mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Petr Mladek <pmladek@suse.com>
To: John Ogness <john.ogness@linutronix.de>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>,
	Sergey Senozhatsky <sergey.senozhatsky@gmail.com>,
	Steven Rostedt <rostedt@goodmis.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	linux-kernel@vger.kernel.org
Subject: recursion handling: Re: [PATCH next v2 3/3] printk: remove logbuf_lock, add syslog_lock
Date: Fri, 4 Dec 2020 17:10:07 +0100	[thread overview]
Message-ID: <X8pfX/qPBuY360k/@alley> (raw)
In-Reply-To: <20201201205341.3871-4-john.ogness@linutronix.de>

On Tue 2020-12-01 21:59:41, John Ogness wrote:
> Since the ringbuffer is lockless, there is no need for it to be
> protected by @logbuf_lock. Remove @logbuf_lock.
> 
> --- a/kernel/printk/printk.c
> +++ b/kernel/printk/printk.c
> @@ -1847,6 +1811,65 @@ static void call_console_drivers(const char *ext_text, size_t ext_len,
>  	}
>  }

The recursion-related code needs some explanation or we should do it
another way. I spent quite some time on it and I am still not sure
that I understand it. Let me describe how I understand it.


> +#ifdef CONFIG_PRINTK_NMI
> +#define NUM_RECURSION_CTX 2
> +#else
> +#define NUM_RECURSION_CTX 1
> +#endif

OK, the number of context is limited because interrupts are disabled inside
print_enter()/printk_exit(). It is basically the same reason why
we have only two printk_safe buffers (NNI + other contexts).

What is the exact reason to disable interrupts around the entire
vprintk_store(), please? It should get documented.

One reason is the use of per-cpu variables. Alternative solution would
be to store printk_context into task_struct. Well, I am not sure if
"current" task is available during early boot. But it might solve
problems with per-cpu variables that are not working during early boot.

That said, I am not sure if it is worth it.


> +
> +struct printk_recursion {
> +	char	count[NUM_RECURSION_CTX];
> +};
>
> +static DEFINE_PER_CPU(struct printk_recursion, percpu_printk_recursion);
> +static char printk_recursion_count[NUM_RECURSION_CTX];

This is pretty confusing. The array is hidden in a struct when per-cpu
variables are used. And a naked array is used for early boot.

Is the structure really needed? What about?

static DEFINE_PER_CPU(char [PRINTK_CTX_NUM], printk_count);
static char printk_count_early[NUM_RECURSION_CTX];

> +
> +static char *get_printk_count(void)
> +{
> +	struct printk_recursion *rec;
> +	char *count;
> +
> +	if (!printk_percpu_data_ready()) {
> +		count = &printk_recursion_count[0];

I see why you avoided per-cpu variables for early boot. I am just
curious how printk_context variable works these days. It is used by
any printk(), including early code, see vprintk_func().


> +	} else {
> +		rec = this_cpu_ptr(&percpu_printk_recursion);
> +
> +		count = &rec->count[0];
> +	}
> +
> +#ifdef CONFIG_PRINTK_NMI
> +	if (in_nmi())
> +		count++;
> +#endif

This is extremely confusing. It is far from obvious that
the pointer and not the value is incremented.

If we really need this to avoid per-cpu variables during early boot
then a more clear implementation would be:

char *get_printk_counter_by_ctx()
{
	int ctx = 0;

	if (in_nmi)
		ctx = 1;

	if (!printk_percpu_data_ready())
		return &printk_count_early[ctx];

	return this_cpu_ptr(printk_count[ctx]);
}

> +
> +	return count;
> +}
> +
> +static bool printk_enter(unsigned long *flags)
> +{
> +	char *count;
> +
> +	local_irq_save(*flags);
> +	count = get_printk_count();
> +	/* Only 1 level of recursion allowed. */

We should allow at least some level of recursion. Otherwise, we would
not see warnings printed from vsprintf code.

