Re: + signal-move-print_dropped_signal.patch added to -mm tree

[Oleg Nesterov]

> From: Tom Rix <trix@xxxxxxxxxx>
> Subject: kernel/signal.c: move print_dropped_signal
>
> If the allocation of 'q' fails, the signal will be dropped.

Well, not necessarily... See the comment above TRACE_SIGNAL_LOSE_INFO
in __send_signal().

> To ensure
> that this is reported, move print_dropped_signal to be inside the '(q ==
> NULL)' if-check.

OK, but print_dropped_signal() says "reached RLIMIT_SIGPENDING", this
is misleading if kmem_cache_alloc() fails.

> Link: http://lkml.kernel.org/r/20191203180221.7038-1-trix@xxxxxxxxxx
> Signed-off-by: Tom Rix <trix@xxxxxxxxxx>
> Reviewed-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> Cc: "Eric W. Biederman" <ebiederm@xxxxxxxxxxxx>
> Cc: Christian Brauner <christian.brauner@xxxxxxxxxx>
> Cc: Arnd Bergmann <arnd@xxxxxxxx>
> Cc: Roman Gushchin <guro@xxxxxx>
> Cc: Deepa Dinamani <deepa.kernel@xxxxxxxxx>
> Cc: Oleg Nesterov <oleg@xxxxxxxxxx>
> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> ---
> 
>  kernel/signal.c |    3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> --- a/kernel/signal.c~signal-move-print_dropped_signal
> +++ a/kernel/signal.c
> @@ -427,11 +427,10 @@ __sigqueue_alloc(int sig, struct task_st
>  	    atomic_read(&user->sigpending) <=
>  			task_rlimit(t, RLIMIT_SIGPENDING)) {
>  		q = kmem_cache_alloc(sigqueue_cachep, flags);
> -	} else {
> -		print_dropped_signal(sig);
>  	}
>  
>  	if (unlikely(q == NULL)) {
> +		print_dropped_signal(sig);
>  		atomic_dec(&user->sigpending);
>  		free_uid(user);
>  	} else {
> _
> 
> Patches currently in -mm which might be from trix@xxxxxxxxxx are
> 
> signal-move-print_dropped_signal.patch
>