Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   John Ogness <john.ogness@linutronix.de>
To:     Petr Mladek <pmladek@suse.com>
Cc:     Sergey Senozhatsky <senozhatsky@chromium.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        linux-kernel@vger.kernel.org,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Subject: Re: [PATCH printk v2 1/7] printk: Move buffer size defines
In-Reply-To: <Y39Q5TjoHUUKXC1S@alley>
References: <20221123231400.614679-1-john.ogness@linutronix.de>
 <20221123231400.614679-2-john.ogness@linutronix.de>
 <Y39Q5TjoHUUKXC1S@alley>
Date:   Thu, 24 Nov 2022 13:44:29 +0106
Message-ID: <87zgcgttmi.fsf@jogness.linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

On 2022-11-24, Petr Mladek <pmladek@suse.com> wrote:
>> Move the buffer size defines to console.h in preparation of adding a
>> buffer structure. The new buffer structure will be embedded within
>> struct console. Therefore console.h was chosen as the new home for
>> these defines.
>
> The buffers are not embedded into struct console in this patchset.
> Are they going to be added directly or via pointer, please?

By "embedded" I mean added directly. The buffers need to be available
immediately and cannot be allocated or assigned dynamically. The console
struct is generally defined by drivers with:

static struct console my_console = {
   ...
};

I could think of no way to statically define the buffers but keep their
sizes hidden.

> IMHO, it is always better to hide these implementation details
> in an internal header or source file. It will be possible
> if struct console contained on a pointer to the buffers.

The problem is not pointers, it is static definition (without knowing
the size of the thing that is statically defined). The new thread/atomic
consoles run in parallel, so they cannot share the single static buffer
like we do now.

>> --- a/include/linux/console.h
>> +++ b/include/linux/console.h
>> @@ -122,6 +122,20 @@ static inline int con_debug_leave(void)
>>  #define CM_ERASE    (2)
>>  #define CM_MOVE     (3)
>>  
>> +#ifdef CONFIG_PRINTK
>> +
>> +/* The maximum size of a formatted record (i.e. with prefix added per line) */
>> +#define CONSOLE_LOG_MAX		1024
>> +
>> +#else
>> +
>> +#define CONSOLE_LOG_MAX		0
>> +
>> +#endif
>> +
>> +/* The maximum size of a formatted extended record */
>> +#define CONSOLE_EXT_LOG_MAX	8192
>
> It looks strange that we need the buffer for extended messages
> and not the normal one when !CONFIG_PRINTK. I can't find any reason
> for this. It looks like a historic inconsistency.

Yes, it looked that way to me as well. I will move it under
CONFIG_PRINTK for v3.

John