|Summary:||install onto i2o RAID5 volume fails with "Unable to get HRT" error|
|Product:||Red Hat Enterprise Linux 4||Reporter:||Tim Mooney <mooney>|
|Component:||kernel||Assignee:||Tom Coughlan <coughlan>|
|Status:||CLOSED WONTFIX||QA Contact:||Brian Brock <bbrock>|
|Version:||4.0||CC:||dledford, jbaron, markus.lidel, n.marjoram, rhbugzillamarcw|
|Fixed In Version:||Doc Type:||Bug Fix|
|Doc Text:||Story Points:||---|
|Last Closed:||2012-06-20 13:32:49 UTC||Type:||---|
|oVirt Team:||---||RHEL 7.3 requirements from Atomic Host:|
|Cloudforms Team:||---||Target Upstream Version:|
Description Tim Mooney 2005-06-26 16:23:01 UTC
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; OSF1 alpha; en-US; rv:1.7.7) Gecko/20050418 Firefox/1.0.3 Description of problem: We have an Intel SCB2 system with Adaptec AIC 7899 (firmware: 2.57S4) and Adaptec I2O RAID controller (firmware: v001.3W). It's currently running Red Hat AS 3, update 5. Everything is installed on a single RAID5 volume. I attempted to re-install the system with RHEL AS 4, and the install failed. Even booting off the RHEL AS 4 CD into "rescue" mode, the kernel loads both the i2o and aic7xxx drivers, but it can't see any disk devices. Looking through the dmesg output, I see the following messages: iop0: Unable to get HRT (status=0x6e) iop0: controller could not activated I2O controller: probe of 0000:01:08.1 failed with error -110 Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: Boot AS 4 in rescue mode, or try install AS 4 on the system. Actual Results: Kernel can't see the single RAID5 volume that the controller is presenting. Expected Results: kernel sees the "hard drive" (really a RAID5 volume) and I partition it and proceed happily. Additional info: Available upon request.
Comment 1 Tom Coughlan 2005-06-29 13:43:58 UTC
Markus, The i2o_iop_activate failed with "iop0: Unable to get HRT (status=0x6e)". Can you decode that for us? Thanks, Tom
Comment 2 Markus Lidel 2005-06-29 15:14:24 UTC
Hello, the status 0x6e means that the function to get the HRT timed out... The timeout is the same as in the dpt_i2o module, so i think it's not a problem of a too short timeout. It could be probably because the aic7xxx driver tried to access the controller too. Best regards, Markus Lidel
Comment 3 Tom Coughlan 2005-06-29 18:24:33 UTC
Another possibility, if this is the first command issued to the board, and the first interrupt we are waiting to recieve, then it might indicate something wrong with interrupt routing. Is that possible Markus?
Comment 4 Markus Lidel 2005-06-29 18:38:56 UTC
Hello, indeed the command is the first which uses interrupts, so it is possible :-) Best regards, Markus Lidel
Comment 5 Tim Mooney 2005-07-01 20:42:38 UTC
We have several boxes with this exact config, including one that's not a production box. I tried flashing the firmware on the I2O controller on that non-production box to the latest version available from Adaptec (v1.62, from 2003). That updated it from v001.3W to v001.62. I booted the Red Hat AS 4 CD via "linux rescue". I get the same error on this test box, even after flashing the firmware to the latest version.
Comment 6 Marc Wallman 2005-08-30 15:19:44 UTC
Is there anything happening on this bug??
Comment 7 Neil Marjoram 2005-09-11 11:35:02 UTC
I have the similar problem. However it has presented itself slightly differently. I have two Dell 410 Precision Workstations with an Adaptec 2400A, both ran FC3 perfectly with these cards, I reinstalled with RHEL4 no problem, running the OS off the RAID on seperate disks, I remounted the old filesystems and away they went. These boxes are due to go into production with a slightly different filesystem setup, so I removed the partitions and recreated what I needed, made the filesystem and mounted up OK. I always like to reboot to check everything reloads ok, but it didn't, it failed with the above errors. Both boxes are patched to the same level, the one I have not changed yet reboots ok. No devices are created in /dev/ for i2o/hda1 or hda2. The RAID checks out OK in the bios.
Comment 8 Markus Lidel 2005-09-13 11:06:35 UTC
Hello, the timeout for retreiving all devices was in kernel <= 2.6.9 too short for some controllers. In 2.6.10 the timeout was increased to the same value as in dpt_i2o. Could you tell me which kernel version you use? Thank you very much. Best regards, Markus Lidel
Comment 9 Tim Mooney 2005-09-13 17:58:45 UTC
Are you asking for Neil's kernel version or ours? We're using the default Red Hat AS 4 kernel, as part of the "update 1" release. That would be 2.6.9 + whatever patches Red Hat has backported. Although Neil is getting the same error message, it's happening under different circumstances. We can't even install Red Hat AS 4 update 1 on a box that's currently running Red Hat 3 AS (update 5).
