kexec-tools/kdump.sysconfig.ppc64le
Pingfan Liu 870ec2ec93 ppc64: tackle SRCU hang issue
Resolves: bz2158296
Upstream: RHEL-only

On PowerPC platform, the following hang is witnessed:

Welcome to
Red Hat Enterprise Linux 9.2 Beta (Plow) dracut-057-13.git20220816.el9 (Initramfs)
!

[    1.631210] systemd[1]: Hostname set to <ibm-p9z-18-lp11.virt.pnr.lab.eng.rdu2.redhat.com>.
[-- MARK -- Mon Sep 26 01:45:00 2022]
[  243.681283] INFO: task systemd:1 blocked for more than 122 seconds.
[  243.681303]       Not tainted 5.14.0-167.el9.ppc64le #1
[  243.681315] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[  243.681329] task:systemd         state:D stack:    0 pid:    1 ppid:     0 flags:0x00042000
[  243.681349] Call Trace:
[  243.681356] [c00000001a603640] [c00000004f990100] 0xc00000004f990100 (unreliable)
[  243.681378] [c00000001a603830] [c00000001001e9cc] __switch_to+0x12c/0x220
[  243.681400] [c00000001a603890] [c000000010ec5b40] __schedule+0x230/0x720
[  243.681418] [c00000001a603950] [c000000010ec6090] schedule+0x60/0x110
[  243.681435] [c00000001a603980] [c000000010ecd948] schedule_timeout+0x168/0x1c0
[  243.681454] [c00000001a603a60] [c000000010ec7214] __wait_for_common+0x134/0x360
[  243.681473] [c00000001a603b00] [c00000001017c98c] __flush_work.isra.0+0x1dc/0x3d0
[  243.681493] [c00000001a603ba0] [c0000000105cbd88] fsnotify_wait_marks_destroyed+0x28/0x40
[  243.681512] [c00000001a603bc0] [c0000000105cb800] fsnotify_destroy_group+0x60/0x150
[  243.681531] [c00000001a603c30] [c0000000105cf640] inotify_release+0x30/0xa0
[  243.681548] [c00000001a603ca0] [c00000001054fad8] __fput+0xc8/0x350
[  243.681565] [c00000001a603cf0] [c000000010183174] task_work_run+0xe4/0x160
[  243.681583] [c00000001a603d40] [c000000010021874] do_notify_resume+0x134/0x140
[  243.681602] [c00000001a603d70] [c000000010030168] interrupt_exit_user_prepare_main+0x198/0x270
[  243.681622] [c00000001a603de0] [c0000000100305ac] syscall_exit_prepare+0x6c/0x180
[  243.681641] [c00000001a603e10] [c00000001000bff4] system_call_vectored_common+0xf4/0x278
[  243.681661] --- interrupt: 3000 at 0x7fffb3015ba4
[  243.681673] NIP:  00007fffb3015ba4 LR: 0000000000000000 CTR: 0000000000000000
[  243.681687] REGS: c00000001a603e80 TRAP: 3000   Not tainted  (5.14.0-167.el9.ppc64le)
[  243.681703] MSR:  800000000000d033 <SF,EE,PR,ME,IR,DR,RI,LE>  CR: 42044440  XER: 00000000
[  243.681737] IRQMASK: 0
[  243.681737] GPR00: 0000000000000006 00007fffd24a31a0 00007fffb3127200 0000000000000000
[  243.681737] GPR04: 0000000000000002 000000000000000a 0000000000000000 0000000000000000
[  243.681737] GPR08: 0000010009ea2d40 0000000000000000 0000000000000000 0000000000000000
[  243.681737] GPR12: 0000000000000000 00007fffb3834bc0 0000000000000000 0000000000000000
[  243.681737] GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  243.681737] GPR20: 000000012c74ddf0 000000000000000e 000000000017cd3f 0000000000000000
[  243.681737] GPR24: 00007fffd24a3570 0000000000000005 0000010009eb5490 0000010009ea24e0
[  243.681737] GPR28: 0000010009ea2900 0000010009eb4850 0000010009ea2d70 00007fffb382dd98
[  243.681896] NIP [00007fffb3015ba4] 0x7fffb3015ba4
[  243.681907] LR [0000000000000000] 0x0
[  243.681917] --- interrupt: 3000
[  243.681928] INFO: task kworker/u16:1:34 blocked for more than 122 seconds.
[  243.681941]       Not tainted 5.14.0-167.el9.ppc64le #1
[  243.681951] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[  243.681964] task:kworker/u16:1   state:D stack:    0 pid:   34 ppid:     2 flags:0x00000800
[  243.681982] Workqueue: events_unbound fsnotify_mark_destroy_workfn
[  243.681998] Call Trace:
[  243.682005] [c00000001a9336d0] [c00000004f990100] 0xc00000004f990100 (unreliable)
[  243.682023] [c00000001a9338c0] [c00000001001e9cc] __switch_to+0x12c/0x220
[  243.682042] [c00000001a933920] [c000000010ec5b40] __schedule+0x230/0x720
[  243.682059] [c00000001a9339e0] [c000000010ec6090] schedule+0x60/0x110
[  243.682075] [c00000001a933a10] [c000000010ecd948] schedule_timeout+0x168/0x1c0
[  243.682094] [c00000001a933af0] [c000000010ec7214] __wait_for_common+0x134/0x360
[  243.682113] [c00000001a933b90] [c000000010213370] __synchronize_srcu.part.0+0xa0/0xe0
[  243.682132] [c00000001a933c00] [c0000000105cc154] fsnotify_mark_destroy_workfn+0xc4/0x1a0
[  243.682151] [c00000001a933c70] [c00000001017acb8] process_one_work+0x298/0x580
[  243.682169] [c00000001a933d10] [c00000001017b048] worker_thread+0xa8/0x630
[  243.682185] [c00000001a933da0] [c000000010188348] kthread+0x1b8/0x1c0
[  243.682203] [c00000001a933e10] [c00000001000cd64] ret_from_kernel_thread+0x5c/0x64
[  366.561279] INFO: task systemd:1 blocked for more than 245 seconds.