> +	if (*count > 1) {
> +		local_irq_restore(*flags);
> +		return false;
> +	}
> +	(*count)++;
> +
> +	return true;
> +}

This should be unified with printk_context, printk_nmi_enter(),
printk_nmi_exit(). It does not make sense to have two separate
printk context counters.

Or is there any plan to remove printk_safe and printk_context?

BTW: I prefer to use the bitmask approach. It allows to check
the normal context by a single operation (no bit is set).
There is no need to go through all counters.

Note that we need at least one more context for gdb.

Best Regards,
Petr

  parent reply	other threads:[~2020-12-04 16:11 UTC|newest]

Thread overview: 47+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-12-01 20:53 [PATCH next v2 0/3] printk: remove logbuf_lock John Ogness
2020-12-01 20:53 ` [PATCH next v2 1/3] printk: inline log_output(),log_store() in vprintk_store() John Ogness
2020-12-03 15:57   ` Petr Mladek
2020-12-03 16:25     ` John Ogness
2020-12-04  6:13       ` Sergey Senozhatsky
2020-12-04  8:26       ` Petr Mladek
2020-12-01 20:53 ` [PATCH next v2 2/3] printk: change @clear_seq to atomic64_t John Ogness
2020-12-04  9:12   ` Petr Mladek
2020-12-06 20:23     ` John Ogness
2020-12-07  9:34       ` Peter Zijlstra
2020-12-07 10:03         ` John Ogness
2020-12-07 12:56           ` Peter Zijlstra
2020-12-07 12:56           ` Petr Mladek
2020-12-07 16:46           ` David Laight
2020-12-08 20:34     ` Sergey Senozhatsky
2020-12-08 22:30       ` John Ogness
2020-12-09  1:04         ` Sergey Senozhatsky
2020-12-09  8:16         ` Peter Zijlstra
2020-12-09  9:22           ` Sergey Senozhatsky
2020-12-09 10:46             ` Sergey Senozhatsky
2020-12-09 11:00               ` Peter Zijlstra
2020-12-09 11:28                 ` Sergey Senozhatsky
2020-12-09 12:29                   ` Peter Zijlstra
2020-12-09  8:07       ` Peter Zijlstra
2020-12-01 20:53 ` [PATCH next v2 3/3] printk: remove logbuf_lock, add syslog_lock John Ogness
2020-12-04  6:41   ` Sergey Senozhatsky
2020-12-06 20:44     ` John Ogness
2020-12-04 15:52   ` devkmsg: was " Petr Mladek
2020-12-06 20:51     ` John Ogness
2020-12-07  9:56       ` Petr Mladek
2020-12-04 15:57   ` syslog: was: " Petr Mladek
2020-12-06 21:06     ` John Ogness
2020-12-07 10:01       ` Petr Mladek
2020-12-04 16:10   ` Petr Mladek [this message]
2020-12-05  4:25     ` recursion handling: " Sergey Senozhatsky
2020-12-06 22:08       ` John Ogness
2020-12-05  9:41     ` Sergey Senozhatsky
2020-12-06 22:17       ` John Ogness
2020-12-06 21:44     ` John Ogness
2020-12-07 11:17       ` Petr Mladek
2020-12-04 16:15   ` vprintk_store: was: " Petr Mladek
2020-12-06 22:30     ` John Ogness
2020-12-07 12:46       ` Petr Mladek
2020-12-04 16:19   ` consoles: " Petr Mladek
2020-12-05  4:39     ` Sergey Senozhatsky
2020-12-07  9:50       ` Petr Mladek
2020-12-08 20:51         ` Sergey Senozhatsky

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=X8pfX/qPBuY360k/@alley \
    --to=pmladek@suse.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=john.ogness@linutronix.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rostedt@goodmis.org \
    --cc=sergey.senozhatsky.work@gmail.com \
    --cc=sergey.senozhatsky@gmail.com \
    --cc=tglx@linutronix.de \
    --cc=torvalds@linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®