linux-kernel - Re: [PATCH v3] can, tty: elmcan CAN/ldisc driver for ELM327 based OBD-II adapters

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]

Date:   Sat, 12 Mar 2022 22:21:42 +0100
From:   Max Staudt <max@...as.org>
To:     Vincent Mailhol <vincent.mailhol@...il.com>,
        Marc Kleine-Budde <mkl@...gutronix.de>
Cc:     Wolfgang Grandegger <wg@...ndegger.com>, linux-can@...r.kernel.org,
        Oliver Neukum <oneukum@...e.com>, linux-kernel@...r.kernel.org,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jiri Slaby <jslaby@...e.com>
Subject: Re: [PATCH v3] can, tty: elmcan CAN/ldisc driver for ELM327 based
 OBD-II adapters

Hi Vincent,
(Hi Marc,)

Thank you for your in-depth look at it - especially regarding stale
things that reflect its age, such as cf->can_dlc vs. cf->len. This
driver has been gathering some dust. I've reworked the code according
to your review.

@Marc - could you please have a look at this?

The elmcan code currently has a dummy mailbox_read() function for
rx_offload because can_rx_offload_add_fifo() requires it - is this
intentional?

@Vincent - two more things have remained, and I hope it's okay once I
explain them:

1. _memstrcmp() - memcmp() vs. str(n)cmp()

The _memstrcmp() function does not compare strings, it compares raw
buffers. I am just using C strings for the fixed buffers to compare
against, as that allows for shorter and easier to read code. The NUL
byte at the end of those strings goes unused.

Also, I have not looked at the assembly produced, since the semantics
are different: str(n)cmp() needs to look for NUL bytes in the buffer(s),
which is unnecessary here. As a bonus, NUL will never even occur
because my code filters those bytes out upon reception from the UART
(it's a documented quirk of the ELM327).

Finally, even if I were to use strcmp(), the code would still look just
as ugly. Except the machine would also look for NUL bytes, and the next
human to read the code would wonder why I'm comparing strings and not
buffers.

Hence memcmp(), to help the code self-document and the compiler
optimise - I hope that's okay.

2. Useless parentheses in a for loop's condition:

I left those in there because it gets hard to read otherwise, IMHO.
It's really a matter of taste though, and if you insist, I'll remove
them. With if, it's easier to keep it readable:

	if (a == 1
         && b == 2) {
		...
	}

Whereas with for, I've already used multiple lines to visually separate
the initialiser, the condition, and the incrementer. Hence the
parentheses to visually separate the two subconditions.

Matter of taste really, and I'll change it if you insist.

That's it!
Everything else is already patched for a future v4 of the code. I'll CC
you once I send that out. But I'd like to wait for Marc's feedback
first.

Thanks,

Max