Red Hat Bugzilla – Bug 228946
I2o driver (Adaptec 2400 et al) not loaded properly
Last modified: 2007-11-30 17:11:57 EST
Description of problem: I'm using Adaptec 2400A raid card with raid-5 configured. The I2o driver not work properly in new kernel but worked in 2.6.18-1.2798.fc6. Version-Release number of selected component (if applicable): kernel-2.6.19-1.2911.fc6 How reproducible: When everytime boot in new kernel, the error messages shown Buffer I/O error on device i2o/hda, logical block 183148784 Buffer I/O error on device i2o/hda, logical block 183148784 Buffer I/O error on device i2o/hda, logical block 0 Buffer I/O error on device i2o/hda, logical block 1 Buffer I/O error on device i2o/hda, logical block 2 Buffer I/O error on device i2o/hda, logical block 3 Buffer I/O error on device i2o/hda, logical block 4 and can't fdisk or mount the partition Steps to Reproduce: 1. 2. 3. Actual results: checked with the modules i2o_block not used with lsmod Module Size Used by i2o_block 17613 0 i2o_core 45029 1 i2o_block Expected results: i2o_block module locked and not be removed using rmmod. Additional info:
I just tried to update an i2o-based machine from FC5 to FC6 (using yum). The new system, with kernel 2.6.19-1.2911.fc6, doesn't seem to cope with having its root filesystem on an i2o device. (I have it temporarily working with the old FC5 kernel.) I have another system with the same hardware running 2.6.18-1.2869.fc6 fine, so I'm wondering if there's a problem with i2o on 2.6.19-1.2911.fc6.
*** Bug 228973 has been marked as a duplicate of this bug. ***
Can you find the messages in your syslog from when the driver loads? Does it throw any error messages? And please post the corresponding log from a working kernel.
Created attachment 148282 [details] Kernel dmesg output from 2.6.19-1.2911.fc6.x86_64 I have just been bitten by this bug. I'm running FC6 (x86_64) with my root filesystem connected to the onboard IDE controller on my mainboard with a separate RAID-5 array on the Adaptec 2400A card (uses the i2o driver). While I can see the device node for the single array configured on the card (/dev/i2o/hda), I cannot read its' partition table via fdisk nor mount any filesystems on it. Please find attached the kernel dmesg output from the most recent FC6 kernel update that appears to cause the behaviour already described in this bug. (I'll be attaching the syslog messages generated by this kernel in a few seconds)
Created attachment 148283 [details] Kernel syslog output from 2.6.19-1.2911.fc6.x86_64 I have attached the syslog entries generated by 2.6.19-1.2911.fc6.x86_64.
Created attachment 148284 [details] Kernel dmesg output from 2.6.18-1.2869.fc6.x86_64 I am attaching the dmesg output from 2.6.18-1.2869.fc6.x86_64 as this is the last known kernel which works correctly with my Adaptec 2400A card. Expect the matching syslog within the next few seconds...
Created attachment 148285 [details] Kernel syslog output from 2.6.18-1.2869.fc6.x86_64 ... and the syslog as promised! I am happy to test any patches submitted to this bug as this box isn't a production system that needs to stay up 24/7.
I've just switched my system from 2.6.18-1.2257.fc5smp to 2.6.18-1.2869.fc6 with no problems, so it definitely looks like a 2.6.19 problem. (If I get a test machine put together I'll try 2.6.19-1.2288.fc5smp - I expect that to fail.)
*** Bug 229614 has been marked as a duplicate of this bug. ***
Problem has been reported to linux-kernel mailing list -- no response.
The bug continues to persist in the 2.6.20 kernel. I've been relying on the 2.6.18 kernel for quite a while now in order to get my system to boot.
Someone has posted a patch which apparently fixes this bug to the linux-kernel mailing list. Details here: http://marc.info/?l=linux-kernel&m=117489656029021&w=2 While I am still happy to test a kernel from Red Hat, I will likely spin my own binary RPM of the most recent FC6 kernel SRPM with this patch applied to it. I will post an update to this bug if the above patch does indeed work.
Go ahead and test. The patch will be in the next Fedora test kernel as well.
The aforementioned patch works for me (x86_64) against 2.6.20-1.2933. If you wish for me to test an official Fedora test kernel, please let me know.
Is this fixed in 2.6.20-1.2944?