The right solution should be in kernel, but since the patch [1] for SRCU
will not be merged into the mainline in near future, it had better to
have a userspace workaround to overcome this test blocker.

The workaround method is to pass the kernel parameter "srcutree.big_cpu_lim=0", so
that the SRCU system will always use srcu_node array.

[1]: https://lore.kernel.org/rcu/20221026032716.78674-1-kernelfans@gmail.com/T/#m6534975507c2abca497a94d81c7abbfea1d0978d

Signed-off-by: Pingfan Liu <piliu@redhat.com>
2023-01-06 11:26:03 +08:00

59 lines
2.5 KiB
Plaintext

# Kernel Version string for the -kdump kernel, such as 2.6.13-1544.FC5kdump
# If no version is specified, then the init script will try to find a
# kdump kernel with the same version number as the running kernel.
KDUMP_KERNELVER=""
# The kdump commandline is the command line that needs to be passed off to
# the kdump kernel. This will likely match the contents of the grub kernel
# line. For example:
# KDUMP_COMMANDLINE="ro root=LABEL=/"
# Dracut depends on proper root= options, so please make sure that appropriate
# root= options are copied from /proc/cmdline. In general it is best to append
# command line options using "KDUMP_COMMANDLINE_APPEND=".
# If a command line is not specified, the default will be taken from
# /proc/cmdline
KDUMP_COMMANDLINE=""
# This variable lets us remove arguments from the current kdump commandline
# as taken from either KDUMP_COMMANDLINE above, or from /proc/cmdline
# NOTE: some arguments such as crashkernel will always be removed
KDUMP_COMMANDLINE_REMOVE="hugepages hugepagesz slub_debug quiet log_buf_len swiotlb hugetlb_cma ignition.firstboot"
# This variable lets us append arguments to the current kdump commandline
# after processed by KDUMP_COMMANDLINE_REMOVE
KDUMP_COMMANDLINE_APPEND="irqpoll maxcpus=1 noirqdistrib reset_devices cgroup_disable=memory numa=off udev.children-max=2 ehea.use_mcs=0 panic=10 kvm_cma_resv_ratio=0 transparent_hugepage=never novmcoredd hugetlb_cma=0 srcutree.big_cpu_lim=0"
# Any additional kexec arguments required. In most situations, this should
# be left empty
#
# Example:
# KEXEC_ARGS="--elf32-core-headers"
KEXEC_ARGS="--dt-no-old-root -s"
#Where to find the boot image
#KDUMP_BOOTDIR="/boot"
#What is the image type used for kdump
KDUMP_IMG="vmlinuz"
#What is the images extension. Relocatable kernels don't have one
KDUMP_IMG_EXT=""
#Specify the action after failure
# Logging is controlled by following variables in the first kernel:
# - @var KDUMP_STDLOGLVL - logging level to standard error (console output)
# - @var KDUMP_SYSLOGLVL - logging level to syslog (by logger command)
# - @var KDUMP_KMSGLOGLVL - logging level to /dev/kmsg (only for boot-time)
#
# In the second kernel, kdump will use the rd.kdumploglvl option to set the
# log level in the above KDUMP_COMMANDLINE_APPEND.
# - @var rd.kdumploglvl - logging level to syslog (by logger command)
# - for example: add the rd.kdumploglvl=3 option to KDUMP_COMMANDLINE_APPEND
#
# Logging levels: no logging(0), error(1),warn(2),info(3),debug(4)
#
# KDUMP_STDLOGLVL=3
# KDUMP_SYSLOGLVL=0
# KDUMP_KMSGLOGLVL=0