Re: [PATCH 4/4] printk: Drop console_sem during panic

[John Ogness]

On 2022-01-24, John Ogness <john.ogness@xxxxxxxxxxxxx> wrote:
> Something like:
>
> @@ -2731,7 +2731,8 @@ void console_unlock(void)
>  	 * there's a new owner and the console_unlock() from them will do the
>  	 * flush, no worries.
>  	 */
> -	retry = prb_read_valid(prb, next_seq, NULL);
> +	retry = (pcpu != raw_smp_processor_id()) &&
> +		prb_read_valid(prb, next_seq, NULL);
>  	if (retry && console_trylock())
>  		goto again;
>  }

Sorry, that hunk is obviously garbage. I meant something like:

@@ -2731,7 +2731,10 @@ void console_unlock(void)
 	 * there's a new owner and the console_unlock() from them will do the
 	 * flush, no worries.
 	 */
-	retry = prb_read_valid(prb, next_seq, NULL);
+	if (panic_in_progress())
+		retry = (pcpu == raw_smp_processor_id()) && prb_read_valid(prb, next_seq, NULL);
+	else
+		retry = prb_read_valid(prb, next_seq, NULL);
 	if (retry && console_trylock())
 		goto again;
 }

I'm sure there is a cleaner way to code that.

John