linux-kernel - Re: [PATCH] tracing/probe: fix typo in pr_fmt in trace

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:   Wed, 6 Dec 2017 20:00:16 +0900
From:   Masahiro Yamada <yamada.masahiro@...ionext.com>
To:     Joe Perches <joe@...ches.com>
Cc:     Steven Rostedt <rostedt@...dmis.org>,
        Michal Marek <michal.lkml@...kovi.net>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        Jiri Kosina <jikos@...nel.org>,
        Song Liu <songliubraving@...com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        kernel-team@...com, Jiri Kosina <trivial@...nel.org>
Subject: Re: [PATCH] tracing/probe: fix typo in pr_fmt in trace_uprobe.c

2017-11-17 2:05 GMT+09:00 Joe Perches <joe@...ches.com>:
> On Wed, 2017-11-15 at 21:55 -0500, Steven Rostedt wrote:
>> On Wed, 15 Nov 2017 14:55:19 -0800
>> Joe Perches <joe@...ches.com> wrote:
>>
>> > Maybe use the much more common below instead?
>> >
>> > #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
>>
>> Is there someplace this is documented. My git grep foo isn't finding it
>> for me.
>
> Not so far as I know.
>
> Maybe Michal Marek or Masahiro Yamada might add some
> documentation in the future about these KBUILD defines.
>
> btw, there are about 1800 #define pr_fmt uses and most
> of them use KBUILD_MODNAME
>
> $ git grep -P "^\s*#\s*define\s+pr_fmt\b" | \
>   cut -f2- -d":" | \
>   sed -r 's/^\s*#\s*define\s+pr_fmt([^ ]+)\s++/#define pr_fmt\1 /' | \
>   sed -r 's/\)\s+/) /' | sort | uniq -c | sort -rn | head -10
>    1166 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
>     134 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
>      32 #define pr_fmt(fmt) "%s: " fmt, __func__
>      22 #define pr_fmt(fmt) KBUILD_MODNAME ":%s: " fmt, __func__
>      17 #define pr_fmt(fmt) KBUILD_MODNAME ":%s(): " fmt, __func__
>      13 #define pr_fmt(fmt) "xen:" KBUILD_MODNAME ": " fmt
>      11 #define pr_fmt(fmt) "arcnet:" KBUILD_MODNAME ": " fmt
>      10 #define pr_fmt(fmt) KBUILD_MODNAME ": %s: " fmt, __func__
>      10 #define pr_fmt(fmt) "IPv6: " fmt
>       9 #define pr_fmt(fmt) fmt
>


Slightly related topic.

Does KBUILD_MODNAME need to be correct,
or defined all the time?



One object could be shared among multiple modules, like follows:

  obj-m += foo.o
  obj-m += bar.o
  foo-objs := foo-bar-common.o foo-main.o
  bar-objs := foo-bar-common.o bar-main.o

In this case, it is debatable
what we should define for KBUILD_MODNAME
when compiling foo-bar-common.o


[1] -DKBUILD_MODNAME='"foo-bar-common"'
     (define base name of object name)

[2] -DKBUILD_MODNAME='"foo:bar"'
     (define colon separated module names)

[3] Do not define KBUILD_MODNAME at all

or any other idea?


[3] is the current behavior of Kbuild.
If KBUILD_MODNAME is referenced, it is a build error.


One object is shared among multiple modules,
for example:
lib/zstd/Makefile
drivers/mtd/tests/Makefile
drivers/net/ethernet/cavium/liquidio/Makefile




-- 
Best Regards
Masahiro Yamada