[PATCH] printk/console: Allow to disable console output by using console="" or console=null

[PATCH] printk/console: Allow to disable console output by using console="" or console=null

[Petr Mladek]

The commit 48021f98130880dd74 ("printk: handle blank console arguments
passed in.") prevented crash caused by empty console= parameter value.

Unfortunately, this value is widely used on Chromebooks to disable
the console output. The above commit caused performance regression
because the messages were pushed on slow console even though nobody
was watching it.

Use ttynull driver explicitly for console="" and console=null
parameters. It has been created for exactly this purpose.

It causes that preferred_console is set. As a result, ttySX and ttyX
are not used as a fallback. And only ttynull console gets registered by
default.

It still allows to register other consoles either by additional console=
parameters or SPCR. It prevents regression because it worked this way even
before. Also it is a sane semantic. Preventing output on all consoles
should be done another way, for example, by introducing mute_console
parameter.

Link: https://lore.kernel.org/r/20201006025935.GA597@jagdpanzerIV.localdomain
Suggested-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Acked-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Signed-off-by: Petr Mladek <pmladek@suse.com>
Link: https://lore.kernel.org/r/20201111135450.11214-3-pmladek@suse.com
---

This is backport of the commit 3cffa06aeef7ece30f6b5ac0e
("printk/console: Allow to disable console output by using
console="" or console=null") for stable release:

    + 4.4, 4.9, 4.14, 4.19, 5.4

Please, use the original upstream commit for stable release:

   + 5.10

It should fix the problem reported at
https://www.spinics.net/lists/stable/msg509616.html

kernel/printk/printk.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index b55dfb3e801f..6d3e1f4961fb 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -2032,8 +2032,15 @@ static int __init console_setup(char *str)
 	char *s, *options, *brl_options = NULL;
 	int idx;
 
-	if (str[0] == 0)
+	/*
+	 * console="" or console=null have been suggested as a way to
+	 * disable console output. Use ttynull that has been created
+	 * for exacly this purpose.
+	 */
+	if (str[0] == 0 || strcmp(str, "null") == 0) {
+		__add_preferred_console("ttynull", 0, NULL, NULL);
 		return 1;
+	}
 
 	if (_braille_console_setup(&str, &brl_options))
 		return 1;
-- 
2.26.2

Re: [PATCH] printk/console: Allow to disable console output by using console="" or console=null

[Greg KH]

On Tue, Nov 09, 2021 at 04:53:26PM +0100, Petr Mladek wrote:
> The commit 48021f98130880dd74 ("printk: handle blank console arguments
> passed in.") prevented crash caused by empty console= parameter value.
> 
> Unfortunately, this value is widely used on Chromebooks to disable
> the console output. The above commit caused performance regression
> because the messages were pushed on slow console even though nobody
> was watching it.
> 
> Use ttynull driver explicitly for console="" and console=null
> parameters. It has been created for exactly this purpose.
> 
> It causes that preferred_console is set. As a result, ttySX and ttyX
> are not used as a fallback. And only ttynull console gets registered by
> default.
> 
> It still allows to register other consoles either by additional console=
> parameters or SPCR. It prevents regression because it worked this way even
> before. Also it is a sane semantic. Preventing output on all consoles
> should be done another way, for example, by introducing mute_console
> parameter.
> 
> Link: https://lore.kernel.org/r/20201006025935.GA597@jagdpanzerIV.localdomain
> Suggested-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
> Reviewed-by: Guenter Roeck <linux@roeck-us.net>
> Tested-by: Guenter Roeck <linux@roeck-us.net>
> Acked-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
> Signed-off-by: Petr Mladek <pmladek@suse.com>
> Link: https://lore.kernel.org/r/20201111135450.11214-3-pmladek@suse.com
> ---
> 
> This is backport of the commit 3cffa06aeef7ece30f6b5ac0e
> ("printk/console: Allow to disable console output by using
> console="" or console=null") for stable release:
> 
>     + 4.4, 4.9, 4.14, 4.19, 5.4
> 
> Please, use the original upstream commit for stable release:
> 
>    + 5.10
> 
> It should fix the problem reported at
> https://www.spinics.net/lists/stable/msg509616.html

