Make ffb driver build on x86-64. The fix is two fold. First, use the

proper DRM_CAS_RESULT to declare the variable used to store the result
of DRM_CAS.  Second, only use the "real" versions of LOCK_HARDWARE and
UNLOCK_HARDWARE on SPARC.  That's the only platform where the hardware
can really exist.
This commit is contained in:
Ian Romanick 2005-08-11 04:25:36 +00:00
parent 967b006f51
commit ac887d4211

View file

@ -8,14 +8,18 @@
extern void ffbXMesaUpdateState(ffbContextPtr fmesa);
#define FFB_UPDATE_STATE(fmesa) ffbXMesaUpdateState(fmesa)
/* Lock the hardware and validate our state. */
#if defined(__i386__)
/* Lock the hardware and validate our state. This hardware can only ever
* exist on SPARC platforms. Don't bother building the real LOCK_HARDWARE and
* UNLOCK_HARDWARE code on non-SPARC platforms. The only reason the driver
* gets built on non-SPARC is to catch build breakages earlier.
*/
#if !defined(__sparc__)
#define LOCK_HARDWARE(fmesa)
#define UNLOCK_HARDWARE(fmesa)
#else
#define LOCK_HARDWARE(fmesa) \
do { \
int __ret=0; \
DRM_CAS_RESULT(__ret); \
DRM_CAS(fmesa->driHwLock, fmesa->hHWContext, \
(DRM_LOCK_HELD | fmesa->hHWContext), __ret);\
if (__ret) { \