Re: [tip:perf/core] perf: Fix endianness argument compatibility with OPT_BOOLEAN() and introduce OPT_INCR()

From: Erik Faye-Lund <kusmabite@googlemail.com>
Date: Apr 14, 2010, 15:12 UTC
Message-ID: <j2k40aa078e1004140812s3d9fab4auf559b1e87af60b5b@mail.gmail.com>
In-Reply-To: <tip-c05556421742eb47f80301767653a4bcb19de9de@git.kernel.org>

I'd just like to point out that I'm commenting from the Git pont-of-view, since this patch was CC'ed to the Git mailing-list.

On Wed, Apr 14, 2010 at 11:36 AM, tip-bot for Ian Munsie <imunsie@au.ibm.com> wrote:

Show 13 quoted lines

> Commit-ID:  c05556421742eb47f80301767653a4bcb19de9de
> Gitweb:     http://git.kernel.org/tip/c05556421742eb47f80301767653a4bcb19de9de
> Author:     Ian Munsie <imunsie@au.ibm.com>
> AuthorDate: Tue, 13 Apr 2010 18:37:33 +1000
> Committer:  Ingo Molnar <mingo@elte.hu>
> CommitDate: Wed, 14 Apr 2010 11:26:44 +0200
>
> perf: Fix endianness argument compatibility with OPT_BOOLEAN() and introduce OPT_INCR()
>
> Parsing an option from the command line with OPT_BOOLEAN on a
> bool data type would not work on a big-endian machine due to the
> manner in which the boolean was being cast into an int and
> incremented.

We don't use C99 features like the "bool"-type in Git, so this shouldn't really be a problem in our end.

Show 5 quoted lines

> For example, running 'perf probe --list' on a
> PowerPC machine would fail to properly set the list_events bool
> and would therefore print out the usage information and
> terminate.
>

I don't know what this 'perf'-thing is, but it sounds like it's some external client for the parse-options-API that uses OPT_BOOLEAN incorrectly.

Looking at linux-next.git, it seems so. The code in tools/perf/ seems to use OPT_BOOLEAN with C99 bools, and it seems to have imported git's parse-options API (in commit 0780060124011b94af55830939c86cc0916be0f5, it seems).

I'd suggest either fixing the call-sites to use int, or just keeping these changes in linux-next.git.

-- 
Erik "kusma" Faye-Lund