[PATCH] sched: Use WARN_ON

The BUG_ON macro simplifies the if condition followed by BUG, but it
will lead to the kernel crashing. Therefore, we can try using WARN_ON
instead of if condition followed by BUG.

Signed-off-by: Jason Wang <redacted>
---
 arch/powerpc/platforms/cell/spufs/sched.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/platforms/cell/spufs/sched.c b/arch/powerpc/platforms/cell/spufs/sched.c
index 369206489895..0f218d9e5733 100644
--- a/arch/powerpc/platforms/cell/spufs/sched.c
+++ b/arch/powerpc/platforms/cell/spufs/sched.c

@@ -904,8 +904,8 @@ static noinline void spusched_tick(struct spu_context *ctx)
 	struct spu_context *new = NULL;
 	struct spu *spu = NULL;
 
-	if (spu_acquire(ctx))
-		BUG();	/* a kernel thread never has signals pending */
+	/* a kernel thread never has signals pending */
+	WARN_ON(spu_acquire(ctx));
 
 	if (ctx->state != SPU_STATE_RUNNABLE)
 		goto out;

-- 
2.32.0

Le 01/07/2021 à 14:50, Jason Wang a écrit :

The BUG_ON macro simplifies the if condition followed by BUG, but it
will lead to the kernel crashing. Therefore, we can try using WARN_ON
instead of if condition followed by BUG.

But are you sure it is ok to continue if spu_acquire(ctx) returned false ?
Shouldn't there be at least for fallback handling ?

Something like:

	if (WARN_ON(spu_acquire(ctx)))
		return;


Christophe

On Thu, Jul 1, 2021 at 2:57 PM Christophe Leroy
[off-list ref] wrote:

Le 01/07/2021 à 14:50, Jason Wang a écrit :

quoted

The BUG_ON macro simplifies the if condition followed by BUG, but it
will lead to the kernel crashing. Therefore, we can try using WARN_ON
instead of if condition followed by BUG.

But are you sure it is ok to continue if spu_acquire(ctx) returned false ?
Shouldn't there be at least for fallback handling ?

Something like:

        if (WARN_ON(spu_acquire(ctx)))
                return;

I think you get a crash in either case:

- with the existing BUG_ON() there is an immediate backtrace and it stops there
- with WARN_ON() and continuing, you operate on a context that is not
  valid
- with the 'return', you get an endless loop, as it keeps calling
spusched_tick()
  without sleeping.

Out of those options, the existing BUG_ON() seems best.

       Arnd