[0/9] fix softlockup in run_timer_softirq

Message

liujian (CE) May 5, 2023, 11:33 a.m. UTC

I encountered the run_timer_softirq lockup issue[1] many times during fuzz
tests. We analyze __run_timers() and find the following problem.

In the while loop of __run_timers(), because there are too many timers or
improper timer handler functions, if the processing time of the expired
timers is always greater than the time wheel's next_expiry, the function
will loop infinitely.

Here base on Peter's softirq_needs_break patchset[2], use the timeout/break
logic jump out of the loop.

[1] https://lore.kernel.org/lkml/[e-mail]/
[2] https://git.kernel.org/pub/scm/linux/kernel/git/peterz/queue.git/log/?h=core/softirq

Liu Jian (1):
  softirq, timer: Use softirq_needs_break()

Peter Zijlstra (8):
  softirq: Rewrite softirq processing loop
  softirq: Use sched_clock() based timeout
  softirq: Factor loop termination condition
  softirq: Allow early break
  softirq: Context aware timeout
  softirq: Provide a softirq_needs_break() API
  softirq,net: Use softirq_needs_break()
  softirq,rcu: Use softirq_needs_break()

 Documentation/admin-guide/sysctl/net.rst | 11 +--
 include/linux/interrupt.h                |  5 ++
 kernel/rcu/tree.c                        | 29 ++-----
 kernel/rcu/tree_nocb.h                   |  2 +-
 kernel/softirq.c                         | 97 +++++++++++++++---------
 kernel/time/timer.c                      | 12 ++-
 net/core/dev.c                           |  6 +-
 net/core/dev.h                           |  1 -
 net/core/sysctl_net_core.c               |  8 --
 9 files changed, 86 insertions(+), 85 deletions(-)