linux-kernel - Re: [PATCH] printk: add per console loglevel

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20150104201945.GA15792@fifo99.com>
Date:	Sun, 4 Jan 2015 20:19:45 +0000
From:	dwalker@...o99.com
To:	Bruno Prémont <bonbons@...ux-vserver.org>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] printk: add per console loglevel

On Tue, Dec 23, 2014 at 10:13:24PM +0100, Bruno Prémont wrote:
> On Tue, 23 December 2014 dwalker@...o99.com wrote:
> > On Sun, Dec 21, 2014 at 07:47:53PM +0100, Bruno Prémont wrote:
> > > On Sat, 20 December 2014 dwalker@...o99.com wrote:
> > > > This adds to to the console= command line options allowing the
> > > > addition of a per console log level setting.
> > > > 
> > > > examples,
> > > > 
> > > > 	console=ttyS0,ll4
> > > > 	console=tty0,ll6
> > > > 
> > > > This can be used on systems which have multiple serial consoles, but
> > > > it's desired for logging to be light on one and heavy on another.
> > > 
> > > Looks useful to me.
> > > What would be the best way to make these per-console loglevels
> > > configurable at runtime? `dmesg -n $LEVEL` only affects the global
> > > limit.
> > > 
> > > One drawback to letting global loglevel have precedence is that
> > > all consoles that should not get detailed log messages need to have
> > > their loglevel explicitly lowered and it's not possible to have
> > > one console forced to show more than the global loglevel.
> > 
> > Right, reason for the patch.
> >  
> > > 
> > > An approach I would prefer is to have all consoles follow global
> > > loglevel except when something different had been explicitly requested
> > > for them.
> > 
> > This patch is mostly like this. It breaks down when you set the "llX"
> > parameter to something above KERN_NOTICE, and the global one is also
> > set above that.
> 
> In your patch global filter applies first, then what passes global filter
> gets filtered with the per-console filter.
> 
> So I can't have `dmesg -n 3` and "console=ttyS0,ll8 console=tty0"
> to get my debugging output over serial console.

Yeah .. This would improve things. I changed it after I sent the first patch, but it
looks like one of your patches changes it also.

I was a little concerned about altering the fast path in call_console_drivers().. With
the changes I made each call into call_console_drivers is forced to go thru the whole list
of consoles, instead of short circuiting prior to the list.
 
I was thinking the max local and global log level could be saved and used for short circuiting
instead of using the global log level.

> > > This way a single console can be added later on (e.g. netconsole)
> > > and set to pass through debug messages without affecting anyone else.
> 
> Netconsole console can be added at any time when system is running
> either through loading of its module or using dynamic configuration
> via configfs.
> 
> See also my patches.

Yeah, I saw them .. Not in too much detail, but it seems worth while for it to
work with netconsole.

Daniel
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/