linux-kernel - Re: [PATCH printk-rework 03/12] printk: consolidate kmsg_dump_get_buffer/syslog_print

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, 2 Feb 2021 13:31:04 +0100
From:   Petr Mladek <pmladek@...e.com>
To:     John Ogness <john.ogness@...utronix.de>
Cc:     Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH printk-rework 03/12] printk: consolidate
 kmsg_dump_get_buffer/syslog_print_all code

On Mon 2021-02-01 10:55:22, John Ogness wrote:
> On 2021-01-29, Petr Mladek <pmladek@...e.com> wrote:
> >> The logic for finding records to fit into a buffer is the same for
> >> kmsg_dump_get_buffer() and syslog_print_all(). Introduce a helper
> >> function find_first_fitting_seq() to handle this logic.
> >> 
> >> Signed-off-by: John Ogness <john.ogness@...utronix.de>
> >> ---
> >>  kernel/printk/printk.c | 71 ++++++++++++++++++++++++------------------
> >>  1 file changed, 41 insertions(+), 30 deletions(-)
> >> 
> >> diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
> >> index 1929aa372e7f..ec2174882b8e 100644
> >> --- a/kernel/printk/printk.c
> >> +++ b/kernel/printk/printk.c
> >> @@ -1421,6 +1421,41 @@ static size_t get_record_print_text_size(struct printk_info *info,
> >>  	return ((prefix_len * line_count) + info->text_len + 1);
> >>  }
> >>  
> >> +/*
> >> + * Beginning with @start_seq, find the first record where it and all following
> >> + * records up to (but not including) @max_seq fit into @size.
> >
> > Please, mention that all existing messages are checked when @max_seq
> > == -1 is used.
> 
> -1 is not special for that purpose, but I will add a comment that if
>  there is no required upper bound, the caller can use -1.

Thanks. I think that it is good to mention that the function is able
to handle this situation. For example, it will not wait until
a message with @max_seq is available ;-)

> >> + */
> >> +static u64 find_first_fitting_seq(u64 start_seq, u64 max_seq, size_t size,
> >> +				  struct printk_info *info, bool syslog, bool time)
> >> +{

> >> +	/*
> >> +	 * Move first record forward until length fits into the buffer. Ignore
> >> +	 * newest messages that were not counted in the above cycle. Messages
> >> +	 * might appear and get lost in the meantime. This is a best effort
> >> +	 * that prevents an infinite loop that could occur with a retry.
> >> +	 */
> >> +	if (seq < max_seq)
> >> +		max_seq = seq;
> >
> > This made my head twist around several times ;-)
> >
> > It should never be true in kmsg_dump_get_buffer().
> 
> Correct. It is there because of syslog_print_all().
> 
> > And there was nothing like this in the original syslog_print_all().
> 
> With logbuf_lock, it is not possible that new messages arrive in between
> these two loops. But without logbuf_lock, it _is_ possible and this
> needs to be handled.

I see.

> I can expand the commit message to mention this necessary change.

Yes, please. I am fine with the code now.



> >> @@ -3436,26 +3461,12 @@ bool kmsg_dump_get_buffer(struct kmsg_dumper *dumper, bool syslog,
> >>  
> >>  	/*
> >>  	 * Find first record that fits, including all following records,
> >> -	 * into the user-provided buffer for this dump.
> >> +	 * into the user-provided buffer for this dump. Pass in size-1
> >> +	 * because this function (by way of record_print_text()) will
> >> +	 * not write more than size-1 bytes of text into @buf.
> >
> > We should do the same also in syslog_print_all(). It must have the
> > same problem. The last message might get lost when there is not
> > a space for the trailing '\0' that was not counted before.
> 
> No, it does not have the same problem. I also made the mistake [0] of
> thinking that.
> 
> copy_to_user() is the function filling the buffer, not
> record_print_text(). And it will (and always has) fill the full
> buffer. Only kmsg_dump_get_buffer() has the bizarre semantics of not
> using the full buffer.

Right. I got it the wrong way.

> > And it might be better to actually change the condition in
> > find_first_fitting_seq(). I mean to replace:
> >
> > 	if (len <= size || info.seq >= max_seq)
> > with
> > 	if (len < size || info.seq >= max_seq)
> 
> I would prefer not twisting syslog_print_all() to act like
> kmsg_dump_get_buffer().

I agree. It is not a common problem after all.

Best Regards,
Petr