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 <senozhatsky@chromium.org>,
	Steven Rostedt <rostedt@goodmis.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	linux-kernel@vger.kernel.org
Subject: Re: preffer_ofload param: was: Re: [PATCH printk v3 11/19] printk: nbcon: Rely on kthreads for normal operation
Date: Fri, 2 Aug 2024 12:19:14 +0200	[thread overview]
Message-ID: <ZqyyovKfUfI1t6Nz@pathway.suse.cz> (raw)
In-Reply-To: <87bk2b8jsi.fsf@jogness.linutronix.de>

On Fri 2024-08-02 09:35:17, John Ogness wrote:
> On 2024-08-01, Petr Mladek <pmladek@suse.com> wrote:
> > I believe that the parameter "prefer_offload" adds more harm than good
> > because:
> >
> >    + It is a non-sense for nbcon consoles. They always prefer offload
> >      except for emergency and panic. But emergency and panic is
> >      handled transparently by nbcon_get_default_prio().
> >
> >    + It is confusing even for legacy consoles after introducing the
> >      kthread. There will be three types of offload:
> >
> > 	+ do console_lock()/unlock() in IRQ work
> > 	+ wake kthread
> > 	+ wake kthread in IRQ work
> 
> I think the confusion comes from my intention of the function. I wanted
> a caller to use it as:
> 
> "Tell me how to flush."

Yes, I understand it the same way.

> This requires input from the caller to know some information about what
> the caller's intentions are.

Does it really need it?
Where?

Please, show me two examples where the function is not able to make
the right decision by global variables?

I know about __wake_up_klogd(). And it is only because
printk_deferred() requires the offload via @level parameter
instead of the per-CPU @printk_context.

Is there any other?

Let me repeat:

    + The parameter makes a difference only when it is "true".
    + The parameter affects only legacy loop.

Let me check where the parameter is true in the current code:

 + nbcon_atomic_flush_pending_con[1548]
    + No effect! The function ignores legacy consoles.

 + nbcon_device_release[1848]
    + No effect! The function ignores legacy consoles.

 + resume_console[2706]
    + Wrong! The function should flush the legacy
	consoles directly by default. Otherwise, we would
	change the legacy behavior.

 + console_flush_all[3102]        printk_get_console_flush_type(&ft, true);
    + No effect! The function ignores legacy consoles.
    + Confusing! The function flushes legacy consoles directly!

  + console_start[3453]
    + Wrong! The function should flush the legacy
      consoles directly by default. Otherwise, we would
	change the legacy behavior.

  + legacy_kthread_should_wakeup[3484]
    + Not needed! The function is called only when
	force_legacy_kthread() == true.

  + __wake_up_klogd[4466]
    + This is the only location where it makes a difference.
    + We could simply check both legacy_direct && legacy_offload
      here.
    + Yes, it is a hack. But it is needed only because of
      printk_deferred() which is already hacky by using
      LOGLEVEL_SCHED.

  + console_try_replay_all[4851]
    + Wrong! It is called under console_lock() => the function will
	flush legacy consoles directly anyway.


> If I change the function so that a caller uses it as:
> 
> "Tell me what flush mechanisms are available to me."

No, this will make the code complicated again.

> Then the function does not need to know the caller's intentions. It only
> needs to know the caller's state, and that information is readily
> available via global/per-cpu variables.
>
> I will drop the @prefer_offload argument, simplifying the function to
> only provide a list of available flush options. The caller will then
> decide itself which option it wants to use. I believe this aligns with
> your intentions as well.

No. Please, keep the original meaning.

Best Regards,
Petr

  reply	other threads:[~2024-08-02 10:19 UTC|newest]

Thread overview: 57+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-07-22 17:19 [PATCH printk v3 00/19] add threaded printing + the rest John Ogness
2024-07-22 17:19 ` [PATCH printk v3 01/19] printk: nbcon: Clarify nbcon_get_default_prio() context John Ogness
2024-07-26  8:57   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 02/19] printk: nbcon: Consolidate alloc() and init() John Ogness
2024-07-26 11:58   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 03/19] printk: nbcon: Add function for printers to reacquire ownership John Ogness
2024-07-26 12:25   ` Petr Mladek
2024-07-29  8:36     ` John Ogness
2024-07-30  9:24       ` Petr Mladek
2024-08-27  1:32         ` John Ogness
2024-07-22 17:19 ` [PATCH printk v3 04/19] printk: nbcon: Clarify rules of the owner/waiter matching John Ogness
2024-07-26 12:55   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 05/19] printk: Fail pr_flush() if before SYSTEM_SCHEDULING John Ogness
2024-07-26 13:14   ` Petr Mladek
2024-07-26 14:45     ` John Ogness
2024-07-30  9:50       ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 06/19] printk: Flush console on unregister_console() John Ogness
2024-07-26 13:23   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 07/19] printk: Add helpers for flush type logic John Ogness
2024-07-23  2:01   ` kernel test robot
2024-07-23  8:39     ` John Ogness
2024-07-23  3:29   ` kernel test robot
2024-07-26 15:51   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 08/19] printk: nbcon: Add context to usable() and emit() John Ogness
2024-07-30 12:30   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 09/19] printk: nbcon: Introduce printer kthreads John Ogness
2024-07-30 14:44   ` John Ogness
2024-07-31  9:59     ` Petr Mladek
2024-07-30 15:16   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 10/19] printk: nbcon: Use thread callback if in task context for legacy John Ogness
2024-07-30 15:35   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 11/19] printk: nbcon: Rely on kthreads for normal operation John Ogness
2024-07-23  3:18   ` kernel test robot
2024-07-23  8:51     ` John Ogness
2024-07-31 13:46   ` preffer_ofload param: was: " Petr Mladek
2024-08-01 14:22     ` John Ogness
2024-08-01 15:40       ` Petr Mladek
2024-08-02  7:29         ` John Ogness
2024-08-02 10:19           ` Petr Mladek [this message]
2024-07-31 14:06   ` Petr Mladek
2024-07-31 15:25     ` John Ogness
2024-08-01  9:36       ` Petr Mladek
2024-08-01  9:52         ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 12/19] printk: Provide helper for message prepending John Ogness
2024-07-22 17:19 ` [PATCH printk v3 13/19] printk: nbcon: Show replay message on takeover John Ogness
2024-07-31 14:59   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 14/19] proc: consoles: Add notation to c_start/c_stop John Ogness
2024-07-22 17:19 ` [PATCH printk v3 15/19] proc: Add nbcon support for /proc/consoles John Ogness
2024-07-31 15:07   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 16/19] tty: sysfs: Add nbcon support for 'active' John Ogness
2024-07-31 15:09   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 17/19] printk: Implement legacy printer kthread for PREEMPT_RT John Ogness
2024-08-02 11:45   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 18/19] printk: nbcon: Assign nice -20 for printing threads John Ogness
2024-08-02 11:47   ` Petr Mladek
2024-07-22 17:19 ` [PATCH printk v3 19/19] printk: Avoid false positive lockdep report for legacy printing John Ogness
2024-08-02 12:34   ` Petr Mladek

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=ZqyyovKfUfI1t6Nz@pathway.suse.cz \
    --to=pmladek@suse.com \
    --cc=john.ogness@linutronix.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rostedt@goodmis.org \
    --cc=senozhatsky@chromium.org \
    --cc=tglx@linutronix.de \
    /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®