Commit da9647e0 authored by Peter Zijlstra's avatar Peter Zijlstra Committed by Ingo Molnar

sched/completions: Fix complete_all() semantics

Documentation/scheduler/completion.txt says this about complete_all():

  "calls complete_all() to signal all current and future waiters."

Which doesn't strictly match the current semantics. Currently
complete_all() is equivalent to UINT_MAX/2 complete() invocations,
which is distinctly less than 'all current and future waiters'
(enumerable vs innumerable), although it has worked in practice.

However, Dmitry had a weird case where it might matter, so change
completions to use saturation semantics for complete()/complete_all().
Once done hits UINT_MAX (and complete_all() sets it there) it will
never again be decremented.
Requested-by: default avatarDmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: default avatarPeter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: der.herr@hofr.at
Cc: linux-kernel@vger.kernel.org
Signed-off-by: default avatarIngo Molnar <mingo@kernel.org>
parent 59f8c298
...@@ -31,7 +31,8 @@ void complete(struct completion *x) ...@@ -31,7 +31,8 @@ void complete(struct completion *x)
unsigned long flags; unsigned long flags;
spin_lock_irqsave(&x->wait.lock, flags); spin_lock_irqsave(&x->wait.lock, flags);
x->done++; if (x->done != UINT_MAX)
x->done++;
__wake_up_locked(&x->wait, TASK_NORMAL, 1); __wake_up_locked(&x->wait, TASK_NORMAL, 1);
spin_unlock_irqrestore(&x->wait.lock, flags); spin_unlock_irqrestore(&x->wait.lock, flags);
} }
...@@ -51,7 +52,7 @@ void complete_all(struct completion *x) ...@@ -51,7 +52,7 @@ void complete_all(struct completion *x)
unsigned long flags; unsigned long flags;
spin_lock_irqsave(&x->wait.lock, flags); spin_lock_irqsave(&x->wait.lock, flags);
x->done += UINT_MAX/2; x->done = UINT_MAX;
__wake_up_locked(&x->wait, TASK_NORMAL, 0); __wake_up_locked(&x->wait, TASK_NORMAL, 0);
spin_unlock_irqrestore(&x->wait.lock, flags); spin_unlock_irqrestore(&x->wait.lock, flags);
} }
...@@ -79,7 +80,8 @@ do_wait_for_common(struct completion *x, ...@@ -79,7 +80,8 @@ do_wait_for_common(struct completion *x,
if (!x->done) if (!x->done)
return timeout; return timeout;
} }
x->done--; if (x->done != UINT_MAX)
x->done--;
return timeout ?: 1; return timeout ?: 1;
} }
...@@ -280,7 +282,7 @@ bool try_wait_for_completion(struct completion *x) ...@@ -280,7 +282,7 @@ bool try_wait_for_completion(struct completion *x)
spin_lock_irqsave(&x->wait.lock, flags); spin_lock_irqsave(&x->wait.lock, flags);
if (!x->done) if (!x->done)
ret = 0; ret = 0;
else else if (x->done != UINT_MAX)
x->done--; x->done--;
spin_unlock_irqrestore(&x->wait.lock, flags); spin_unlock_irqrestore(&x->wait.lock, flags);
return ret; return ret;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment