linux-kernel - Re: [PATCH v2 1/8] lib/printbuf: New data structure for heap-allocated strings

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: <20220422215146.i663tn6zzn6blzo3@moria.home.lan>
Date:   Fri, 22 Apr 2022 17:51:46 -0400
From:   Kent Overstreet <kent.overstreet@...il.com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     Christoph Hellwig <hch@....de>, linux-kernel@...r.kernel.org,
        linux-mm@...ck.org, linux-fsdevel@...r.kernel.org,
        hannes@...xchg.org, akpm@...ux-foundation.org,
        linux-clk@...r.kernel.org, linux-tegra@...r.kernel.org,
        linux-input@...r.kernel.org, roman.gushchin@...ux.dev
Subject: Re: [PATCH v2 1/8] lib/printbuf: New data structure for
 heap-allocated strings

On Fri, Apr 22, 2022 at 04:47:44PM -0400, Steven Rostedt wrote:
> Which is something you could do on top of seq_buf. Point being, you do not
> need to re-implement printbuf, and I have not looked at the code, but
> instead, implement printbuf on top of seq_buf, and extend seq_buf where
> needed. Like trace_seq does, and the patches I have for seq_file would do.
> It would leave the string processing and buffer space management to
> seq_buf, as there's ways to see "oh, we need more space, let's allocate
> more" and then increase the heap.

That sounds like it could work. 

> I would be more willing to accept a printbuf, if it was built on top of
> seq_buf. That is, you don't need to change all your user cases, you just
> need to make printbuf an extension of seq_buf by using it underneath, like
> trace_seq does. Then it would not be re-inventing the wheel, but just
> building on top of it.

Hmm... At first glance, redoing printbuf on top of seq_buf looks like it would
save a pretty trivial amount of code - and my engineering taste these days leans
more toward less layering if it's only slightly more code; I think I might like
printbuf and seq_buf to stay separate things (and both of them are pretty
small).

But it's definitely not an unreasonable idea - I can try it out and see how it
turns out. Would you have any objections to making some changes to seq_buf?

 - You've got size and len as size_t, I've got them as unsigned. Given that we
   need to be checking for overflow anyways for correctens, I like having them
   as u32s.
 - seq_buf->readpos - it looks like this is only used by seq_buf_to_user(), does
   it need to be in seq_buf?
 - in printbufs, I make sure the buffer is always nul-terminated - seems
   simplest, given that we need to make sure there's always room for the
   terminating nul anyways.

A downside of having printbuf on top of seq_buf is that now we've got two apis
that functions can output to - vs. if we modified printbuf by adding a flag for
"this is an external buffer, don't reallocate it". That approach would be less
code overall, for sure.

Could I get you to look over printbuf and share your thoughts on the different
approaches?