linux-kernel - Re: [RFC PATCH V2] checkpatch: Check output format style of __func_

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:	Mon, 14 Mar 2016 07:07:38 -0700
From:	Joe Perches <joe@...ches.com>
To:	Julia Lawall <julia.lawall@...6.fr>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	Andy Whitcroft <apw@...onical.com>,
	Dan Carpenter <error27@...il.com>,
	kernel-janitors <kernel-janitors@...r.kernel.org>,
	linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH V2] checkpatch: Check output format style of
 __func__ uses

On Mon, 2016-03-14 at 09:29 +0100, Julia Lawall wrote:
> On Mon, 14 Mar 2016, Joe Perches wrote:
> > On Mon, 2016-03-14 at 06:19 +0100, Julia Lawall wrote:
> > > On Sun, 13 Mar 2016, Joe Perches wrote:
> > > > Loggng messages that emit function names have many different forms.
> > > > Perhaps it'd be better for logging consistency and grep ease to
> > > > exclusively use "%s:"
> > > > 
> > > > As well, function tracing logging uses are generally unnecessary given
> > > > the kernel's function tracing (ftrace) capability.
> > > > 
> > > > Right now, grep shows these mixtures of forms:
> > > > 
> > > > 13704 "%s:"
> > > > 3839  "%s "
> > > > 2787  "%s()"
> > > > 
> > > > Some of these are macros definitions of various styles.
> > > > 
> > > > Unfortunately, given the complexity of these macro definition styles,
> > > > checkpatch isn't an ideal tool to find these macros.
> > > > 
> > > > Maybe a coccinelle script might be better suited to find and fix all
> > > > the various types of uses.
> > > > 
> > > > Add a --fix option for these logging messages with __func__.
> > >
> > > I'm not good enough at perl to really understand this.  Coudl you give an 
> > > example of what it does, and of what it does not do?
> >
> > For instance, this could do simple conversions like:
> >
> > $ diff --git a/arch/arm/common/mcpm_entry.c b/arch/arm/common/mcpm_entry.c
> > @@ -416 +416 @@ int __init mcpm_loopback(void (*cache_disable)(void))
> > -               pr_err("%s returned %d\n", __func__, ret);
> > +               pr_err("%s: returned %d\n", __func__, ret);
> >
> > But it couldn't find/convert a string concatenation:
> >
> > #define pch_dbg(adap, fmt, arg...)  \
> >         dev_dbg(adap->pch_adapter.dev.parent, "%s :" fmt, __func__, ##arg)
> 
> OK, are there any thoughts about what to do when __func__ is not in the
> first position?

Hard to say,

Uses with any or all of __FILE__, __LINE__, or __func__
in various combinations might be standardized.

A lot of those are debugging style macro and probably
could be converted to dynamic_debug style uses without
any mention of __FILE__, __LINE__, or __func__ at all
as dynamic_debug can optionally output these.

For macros like:

drivers/mtd/ubi/ubi.h-#define ubi_err(ubi, fmt, ...) pr_err(UBI_NAME_STR "%d error: %s: " fmt "\n", \
drivers/mtd/ubi/ubi.h:				      ubi->ubi_num, __func__, ##__VA_ARGS__)

it might be best to leave it alone. as that lets
grep find these a bit easier using "^ubi" though
code style this might be nicer on 3 lines

#define ubi_err(ubi, fmt, ...)						\
	pr_err(UBI_NAME_STR "%d error: %s: " fmt "\n", 			\
	       ubi->ubi_num, __func__, ##__VA_ARGS__)

or maybe even as a function rather than a macro
if it saves code space.  __func__ could then be
converted to __builtin_return_address(0)

for instance: https://lkml.org/lkml/2016/2/25/554