- Dec 14, 2018
-
-
Neil Armstrong authored
-
Neil Armstrong authored
-
Neil Armstrong authored
-
Neil Armstrong authored
-
Signed-off-by:Tomeu Vizoso <tomeu.vizoso@collabora.com>
-
Signed-off-by: -
Otherwise instead of passing a pointer to the next atom, we pass a pointer 1-byte into the first atom. Signed-off-by: -
Signed-off-by:
-
- Dec 11, 2018
-
-
As at that moment no framebuffer has been set yet. Signed-off-by: -
Neil Armstrong authored
-
Some hardware supports source mods only for float operations. Make it possible to skip lowering to source mods in these cases. v2: use option flags instead of a boolean (Jason Ekstrand) Signed-off-by:
Gert Wollny <gert.wollny@collabora.com> Reviewed-by:
Jason Ekstrand <jason@jlekstrand.net>
-
Neil Armstrong authored
-
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
-
Signed-off-by: -
Signed-off-by: -
-
-
Signed-off-by: -
Signed-off-by: -
Nowadays mali_kbase tells us when we need to mmap() things, not the other way around. Unfortunately it's a little rude and doesn't seem to specify the right output flag for requiring an mmap, so we just tell if an mmap is required by checking for the magic number that's always the first cookie. Let's hope we never try multithreading with this! This might have something to do with the kbase API we're reporting.
-
When we started this project, I had 'interesting' ideas. Mainly, I had made our own custom headers for interacting with mali_kbase because the headers they had were a humongous mess. I wasn't aware the burden of maintainership it'd incur, and nowadays the kbase headers are a lot more acceptable anyway. Since we're updating to the latest mali_kbase uapi, I decided to just throw them out and start using the actual headers. Now we do that using a simple bridging header, panfrost-mali-base.h, which handles the various defines and other things needed to make including the kbase headers 'just work'. Additionally-move panfrost-int.h to panfrost-misc.h so we can stick some other required types into it. This commit doesn't get things onto the screen on my T82x board quite yet, but it does communicate with the kernel properly!
-
Since we're about to pull in all of ARM's official headers for their kernel driver instead of our own, and all of those are prefixed with mali-* or mali_*, let's rename ours to panfrost-* so there isn't any confusion. No functional changes.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