Thanks, now all queued up!

greg k-h

Re: [PATCH] printk/console: Allow to disable console output by using console="" or console=null

[Guenter Roeck]

On 11/9/21 7:53 AM, Petr Mladek wrote:
> The commit 48021f98130880dd74 ("printk: handle blank console arguments
> passed in.") prevented crash caused by empty console= parameter value.
> 
> Unfortunately, this value is widely used on Chromebooks to disable
> the console output. The above commit caused performance regression
> because the messages were pushed on slow console even though nobody
> was watching it.
> 

We actually had to revert this patch on Chromebooks, so we'll have to revert
it again from stable releases after it gets there. The problem is two-fold:

First, it is used in Chromebooks to disable the default console in production
images; that default console may be set in a devicetree file, and this patch
doesn't really disable it. In other words, Chromebooks use "console=" to
implement "mute_console" as suggested below, and this patch does not address
that use case.

Second, the patch causes some unexplained problems with dm-verity, which
inexplicably fails on some Chromebooks when the patch is in place.
We never tracked down the root cause because the patch doesn't work
for us anyway.

Guenter

> Use ttynull driver explicitly for console="" and console=null
> parameters. It has been created for exactly this purpose.
> 
> It causes that preferred_console is set. As a result, ttySX and ttyX
> are not used as a fallback. And only ttynull console gets registered by
> default.
> 
> It still allows to register other consoles either by additional console=
> parameters or SPCR. It prevents regression because it worked this way even
> before. Also it is a sane semantic. Preventing output on all consoles
> should be done another way, for example, by introducing mute_console
> parameter.
> 
> Link: https://lore.kernel.org/r/20201006025935.GA597@jagdpanzerIV.localdomain
> Suggested-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
> Reviewed-by: Guenter Roeck <linux@roeck-us.net>
> Tested-by: Guenter Roeck <linux@roeck-us.net>
> Acked-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
> Signed-off-by: Petr Mladek <pmladek@suse.com>
> Link: https://lore.kernel.org/r/20201111135450.11214-3-pmladek@suse.com
> ---
> 
> This is backport of the commit 3cffa06aeef7ece30f6b5ac0e
> ("printk/console: Allow to disable console output by using
> console="" or console=null") for stable release:
> 
>      + 4.4, 4.9, 4.14, 4.19, 5.4
> 
> Please, use the original upstream commit for stable release:
> 
>     + 5.10
> 
> It should fix the problem reported at
> https://www.spinics.net/lists/stable/msg509616.html
> 
> kernel/printk/printk.c | 9 ++++++++-
>   1 file changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
> index b55dfb3e801f..6d3e1f4961fb 100644
> --- a/kernel/printk/printk.c
> +++ b/kernel/printk/printk.c
> @@ -2032,8 +2032,15 @@ static int __init console_setup(char *str)
>   	char *s, *options, *brl_options = NULL;
>   	int idx;
>   
> -	if (str[0] == 0)
> +	/*
> +	 * console="" or console=null have been suggested as a way to
> +	 * disable console output. Use ttynull that has been created
> +	 * for exacly this purpose.
> +	 */
> +	if (str[0] == 0 || strcmp(str, "null") == 0) {
> +		__add_preferred_console("ttynull", 0, NULL, NULL);
>   		return 1;
> +	}
>   
>   	if (_braille_console_setup(&str, &brl_options))
>   		return 1;
> 

Re: [PATCH] printk/console: Allow to disable console output by using console="" or console=null

[Petr Mladek]

On Tue 2021-11-09 10:15:12, Guenter Roeck wrote:
> On 11/9/21 7:53 AM, Petr Mladek wrote:
> > The commit 48021f98130880dd74 ("printk: handle blank console arguments
> > passed in.") prevented crash caused by empty console= parameter value.
> > 
> > Unfortunately, this value is widely used on Chromebooks to disable
> > the console output. The above commit caused performance regression
> > because the messages were pushed on slow console even though nobody
> > was watching it.
> > 
> 
> We actually had to revert this patch on Chromebooks, so we'll have to revert
> it again from stable releases after it gets there.

