gh0s7/mckernel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

runq_lock: Fix deadlock due to cpu migration.

Browse Source 
Symptom and analysis:
runq_lock of the migration source is acquired on
the migration destination CPU.

This happens in the following steps:
 (1) The thread stores value of cpu_local_var(runq_lock)
     to its register when trying to perform
     ihk_mc_spinlock_lock() on the lock variable.
 (2) The thread takes IPI and migrates to another CPU.
 (3) The thread resumes execution and acquires the wrong lock.

Solution:
* Disable interrupts before getting the value of
  cpu_local_var(runq_lock)

Change-Id: Ia0ea450b97f872dd6116252537e4a79f85adfc88
Refs: #1400
This commit is contained in:
Shiratori, Takehiro
2020-02-13 13:02:40 +09:00
committed by Masamichi Takagi
parent 1a204b6674
commit edf7b36669
4 changed files with 38 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
kernel/syscall.c
View File

@ -251,8 +251,9 @@ long do_syscall(struct syscall_request *req, int cpu)
/* Spin by default, but if re-schedule is requested let
* the other thread run */
runq_irqstate =
ihk_mc_spinlock_lock(&(get_this_cpu_local_var()->runq_lock));
runq_irqstate = cpu_disable_interrupt_save();
ihk_mc_spinlock_lock_noirq(
&(get_this_cpu_local_var()->runq_lock));
v = get_this_cpu_local_var();
if (v->flags & CPU_FLAG_NEED_RESCHED ||
@ -261,7 +262,8 @@ long do_syscall(struct syscall_request *req, int cpu)
do_schedule = 1;
}
ihk_mc_spinlock_unlock(&v->runq_lock, runq_irqstate);
ihk_mc_spinlock_unlock_noirq(&v->runq_lock);
cpu_restore_interrupt(runq_irqstate);
if (!do_schedule) {
continue;
@ -2672,9 +2674,13 @@ end:
ihk_mc_free_pages(desc, 4);
if (!ret) {
unsigned long irqstate;
/* Lock run queue because enter_user_mode expects to release it */
cpu_local_var(runq_irqstate) =
ihk_mc_spinlock_lock(&(get_this_cpu_local_var()->runq_lock));
irqstate = cpu_disable_interrupt_save();
ihk_mc_spinlock_lock_noirq(
&(get_this_cpu_local_var()->runq_lock));
cpu_local_var(runq_irqstate) = irqstate;
preempt_enable();
ihk_mc_switch_context(NULL, &thread->ctx, thread);
@ -4957,15 +4963,17 @@ do_sigsuspend(struct thread *thread, const sigset_t *set)
long runq_irqstate;
thread->status = PS_INTERRUPTIBLE;
runq_irqstate =
ihk_mc_spinlock_lock(&(get_this_cpu_local_var()->runq_lock));
runq_irqstate = cpu_disable_interrupt_save();
ihk_mc_spinlock_lock_noirq(
&(get_this_cpu_local_var()->runq_lock));
v = get_this_cpu_local_var();
if (v->flags & CPU_FLAG_NEED_RESCHED) {
do_schedule = 1;
}
ihk_mc_spinlock_unlock(&v->runq_lock, runq_irqstate);
ihk_mc_spinlock_unlock_noirq(&v->runq_lock);
cpu_restore_interrupt(runq_irqstate);
if (do_schedule) {
schedule();