[PATCH] Massive clean-up to MGA server-side DRI code

Ian Romanick idr at us.ibm.com
Mon Jun 6 19:23:12 PDT 2005

Hash: SHA1

Eric Anholt wrote:
> On Mon, 2005-06-06 at 15:33 -0700, Ian Romanick wrote:
>>Here's a largely "cosmetic" patch for the MGA server-side DRI code.
>>This is something of a "pre-patch" before I send out the patch for
>>bugzilla #3259 (MGA: Add support for PCI cards).
>>If people are okay with these changes, I'll go ahead and commit them,
>>and then I'll send out the *real* 3259 patch right quick.
>>I have attached my intended CVS log and the patch itself.
>>plain text document attachment (mga_pre-3259-01.explain.txt)
>>Numerous comments, clean-ups, and refactorings for the DRI portion of
>>the MGA driver.
>>- Use mga_drm.h instead of mga_common.h.  Due to this, all the
>>  MGA-specific DRM names were changed to generic versions (i.e.,
>>  drmMGALock became drm_lock_t).
> Previously we have intentionally not used *_drm.h from the DDX because
> we couldn't guarantee its existence, I thought.  I do very much like
> using the same type names for things at least.  I don't think I object
> to the change to *_drm.h anyway, just want to make sure we're clear if
> this is a good idea.

The only discussion that I recall, and it was quite a long time ago, was
over whether or not the *_drm.h files were platform independent.  That
is, if you #include mga_drm.h on Linux you get the "same" thing as on
*BSD.  Since we moved to the shared-core setup, this is the case.  If
there were other issues, I sure don't remember them.

The main reason I made this change was that I'll be adding a couple
ioctls and associated data structures in the MGA DRM.  I don't want to
have to maintain the same information in 2 different places, whether
they have the same names or not.
Version: GnuPG v1.4.1 (MingW32)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org


More information about the xorg mailing list