: reading `nedit-5.3-4.x86_64.rpm': Input/output error Actual results: Emty file is created in /tmp and dmesg will show "attempt to access beyond end of device" errors. Expected results: file copied to /tmp Additional info: I can mount the cdroms and look at directory listings but when I copy files from it I get this error: Input/output error [root@xxxxxx root]# cd /mnt/cdrom/RedHat/RPMS/ [root@xxxxxx RPMS]# cp nedit-5.3-4.x86_64.rpm /tmp cp: reading `nedit-5.3-4.x86_64.rpm': Input/output error dmesg will also show beyond end of device errors: ISO 9660 Extensions: Microsoft Joliet Level 3 ISO 9660 Extensions: RRIP_1991A attempt to access beyond end of device 03:00: rw=0, want=393400, limit=10560 attempt to access beyond end of device 03:00: rw=0, want=393402, limit=10560 I installed fedora core 1, and everything works fine. All the 'faulty' drives and cdrom disks work fine again. While installing the system, it also read the same cdroms. So it is clearly a RHEL AS v3 x86_64 problem. mount show that the cdrom in mounted as an iso9660 type filesystem. [root@gerald4 RPMS]# mount /dev/cdrom on /mnt/cdrom type iso9660 (ro,nosuid,nodev)
I upgraded the fedora core 1 to the new fedora core 2 (x86_64) and the cdrom input/output error is not happening. So it must be a driver problem in AS v3. I tested some more in RHEL AS v3: it looks like if the system is booted with the cdrom inside, it works ok. I can mount read/ unmount, mount again and read again. But when I change the cdrom (open close drive) it will mount ok, give me directory listings, but I cannot copy RPM files from it. Same input/output error and "attempt to access beyond end of device" errors in dmesg. So it has also something to do with the size of the file that is accessed. What kind of tests/commands should I run to provide you guys with more technical input?
For starters, do a "dmesg -s 65536" and attach the output to this bug report. This will give us some idea of how the system sees the device at startup.
Created attachment 102605 [details] output of dmesg -s 65536 this is the dmesg on the RHEL AS v3. the cdrom driver is version 3.12. on fedore core 2 boot the cdrom driver loaded is version 3.20.
IMPORTANT UPDATE: This problem only happens in gnome. After booting to runlevel 3 the cdrom works fine. (So there is probably nothing wrong with the CDROM driver.) I try different cdrom options in the gnome control panel, unchecking automount, unchecking the various autorun options. But I have not been able to fix it. So I'm unable to use the add/remove applications function in gnome. It only works properly if the packages are on the current cd. After one cdrom change the problem keeps happening, even if I do an init 3 (after mounting a cdrom in gnome) and continue to mount and unmount from the command prompt at runlevel 3 I keep getting the input/output error in combination of the "attempt to access beyond end of device" errors in dmesg.
I think I have found the work-around. After the tip of Michael K. Johnson in bug 11144 I removed magicdev. And I have not had the problem anymore. Fedora core 2 did not have the problem, so I'm going to try to update magicdev and maybe my kernel. But for this production system I rather stay in line with the RHN updates, and not go for the bleeding edge. All suggestions on what si the wised path are welcome. I never expected this to happen with RHEL AS, I thouight it would be more stable than the fedora, but with my opteron server I had more problems with RHEL AS than with fedora. I even had to trade in my adaptec raid controller (see bug 128482) for a megaraid one to get it to work with RHEL AS.
Reassigning bug to the magicdev maintainer.
It seems that the only change was an upgrade from magicdev 1.1.4 to 1.1.6. I will take a closer look at the change set between the two version and see if I can backport a patch or perhaps just bump it up if not much has changed. Magicdev will be obsoleted in the future by gnome-volume-manager.
Looks like it is an issue between magicdev and the kernel because all the new functionality in magicdev-1.1.6 had been backported to the 1.1.4 package. Upgrading to the new magicdev will not fix this problem.
This bug is filed against RHEL 3, which is in maintenance phase. During the maintenance phase, only security errata and select mission critical bug fixes will be released for enterprise products. Since this bug does not meet that criteria, it is now being closed. For more information of the RHEL errata support policy, please visit: http://www.redhat.com/security/updates/errata/ If you feel this bug is indeed mission critical, please contact your support representative. You may be asked to provide detailed information on how this bug is affecting you.