Intel driver build failed (tried 2.4.3 and 2.6.1) when --disable-dri passed to 'configure'. The errors are (from 2.6.1):
In file included from i830.h:80,
/usr/include/intel_bufmgr.h:92: error: expected declaration specifiers or ‘...’ before ‘drm_clip_rect_t’
In file included from i810_driver.c:88:
i830.h:734: error: expected specifier-qualifier-list before ‘drmmode_rec’
In file included from i830.h:766,
i830_batchbuffer.h:46: error: expected declaration specifiers or ‘...’ before ‘GLuint’
i830_batchbuffer.h: In function ‘intel_batch_require_space’:
i830_batchbuffer.h:49: error: ‘sz’ undeclared (first use in this function)
i830_batchbuffer.h:49: error: (Each undeclared identifier is reported only once
i830_batchbuffer.h:49: error: for each function it appears in.)
i830_batchbuffer.h: In function ‘intel_batch_start_atomic’:
i830_batchbuffer.h:59: error: too many arguments to function ‘intel_batch_require_space’
i810_driver.c: In function ‘I810ScreenInit’:
i810_driver.c:2389: warning: unused variable ‘driFrom’
make: *** [i810_driver.lo] Error 1
We don't actively support disabling DRI, but accept patches to fix it currently (this will change at some point).
Created attachment 23448 [details] [review]
The patch fixes the compilation when --disable-dri passed to configure
The attached patch fixes the compilation when --disable-dri passed to configure. Perhaps the approach totally broken, but at least the driver works (only tried with --disable-dri)
I'd like to remove --disable-dri option than having this kind of patch.
I think we're in agreement that we don't want to support a non-DRI path, and patches since 2.7 have taken us further down that path.