linux-kernel - Re: [PATCH v5 3/7] printk: initialize syslog_prev and console

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]

Date:	Tue, 22 Jul 2014 14:01:35 +0200
From:	Petr Mládek <pmladek@...e.cz>
To:	Alex Elder <elder@...aro.org>
Cc:	akpm@...ux-foundation.org, kay@...y.org, bp@...e.de,
	john.stultz@...aro.org, jack@...e.cz, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v5 3/7] printk: initialize syslog_prev and console_prev

On Tue 2014-07-22 06:48:22, Alex Elder wrote:
> On 07/21/2014 08:02 AM, Alex Elder wrote:
> > Two global variables, "syslog_prev" and "console_prev", maintain a
> > copy of the flags value used in the log record most recently
> > formatted for syslog or the console, respectively.
> > 
> > Initially there is no previous formatted log record, and these
> > variables simply have an initial value 0.  And occasionally log
> > records can get consumed at a rate such that syslog or the console
> > can't keep up, in which case those variables (along with their
> > corresponding position variables) must be reset.  Here too, they're
> > reset to 0.
> > 
> > This patch changes it so the initial value used is LOG_NEWLINE.
> > That is, if we know nothing about the prevously-formatted log
> > record, we can assume it was complete, and ended with a newline.
> > One exception is that occasionally we reset our syslog or console
> > (etc.) position variables.  In that case the previously-formatted
> > record flags value is still valid, so we preserve that information.
> > 
> > This is being done to support the next patch.  Initializing
> > these variables this way makes LOG_NEWLINE and LOG_CONT be
> > mutually exclusive, and the next patch uses that fact to simplify
> > some logic.
> > 
> > Signed-off-by: Alex Elder <elder@...aro.org>
> > Reviewed-by: Petr Mládek <pmladek@...e.cz>
> 
> I have one change I'd like to suggest on this one.
> 
> Petr, could you offer your opinion?
> 
> 
> > ---
> >  kernel/printk/printk.c | 44 ++++++++++++++++++++++++++++++--------------
> >  1 file changed, 30 insertions(+), 14 deletions(-)
> > 
> > diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
> > index 9f11eab..2f43116 100644
> > --- a/kernel/printk/printk.c
> > +++ b/kernel/printk/printk.c
> 
> 
> . . .
> 
> > @@ -2156,10 +2168,14 @@ again:
> >  				      "%s** %u printk messages dropped **\n",
> >  				      (console_prev & LOG_CONT) ? "\n" : "",
> >  				      (unsigned)(log_first_seq - console_seq));
> > -			/* messages are gone, move to first one */
> > +			/*
> > +			 * Messages are gone, move to first one.
> > +			 * Don't discard what we know about the
> > +			 * previously-formatted record.
> > +			 */
> >  			console_seq = log_first_seq;
> >  			console_idx = log_first_idx;
> > -			console_prev = 0;
> > +			console_prev &= LOG_NEWLINE|LOG_CONT;
> 
> In this one spot, I think console_prev should simply be
> initialized with LOG_NEWLINE.
> 
> The reason is that the "printk messages dropped" message will
> be inserted into the formatted output, and is hence the last
> formatted line.  And that message is (now) terminated with
> a newline.

Sure. Great catch! I actually mentioned it when commenting v4, see
https://lkml.org/lkml/2014/7/21/153 . But I forgot it today :-/


Best Regards,
Petr
--
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/