[RFC PATCH 1/1] rseq: Eliminate useless task_work on execve

[RFC PATCH 1/1] rseq: Eliminate useless task_work on execve

[Mathieu Desnoyers]

Eliminate a useless task_work on execve by moving the call to
rseq_set_notify_resume() from sched_mm_cid_after_execve() to the error
path of bprm_execve(),

The call to rseq_set_notify_resume() from sched_mm_cid_after_execve() is
pointless in the success case, because rseq_execve() will clear the rseq
pointer before returning to userspace.

sched_mm_cid_after_execve() is called from both the success and error
paths of bprm_execve(). The call to rseq_set_notify_resume() is needed
on error because the mm_cid may have changed.

Reported-by: Mateusz Guzik <mjguzik@gmail.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Michael Jeanson <mjeanson@efficios.com>
Cc: Mateusz Guzik <mjguzik@gmail.com>
---
 fs/exec.c           | 3 ++-
 kernel/sched/core.c | 1 -
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/exec.c b/fs/exec.c
index 506cd411f4ac..35ee689ec00a 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -1861,10 +1861,10 @@ static int bprm_execve(struct linux_binprm *bprm)
 		goto out;
 
 	sched_mm_cid_after_execve(current);
+	rseq_execve(current);
 	/* execve succeeded */
 	current->fs->in_exec = 0;
 	current->in_execve = 0;
-	rseq_execve(current);
 	user_events_execve(current);
 	acct_update_integrals(current);
 	task_numa_free(current, false);
@@ -1881,6 +1881,7 @@ static int bprm_execve(struct linux_binprm *bprm)
 		force_fatal_sig(SIGSEGV);
 
 	sched_mm_cid_after_execve(current);
+	rseq_set_notify_resume(t);
 	current->fs->in_exec = 0;
 	current->in_execve = 0;
 
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 042351c7afce..e59a4e2a6453 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -10652,7 +10652,6 @@ void sched_mm_cid_after_execve(struct task_struct *t)
 		smp_mb();
 		t->last_mm_cid = t->mm_cid = mm_cid_get(rq, t, mm);
 	}
-	rseq_set_notify_resume(t);
 }
 
 void sched_mm_cid_fork(struct task_struct *t)
-- 
2.39.5

Re: [RFC PATCH 1/1] rseq: Eliminate useless task_work on execve

[Mathieu Desnoyers]

On 2025-03-27 09:20, Mathieu Desnoyers wrote:
> Eliminate a useless task_work on execve by moving the call to
> rseq_set_notify_resume() from sched_mm_cid_after_execve() to the error
> path of bprm_execve(),
> 
> The call to rseq_set_notify_resume() from sched_mm_cid_after_execve() is
> pointless in the success case, because rseq_execve() will clear the rseq
> pointer before returning to userspace.
> 
> sched_mm_cid_after_execve() is called from both the success and error
> paths of bprm_execve(). The call to rseq_set_notify_resume() is needed
> on error because the mm_cid may have changed.
> 
> Reported-by: Mateusz Guzik <mjguzik@gmail.com>
> Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
> Cc: Ingo Molnar <mingo@kernel.org>
> Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
> Cc: "Paul E. McKenney" <paulmck@kernel.org>
> Cc: Boqun Feng <boqun.feng@gmail.com>
> Cc: Michael Jeanson <mjeanson@efficios.com>
> Cc: Mateusz Guzik <mjguzik@gmail.com>
> ---
>   fs/exec.c           | 3 ++-
>   kernel/sched/core.c | 1 -
>   2 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/exec.c b/fs/exec.c
> index 506cd411f4ac..35ee689ec00a 100644
> --- a/fs/exec.c
> +++ b/fs/exec.c
> @@ -1861,10 +1861,10 @@ static int bprm_execve(struct linux_binprm *bprm)
>   		goto out;
>   
>   	sched_mm_cid_after_execve(current);
> +	rseq_execve(current);
>   	/* execve succeeded */
>   	current->fs->in_exec = 0;
>   	current->in_execve = 0;
> -	rseq_execve(current);
>   	user_events_execve(current);
>   	acct_update_integrals(current);
>   	task_numa_free(current, false);
> @@ -1881,6 +1881,7 @@ static int bprm_execve(struct linux_binprm *bprm)
>   		force_fatal_sig(SIGSEGV);
>   
>   	sched_mm_cid_after_execve(current);
> +	rseq_set_notify_resume(t);

t -> current. will send a v2.

>   	current->fs->in_exec = 0;
>   	current->in_execve = 0;
>   
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 042351c7afce..e59a4e2a6453 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -10652,7 +10652,6 @@ void sched_mm_cid_after_execve(struct task_struct *t)
>   		smp_mb();
>   		t->last_mm_cid = t->mm_cid = mm_cid_get(rq, t, mm);
>   	}
> -	rseq_set_notify_resume(t);
>   }
>   
>   void sched_mm_cid_fork(struct task_struct *t)


-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com