linux-kernel - Re: devkmsg: was [PATCH next v2 3/3] printk: remove logbuf_lock, add syslog

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

Message-ID: <87y2iamzdp.fsf@jogness.linutronix.de>
Date:   Sun, 06 Dec 2020 21:57:46 +0106
From:   John Ogness <john.ogness@...utronix.de>
To:     Petr Mladek <pmladek@...e.com>
Cc:     Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        linux-kernel@...r.kernel.org
Subject: Re: devkmsg: was [PATCH next v2 3/3] printk: remove logbuf_lock, add syslog_lock

On 2020-12-04, Petr Mladek <pmladek@...e.com> wrote:
>> Since the ringbuffer is lockless, there is no need for it to be
>> protected by @logbuf_lock. Remove @logbuf_lock.
>
> It might make sense to split also this patch into few more pieces that
> would remove the lock from a particular interface.

OK.

>> diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
>> index e9018c4e1b66..7385101210be 100644
>> --- a/kernel/printk/printk.c
>> +++ b/kernel/printk/printk.c
>> @@ -785,7 +749,6 @@ static loff_t devkmsg_llseek(struct file *file, loff_t offset, int whence)
>>  	if (offset)
>>  		return -ESPIPE;
>>  
>> -	logbuf_lock_irq();
>
> user->seq manipulation is not longer safe from the atomicity point of
> view.
>
> One solution would be to use atomic variable in struct devkmsg_user().
> Another solution would be to synchronize it with user->lock like we do
> in devkmsg_read().
>
> user->lock looks like an overhead. But it actually would make sense to
> prevent seek in the middle of a read.

I would prefer using atomic64_t. Using user->lock could introduce some
wacky regression.

>>  	switch (whence) {
>>  	case SEEK_SET:
>>  		/* the first record */
>> @@ -820,7 +782,6 @@ static __poll_t devkmsg_poll(struct file *file, poll_table *wait)
>>  
>>  	poll_wait(file, &log_wait, wait);
>>  
>> -	logbuf_lock_irq();
>>  	if (prb_read_valid(prb, user->seq, NULL)) {
>
> Same here. The atomicity of user->seq read/write is not guaranteed.

Right.

John Ogness