linux-kernel - Re: [PATCH] softirq: Be more verbose on t->state BUG()

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <87k0q7186h.fsf@nanos.tec.linutronix.de>
Date:   Tue, 16 Mar 2021 15:31:50 +0100
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Eugeniu Rosca <erosca@...adit-jv.com>,
        linux-kernel@...r.kernel.org,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...nel.org>, Will Deacon <will@...nel.org>,
        Frederic Weisbecker <frederic@...nel.org>,
        Jiafei Pan <Jiafei.Pan@....com>,
        Romain Perier <romain.perier@...il.com>
Cc:     Andrew Gabbasov <andrew_gabbasov@...tor.com>,
        Dirk Behme <dirk.behme@...bosch.com>,
        Eugeniu Rosca <roscaeugeniu@...il.com>,
        Eugeniu Rosca <erosca@...adit-jv.com>
Subject: Re: [PATCH] softirq: Be more verbose on t->state BUG()

On Mon, Mar 15 2021 at 16:44, Eugeniu Rosca wrote:
> From: Dirk Behme <dirk.behme@...bosch.com>
>
> In case this BUG() is hit, it helps debugging a lot to get an idea
> what tasklet is the root cause. So, be slightly more verbose here.
>
> Signed-off-by: Dirk Behme <dirk.behme@...bosch.com>
> Signed-off-by: Eugeniu Rosca <erosca@...adit-jv.com>
> ---
>  kernel/softirq.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/softirq.c b/kernel/softirq.c
> index 9908ec4a9bfe..a6b602ad48d6 100644
> --- a/kernel/softirq.c
> +++ b/kernel/softirq.c
> @@ -550,9 +550,13 @@ static void tasklet_action_common(struct softirq_action *a,
>  
>  		if (tasklet_trylock(t)) {
>  			if (!atomic_read(&t->count)) {
> -				if (!test_and_clear_bit(TASKLET_STATE_SCHED,
> -							&t->state))
> +				if (!test_and_clear_bit(TASKLET_STATE_SCHED, &t->state)) {
> +					if (t->use_callback)
> +						pr_emerg("tasklet failed, cb: %pS\n", t->callback);
> +					else
> +						pr_emerg("tasklet failed, func: %pS\n", t->func);
>  					BUG();
> +				}
>  				if (t->use_callback)
>  					t->callback(t);
>  				else

This belongs into unreadable land and actually the BUG() should just be
replaced by a WARN_ONCE(). Something like the below. Hmm?

Thanks,

        tglx
---
 
+static bool tasklet_should_run(struct tasklet_struct *t)
+{
+	if (test_and_clear_bit(TASKLET_STATE_SCHED, &t->state))
+		return true;
+
+	WARN_ONCE(1, "tasklet SCHED state not set: %s %pS\n",
+		  t->use_callback ? "callback" : "func",
+		  t->use_callback ? (void*)t->callback : (void*)t->func);
+
+	return false;
+}
+
 static void tasklet_action_common(struct softirq_action *a,
 				  struct tasklet_head *tl_head,
 				  unsigned int softirq_nr)
@@ -550,13 +562,12 @@ static void tasklet_action_common(struct
 
 		if (tasklet_trylock(t)) {
 			if (!atomic_read(&t->count)) {
-				if (!test_and_clear_bit(TASKLET_STATE_SCHED,
-							&t->state))
-					BUG();
-				if (t->use_callback)
-					t->callback(t);
-				else
-					t->func(t->data);
+				if (tasklet_should_run(t)) {
+					if (t->use_callback)
+						t->callback(t);
+					else
+						t->func(t->data);
+				}
 				tasklet_unlock(t);
 				continue;
 			}