Comment 10 Markus Lidel 2005-09-13 19:24:48 UTC
Hello, then it's most likely that the patch is missing. It is easy to fix this bug in a running system, but it is a little bit complicated to do it when booting from CD. In any case you have to increase the timeout in the kernel header "include/linux/i2o.h" under the kernel source tree. There should be a line #define I2O_TIMEOUT_LCT_GET 20 and this should be changed into #define I2O_TIMEOUT_LCT_GET 360 After you compile and install the new kernel everything should be fine. But when booting from CD you have to download the module from another machine somehow. You may find this link usefull. http://www.redhat.com/archives/fedora-list/2004-May/msg03906.html It describes how you could download the module (in your case it's i2o_core.ko and not i2o_proc.ko). Hope i could help you a little bit. If you have further questions, please feel free to contact me. Best regards, Markus Lidel
Comment 11 Tim Mooney 2005-09-13 19:50:45 UTC
Thanks for the link, that looks very useful. It may be a couple weeks before we have a down-time window to test this, but I will test and report back. Mr. Coughlan, I assume this is something Red Hat is looking at backporting into a future kernel release for AS 4?
Comment 12 Tom Coughlan 2005-09-13 20:20:01 UTC
The patch to increase the timeout from 20 to 360 is in RHEL 4 U2 (BZ 140002). This is available in beta now, and will be final later this month. U2 includes install media, so it will solve that issue as well. I will leave this BZ open until we have confirmation that this patch fixes the problems reported here. Please provide an update when you have tested it. Thanks.
Comment 13 Tim Mooney 2005-09-13 21:42:05 UTC
Trying the 4 AS U2 beta sounded like the easier route, so I downloaded ISO disc 1 for RHEL4 AS U2 from the "Betas" channel of rhn.redhat.com. I tested the install from the 4 AS U2 CD on a non-production system we have that also has the I2O card, and the problem is NOT fixed. The installer still gets the same error, and it still cannot see any devices to install onto. Is it possible that the install kernel on the ISO that's on RHN hasn't yet picked up the patch?
Comment 14 Markus Lidel 2005-09-22 20:01:50 UTC
Hello, today another user has reported that he tried out U2 and it worked. Could you try it out again? Best regards, Markus Lidel
Comment 15 Tim Mooney 2005-09-26 23:21:14 UTC
Using the same CD I mentioned in comment #13, I tried another system with an I2O card, and received the same problem. I'll try fetching the AS 4 U2 beta image again, to make sure I have the correct ISO. If U2 is going to be released to production in the next week, I may wait until after that's released before doing much more testing. As far as I can tell right now, the problem is not fixed using the stock U2 beta. I have not yet tried the custom module method that Markus mentions in comment #10, simply because using the U2 beta ISO seemed a lot easier.
Comment 16 Tim Mooney 2005-10-10 18:40:42 UTC
I just tested today with the released RH AS 4 u2 disc 1, and the problem is still there. If I look at the i2o.h header that's installed as part of the Red Hat 2.6.9-22.EL kernel, I can see that the change is included in that header, but either the kernel on the disc1 ISO is older than 2.6.9-22.EL or the timeout still isn't long enough. If I drop to an ALT-F2 prompt after the installer tells me it couldn't find any disks to install onto, I can see that i2o_core and i2o_block are loaded. If I rmmod them, fetch versions (kernel 2.6.9-22.EL) of those modules from a different system and then modprobe them and go back into the installer, then the installer can see system storage and I can proceed. So, using the workaround that Markus provided in comment #10, it should be possible to proceed. Tom's comment #10 appears to be only partially correct; it looks like the fix did make it into U2, but the fix appears to *NOT* be included in the install media (even though the kernel says it's 2.6.9-22.EL), because the problem is still there for the install off the U2 media.
Comment 17 Tom Coughlan 2007-02-19 22:09:21 UTC
Please try this again with RHEL 4.4. Thanks.
Comment 18 Jiri Pallich 2012-06-20 13:32:49 UTC
Thank you for submitting this issue for consideration in Red Hat Enterprise Linux. The release for which you requested us to review is now End of Life. Please See https://access.redhat.com/support/policy/updates/errata/ If you would like Red Hat to re-consider your feature request for an active release, please re-open the request via appropriate support channels and provide additional supporting details about the importance of this issue.