xfs
[Top] [All Lists]

Re: [PATCH 2/2] OS X partial support

To: Jan Tulak <jtulak@xxxxxxxxxx>
Subject: Re: [PATCH 2/2] OS X partial support
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Thu, 16 Jul 2015 11:25:08 +1000
Cc: xfs@xxxxxxxxxxx
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <1436964412-12093-3-git-send-email-jtulak@xxxxxxxxxx>
References: <1436964412-12093-1-git-send-email-jtulak@xxxxxxxxxx> <1436964412-12093-3-git-send-email-jtulak@xxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Wed, Jul 15, 2015 at 02:46:52PM +0200, Jan Tulak wrote:
> This patch introduces changes to support current OS X (10.10 Yosemite).
> The previous OS X code base is about 4 years old and both OS X and
> XFS changed, so there had to be done some work to make it compilable.
> 
> - Because some missing APIs in OS X (like fstab/mtab) and the
>   unability to mount XFS, not all tools are compiled.

hmmmm. We should still try to compile them, otherwise anyone
doing dev work on OS X is going to make changes that aren't
complete (i.e. cause compile errors on other platforms).

> - fls (find last bit) calls were moved to platform headers,
>   as OS X has a native implementation.

I think that's the wrong way to do this - this is what autoconf is
for. i.e. create a "HAVE_FLS" check, and surround the fls code
in bitops.h with "#ifndef HAVE_FLS"....

> - xfs_db currently has only a dummy timer implementation

xfs_repair, actually ;)

> - Various other edits  - some API translations, some, missing
>   defines.
> 
> The code compiles using Apple's clang, although it throws some
> warnings. Some of them are complaints about deprecated calls, some
> are false-positives and there are also some incompatible type
> assignments/comparisons, so there is a place for further patches.

Let's try and keep the changes within the darwin and autoconf code
as much as possible. 

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>