xf86UDelay was used - removal breaks drivers

[Alan Coopersmith]

Matt Turner wrote:
>> Are updates going to be issued to all these drivers to use a different
>> function, or should the removal be reverted?
> 
> Attached are patches that replace calls to xf86UDelay with usleep. I
> couldn't find anywhere in -glint or -neomagic that xf86UDelay was
> called--so if there is a place, point it out to me. Also attached is a
> patch to chmod +x autogen.sh in -ast. The patches are compile tested
> only.

ajax beat you to neomagic:

commit 4e2afb88212582df60bca3ec8c447965a5a66fd5
Author: Adam Jackson <ajax at redhat.com>
Date:   Mon Oct 26 15:43:50 2009 -0400

    s/xf86UDelay/usleep/g

glint seems to have been a false match in my grep - I hit the call to
Permedia2I2CUDelay.

> While I did not mean to break these drivers, the fact that they went
> for a couple months without any feedback tells me that they're
> entirely unused.

You can't assume that - just that few people run directly from git on
lesser used hardware.   Most developers running from git do so on
widely used desktop or laptop hardware, not old machines or rackmount
servers.

I know we have -ast users, since that's the remote console in a number
of current Sun servers (both SPARC & x64) - but they get stable builds
shipped, not git head.

Neomagic was the source of the two bugs I saw, so it has some users.

I expect -rendition is entirely unused since it's ancient.   (Last time
I looked for info on it, I found that Windows 98 dropped the driver for
it, since it was obsolete even then.)  glint I don't know.

-- 
	-Alan Coopersmith-           alan.coopersmith at sun.com
	 Sun Microsystems, Inc. - X Window System Engineering