LTC Owner is: thinh.com LTC Originator is: marksmit.com Problem description: kernel panic starting install. recreates with FC5 kernel through today's May15, rawhide ppc64.img netboot image. Does not recreate with Rhel4U3. Recreates with/without "nodmraid" ... md: Autodetecting RAID arrays. md: autorun ... md: ... autorun DONE. RAMDISK: Compressed image found at block 0 RAMDISK: ran out of compressed data invalid compressed format (err=1) Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(1,0) Call Trace: [C0000000030B3C30] [C00000000000ED40] .show_stack+0x68/0x1b0 (unreliable) [C0000000030B3CD0] [C00000000005C680] .panic+0x84/0x218 [C0000000030B3D70] [C0000000003FCEE0] .mount_block_root+0x254/0x28c [C0000000030B3E30] [C0000000003FD094] .prepare_namespace+0xf4/0x164 [C0000000030B3EC0] [C0000000000095B4] .init+0x2e8/0x3c0 [C0000000030B3F90] [C000000000025BF0] .kernel_thread+0x4c/0x68 P5 570 - floralp2.upt.austin.ibm.com ethernet adapter is 1st/only boot device in the SMS bootlist. to recreate, reboot lpar, hit '8' during sms menu to goto openfirmware prompt. of> boot <add any extra boot commands you want> access: telnet sqh7lte.upt (hscroot : abc123) chsysstate -m fsp-flora --id 2 -o shutdown -r lpar --immed --restart rmvterm -m fsp-flora --id 2 (note: that is two dashes in front of id) mkvterm -m fsp-flora --id 2 ~. <enter> tilde, dot to exit from vterm. Lpar recreates in 2minutes. I left it at the open firmware prompt, ready for a recrate. (first time just do the mkvterm line, no need for the chsysstate) Awaiting suggestions on how to debug. >RAMDISK: ran out of compressed data >invalid compressed format (err=1) this is the same as RHEL4 U3 Beta bug 20403 - RIT 85890 the patch is: https://enterprise.redhat.com/issue-tracker/?module=download&fid=53180 on IBM LTC: https://bugzilla.linux.ibm.com/show_bug.cgi?id=20403#c58 since this also recreates on FC5 as well as rawhide, I suspect this is not upstream. ie. The version in rawhide today is yaboot-1.3.13-0.18.src.rpm from March6,2006. That is the same version as is in the March "gold" publish of FC5. I tried going here but search does not find yaboot.c http://sosdg.org/~coywolf/lxr/find?a=ppc&string=yaboot.c So I assume the patch is made _not_ in the kernel, but only in the yaboot SRPM.
----- Additional Comments From marksmit.com 2006-05-17 16:49 EDT ------- As this problem was encountered on JS21 on RHEL4U3, there appears to be a patch dropped into RHEL4U3. yaboot-1.3.12/second/yaboot.c What is unique about this lpar is that it 1) recreated for the first time on a power5 systems, not JS21, and 2) there is an existing AIX install plus RHEL4 install on the disks for this lpar. Whether or not that is pertinent, I do not know, because I am not sure how far along in the netboot process we are when meeting the kernel panic.
Please refer to patches by name not by file - as it's impossible to guess what you mean from the context. If you have AIX installed has it changed load-base - can you boot into OF and check that please. Also please do not switch between a bug report on rawhide/Fedora development and discussing RHEL (other than to point out regressions) as they are seperate bug streams with different processes. This bug was filed against Fedora please lets keep it focussed.
Manoj - please can you try and replicate with a rawhide tree here.
----- Additional Comments From thinh.com(prefers email via th2tran.com) 2006-05-19 15:57 EDT ------- the patch name is reduce-initrd-alloc-increments.patch the link to the patch from RedHat issue tracker #85890: http://bugmail.austin.ibm.com/rhitLoginGetAttachFr.php?&bug_id=191995&fid=53180
changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |natlynch.com Version|FC5 |Other ------- Additional Comments From marksmit.com 2006-05-27 00:22 EDT ------- > If you have AIX installed has it changed load-base - can you boot into OF and > check that please. I am unsure of the OF command syntax to discover this. If you know it, please append; else, I will ask around. I have verified this is easily recreatable, after booting one of the Linux installations and after attempting the AIX installation. But in all our testing , it only seems to recreate on this one power5 lpar. I am changing the version on the LTC Bugzilla from FC5 to "other" to reflect that this is found on Fedora-devel (rawhide). The status remains the same. This continues to recreate on today's May26 snapshot of Fedora-devel. Here is a picture of the drives and bootable partitions on this lpar. from SMS, the 1st boot device, used to network boot: 7. 1 Ethernet ( loc=U7879.001.0395562-P1-C1-T1 ) (1. = Virtual eth, 2-5, 7-10 are physical eth NICs, 6=IDE cdrom) from SMS, the bootable disk devices 11,12,13: 11. - SCSI 73407 MB Harddisk, part=2 (AIX 5.3.0) ( loc=U7879.001.0395562-P1-T14-L3-L0 ) 12. - SCSI 73407 MB Harddisk, part=1 () ( loc=U7879.001.0395562-P1-T14-L4-L0 ) 2.6.9-20 installation 13. - SCSI 73407 MB Harddisk, part=1 () ( loc=U7879.001.0395562-P1-T12-L3-L0 ) 2.6.9-22 installation And from open firmware: of> scsiinfo all the disks visible: 1 U7879.001.0395562-P1-T14 /pci@800000020000001/pci@2/pci1069,b166@1/scsi@0 ID 3 - 73407 MB Disk drive (bootable) (AIX 5.3.0) (VIO server) (does not boot as floralp2 is not configured as a VIOS partition, but instead is an RPA lpar) ID 4 - 73407 MB Disk drive (bootable) (2.6.9-20) 4 U7879.001.0395562-P1-T12 /pci@800000020000003/pci@2,2/pci1069,b166@1/scsi@0 ID 3 - 73407 MB Disk drive (bootable) (2.6.9-22) ID 4 - 73407 MB Disk drive ID 5 - 73407 MB Disk drive
changed: What |Removed |Added ---------------------------------------------------------------------------- Version|Other |devel ------- Additional Comments From marksmit.com 2006-06-06 00:14 EDT ------- I am changing the version to 'devel' to properly reflect that this recreates on rawhide. This still recreates on the one lpar: floralp2 using today's June5 version of rawhide.
changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |REJECTED Resolution| |UNREPRODUCIBLE ------- Additional Comments From marksmit.com 2006-08-30 19:53 EDT ------- After loaning the victim system to Enterprise Linux testing, and getting it back, I can no longer recreate this on floralp2 (vio server lpar, running Linux). I will have to cancel this as unreproducible. I tried to recreate using several versions of OS that formerly all recreated.
Closing Red Hat side based on comment #7