Description of problem: Customer reported that on his Altix with a ql12160 the performance of the attached devices significantly drops after a bus reset had occured. The problem is reproducible. Here the log of what has been done. Version-Release number of selected component (if applicable): How reproducible: Reliably. Steps to Reproduce: 1. Reboot system 2. time dd if=/dev/sdb of=/dev/null bs=64k count=10000 10000+0 records in 10000+0 records out real 0m9.543s user 0m0.011s sys 0m0.577s As you see it takes about 10 seconds to complete the transfer. 3. Now doing a bus reset. taclab26:~ # sg_reset -b /dev/sg1 sg_reset: starting bus reset sg_reset: completed bus reset Actual results: After the reset completed the transfer of the 10000 blocks takes about 7.5 times longer as before. taclab26:~ # time dd if=/dev/sdb of=/dev/null bs=64k count=10000 10000+0 records in 10000+0 records out real 1m14.443s user 0m0.007s sys 0m0.595s Expected results: faster :) Additional info: Fix is changeset 40552 by Jes Sorensen
Created attachment 142488 [details] Kernel changeset 40552
I assume this is reported against rhel4.
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux maintenance release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux Update release for currently deployed products. This request is not yet committed for inclusion in an Update release.
It should be applied to RHEL4.5 and RHEL5.0. I suspect I should have created two BZs.
Yes, please dup this to RHEL5 and resubmit the patch for RHEL5.
Associating this with PV#957504 and added CC's. I tried to change the product to Public Beta to clarify that we are interested in RHEL 4.5, but was told only the submitter or owner could do that and since I am both it is therefore not allowed :-(.
QE ack for RHEL4.5.
Requesting an exception. There is a mis-placed closing paren in the qla1280 driver. - if (qla1280_bus_reset(ha, bus == 0)) + if (qla1280_bus_reset(ha, bus) == 0) This results in reduced performance after a bus reset. The fix is clearly correct, it has been upstream for quite a while, and it has been ACKed on rhkernel-list. This driver is for old hardware that is not widely used by RHEL customers. The bug has apparently been present for quite a while. So, we could defer this, but that does not seem necessary in this case.
committed in stream U5 build 51. A test kernel with this patch is available from http://people.redhat.com/~jbaron/rhel4/
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2007-0304.html