On 12/31/2012 02:22 PM, Terje Bergström wrote:
On 28.12.2012 22:48, Thierry Reding wrote:
I disagree. We shouldn't be hiding this kind of detail behind an #ifdef. Instead it should be detected at runtime. Otherwise you'll need to build different versions of libdrm for every generation of Tegra. That may be fine for NVIDIA provided BSPs, but distributions would have a very hard time dealing with that. What we want is software that works unmodified on as many generations of Tegra as possible.
I agree. The fences will be rejected by kernel and kernel knows about number of sync points in each host1x revision. So, we could just submit and look at return code.
Agree. If we are able to do something dynamically, normally that'll be better.
Terje, we can get the Tegra version in FUSE. I think we don't need this kind of try-catch logics.
Mark
Terje