linux-kernel - Re: [PATCH] tracing: Do not allow mmap() of persistent ring buffer

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 for Android: free password hash cracker in your pocket

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

Message-Id: <20250219001424.c457366901b9327a79e2b83d@kernel.org>
Date: Wed, 19 Feb 2025 00:14:24 +0900
From: Masami Hiramatsu (Google) <mhiramat@...nel.org>
To: Steven Rostedt <rostedt@...dmis.org>
Cc: LKML <linux-kernel@...r.kernel.org>, Linux Trace Kernel
 <linux-trace-kernel@...r.kernel.org>, Mathieu Desnoyers
 <mathieu.desnoyers@...icios.com>, Vincent Donnefort <vdonnefort@...gle.com>
Subject: Re: [PATCH] tracing: Do not allow mmap() of persistent ring buffer

On Sat, 15 Feb 2025 11:45:40 -0500
Steven Rostedt <rostedt@...dmis.org> wrote:

> On Sat, 15 Feb 2025 11:21:53 -0500
> Steven Rostedt <rostedt@...dmis.org> wrote:
> 
> > On Sun, 16 Feb 2025 00:37:02 +0900
> > Masami Hiramatsu (Google) <mhiramat@...nel.org> wrote:
> > 
> > > My concern is related to the fixes policy. If this is a "fix", we will
> > > backport the new "disables mmap on persistent ring buffer" limitation
> > > to the stable kernel (that was not documented previously.)
> > > 
> > > However, from the user point of view, "mmap() ring buffers" is already
> > > supported (although it did not work on stable kernel for now). Thus I think
> > > the "Fix" is expected as "fixing mmap() persistent ring buffer".   
> > 
> > This only disables mmapping of the persistent ring buffer. Other ring
> > buffers can be mapped. We never supported mmapping the persistent ring
> > buffer. Even in stable kernels, if you mmap it, it will crash just like
> > it does now. Thus, this doesn't cause any regressions. It's a fix even
> > for stable kernels.
> > 
> > Or did the virt_to_page() change recently where that wasn't the case?
> > 
> 
> Although the fixes tag is wrong. As the persistent ring buffer didn't
> even exist then. It should be:
> 
> Fixes: 9b7bdf6f6ece6 ("tracing: Have trace_printk not use binary prints if boot buffer")
> 
> As that's what added the BOOT flag and is in the same kernel version
> that added the persistent ring buffer.

OK, so this fix is for the limitation of the buffer which has
TRACE_ARRAY_FL_BOOT. When this flag is introduced, it should also
prohibit the mmap.

Thank you,

> 
> -- Steve


-- 
Masami Hiramatsu (Google) <mhiramat@...nel.org>