Fedora Account System
Red Hat Associate
Red Hat Customer
LTC Owner is: ed_pollard.com Problem description: kernel BUG in shrink_dcache_for_umount_subtree at fs/dcache.c:615! Aug 29 snapshot of rawhide: kernel-2.6.17-1.2597.fc6.ppc64.rpm development/source/SRPMS/ kernel-2.6.17-1.2597.fc6.src.rpm anaconda-11.1.0.84-1.src.rpm Hardware Environment Machine type (p650, x235, SF2, etc.): OpenPower720 (9124-720) Cpu type (Power4, Power5, IA-64, etc.):Power5 Describe the steps: choose dhcp for eth0 and start an NFS install (recreates with ftp install too) welcome to Fedora screen (ok) Warning: this is a beta (choose: install anyway) anaconda starts to run and lpar crashes. Additional information: ┌──────────────────────┤ Searching ├──────────────────────�? │ │ │ Searching for Fedora Core installations... │ │ │ │ 0% cpu 0x3: Vector: 700 (Program Check) at [c0000000484c3690] │ pc: c0000000000ff550: .shrink_dcache_for_umount_subtree+0x204/0x304 lr: c0000000000ff54c: .shrink_dcache_for_umount_subtree+0x200/0x304 sp: c0000000484c3910 msr: 8000000000029032 current = 0xc00000008556a250 paca = 0xc000000000494980 pid = 914, comm = anaconda kernel BUG in shrink_dcache_for_umount_subtree at fs/dcache.c:615! enter ? for help 3:mon> [c0000000484c39a0] c00000000010036c .shrink_dcache_for_umount+0x50/0x74 [c0000000484c3a20] c0000000000e8620 .generic_shutdown_super+0x3c/0x1b0 [c0000000484c3ab0] c0000000000e87c4 .kill_block_super+0x30/0x60 [c0000000484c3b40] c0000000000e8980 .deactivate_super+0xd4/0x128 [c0000000484c3bd0] c000000000105ac4 .mntput_no_expire+0x84/0x10c [c0000000484c3c60] c0000000000f15ec .path_release_on_umount+0x30/0x48 [c0000000484c3cf0] c00000000010671c .sys_umount+0x300/0x36c [c0000000484c3e30] c00000000000871c syscall_exit+0x0/0x40 --- Exception: c01 (System Call) at 000000000fbdf438 SP (fa16f150) is in userspace 3:mon> r R00 = c0000000000ff54c R16 = 00000000f7de9e90 R01 = c0000000484c3910 R17 = 00000000f7d52260 R02 = c0000000005b2c80 R18 = 0000000000000000 R03 = 0000000000000062 R19 = 00000000100201b8 R04 = 8000000000001032 R20 = 0000000000000000 R05 = 0000000000000000 R21 = 00000000f7d540ad R06 = 0000000000000000 R22 = 0000000010048f90 R07 = 0000000000000000 R23 = 00000000100490e8 R08 = 0000000000000002 R24 = 000000000fea41b0 R09 = c000000000650000 R25 = 0000000000000000 R10 = 000000001c00000f R26 = 0000000010048f90 R11 = 0000000000000000 R27 = 0000000000000000 R12 = 0000000000004000 R28 = c0000000874f7450 R13 = c000000000494980 R29 = c000000086dec528 R14 = 00000000ffffffff R30 = c0000000004e7fa0 R15 = 00000000f7d54034 R31 = c000000086dec428 pc = c0000000000ff550 .shrink_dcache_for_umount_subtree+0x204/0x304 lr = c0000000000ff54c .shrink_dcache_for_umount_subtree+0x200/0x304 msr = 8000000000029032 cr = 24228882 ctr = 80000000001c5840 xer = 0000000000000000 trap = 700 I set this to anaconda because I can't find anything closer for an install problem, please let me know what component would better suit to avoid misdirection in teh future if this is incorrect.
----- Additional Comments From marksmit.com 2006-08-30 19:39 EDT ------- This again recreates on multiple Power5 systems using todays Aug 30 snapshot of rawhide.
setting the priority to high to be consistent with the severity field.
possibly this is a duplicate of http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=200645
given this bug is found in 2597.fc6, and the above was fixed by 2505.fc6, this looks related but not identical to 200645. Adding davej to cc list, as he posted the fix for 200645.
changed: What |Removed |Added ---------------------------------------------------------------------------- Component|Installation and Packaging |Kernel ------- Additional Comments From marksmit.com 2006-08-31 16:40 EDT ------- changing component to kernel to match similar bug 200645
----- Additional Comments From marksmit.com 2006-09-01 13:41 EDT ------- recreates on Aug31 and Sept1 snapshots of rawhide
Changing to proper owner, kernel-maint.
changed: What |Removed |Added ---------------------------------------------------------------------------- Severity|block |high ------- Additional Comments From marksmit.com 2006-09-13 17:37 EDT ------- I have experimented enough to see that this still recreates on the Sept 11 version of rawhide. However, I can make it come or go by putting a reiserfs partition on the disk. Beginning the install process, when anaconda searches for existing Fedora Core installations, if it encounters a reiserfs partition, this problem recreates. If I boot in rescue mode and use parted to remove the reiserfs partition, then this problem does not recreate.
----- Additional Comments From marksmit.com 2006-09-13 19:52 EDT ------- reducing ltc severity to high from blocker. I have verified that this recreates on Fedora Core6 - test3. However, I can make it come or go by putting a reiserfs partition on the disk. Beginning the install process, when anaconda searches for existing Fedora Core installations, if it encounters a reiserfs partition, this problem recreates. If I boot in rescue mode and use parted to remove the reiserfs partition, then this problem does not recreate.
*** This bug has been marked as a duplicate of 206036 ***