What patch was or need to get reverted on Chromebooks, please?

  1. commit 48021f98130880dd74 ("printk: handle blank console
     arguments passed in.")

or

  2. commit commit 3cffa06aeef7ece30f6b5ac0e ("printk/console: Allow
     to disable console output by using console="" or console=null")


I know that the 1st patch caused problems on Chromebook. The 2nd one
was supposed to fix the problem.

The 2nd patch is being backported here? Do you still see the problems
with it, please?


> The problem is two-fold:
> 
> First, it is used in Chromebooks to disable the default console in production
> images; that default console may be set in a devicetree file, and this patch
> doesn't really disable it. In other words, Chromebooks use "console=" to
> implement "mute_console" as suggested below, and this patch does not address
> that use case.

I guess that you are talking about the 1st patch.

The 2nd patch should make it working basically the same way as when reverting
the 1st patch. The difference is that it prefers the fake ttynull
console driver instead of none. It should be better because it will
provide a kind of null console for stdin/stdou/stderr of the init
process. But it still should result into a none-driver when ttynull
driver is not available.

Or do you use another extra patch for Chromebooks, please?

> Second, the patch causes some unexplained problems with dm-verity, which
> inexplicably fails on some Chromebooks when the patch is in place.
> We never tracked down the root cause because the patch doesn't work
> for us anyway.

Interesting. I wonder what console was really registered when it complained.

Best Regards,
Petr

Re: [PATCH] printk/console: Allow to disable console output by using console="" or console=null

[Guenter Roeck]

On 11/10/21 4:03 AM, Petr Mladek wrote:
> On Tue 2021-11-09 10:15:12, Guenter Roeck wrote:
>> On 11/9/21 7:53 AM, Petr Mladek wrote:
>>> The commit 48021f98130880dd74 ("printk: handle blank console arguments
>>> passed in.") prevented crash caused by empty console= parameter value.
>>>
>>> Unfortunately, this value is widely used on Chromebooks to disable
>>> the console output. The above commit caused performance regression
>>> because the messages were pushed on slow console even though nobody
>>> was watching it.
>>>
>>
>> We actually had to revert this patch on Chromebooks, so we'll have to revert
>> it again from stable releases after it gets there.
> 
> What patch was or need to get reverted on Chromebooks, please?
> 
>    1. commit 48021f98130880dd74 ("printk: handle blank console
>       arguments passed in.")
> 
> or
> 
>    2. commit commit 3cffa06aeef7ece30f6b5ac0e ("printk/console: Allow
>       to disable console output by using console="" or console=null")
> 
Both.

> 
> I know that the 1st patch caused problems on Chromebook. The 2nd one
> was supposed to fix the problem.
> 
> The 2nd patch is being backported here? Do you still see the problems
> with it, please?
> 
Yes.

> 
>> The problem is two-fold:
>>
>> First, it is used in Chromebooks to disable the default console in production
>> images; that default console may be set in a devicetree file, and this patch
>> doesn't really disable it. In other words, Chromebooks use "console=" to
>> implement "mute_console" as suggested below, and this patch does not address
>> that use case.
> 
> I guess that you are talking about the 1st patch.
> 
> The 2nd patch should make it working basically the same way as when reverting
> the 1st patch. The difference is that it prefers the fake ttynull
> console driver instead of none. It should be better because it will
> provide a kind of null console for stdin/stdou/stderr of the init
> process. But it still should result into a none-driver when ttynull
> driver is not available.
> 

Yes, that was what we initially assumed as well - only, as the patch
states, it doesn't work as needed for Chromebooks if another console
is registered by other means.

> Or do you use another extra patch for Chromebooks, please?
> 

Not that I know of. We just rely on the old behavior.

>> Second, the patch causes some unexplained problems with dm-verity, which
>> inexplicably fails on some Chromebooks when the patch is in place.
>> We never tracked down the root cause because the patch doesn't work
>> for us anyway.
> 
> Interesting. I wonder what console was really registered when it complained.
> 

It was with a Chromebook using devicetree, where the console is primarily
passed via devicetree and not as boot parameter. I don't recall the exact
model(s).

Guenter