summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorRussell King <rmk@armlinux.org.uk>2017-03-16 12:09:30 +0000
committerRussell King <rmk@armlinux.org.uk>2017-03-16 12:09:30 +0000
commitfc0502197b821f601e8598c6465d4d2e1c8533fb (patch)
tree7fd15a782bbe6a6fb9bb7b90a2306250dd6bf557
parent6e7d019fffee3f2adfbe2c523bc8f9e48518d085 (diff)
FAQ: add some more questions
Add questions about libdrm_armada and libdrm_etnaviv. Signed-off-by: Russell King <rmk@armlinux.org.uk>
-rw-r--r--FAQ25
1 files changed, 25 insertions, 0 deletions
diff --git a/FAQ b/FAQ
index 142d00d..f99201a 100644
--- a/FAQ
+++ b/FAQ
@@ -50,3 +50,28 @@ A2. No, it should not be necessary to build etna_viv or configure it
etnaviv/state.xml.h
etnaviv/state_2d.xml.h
etnaviv/viv.h
+
+Q3. Why does xf86-video-armada use libdrm_armada and not libdrm_etnaviv?
+
+A3. libdrm_armada provides the DDX with the ability to interface with
+ the KMS (display) side, it has almost nothing to do with the GPU
+ drivers. It provides efficient allocation of buffers for the KMS,
+ and provides various features missing from the core libdrm that are
+ necessary to interwork with the GPU drivers.
+
+ libdrm_etnaviv is not used by the etnaviv DRM GPU driver as the
+ libdrm_etnaviv interfaces are too restrictive to allow efficient
+ command stream generation, lacks proper buffer caching for efficient
+ buffer handling, and implementation would mean yours truely would
+ need to maintain three separate GPU drivers for the same hardware.
+
+Q4. Is the dependency on libdrm_armada going to be removed?
+
+A4. Not any time soon. The libdrm_armada is necessary for the KMS side
+ as explained in Q3.
+
+Q5. Is libdrm_etnaviv going to be integrated?
+
+A5. As long as the DDX supports the libetnaviv (GALcore compatible) GPU
+ backend, and libdrm_etnaviv suffers the problems mentioned in Q3,
+ moving to libdrm_etnaviv would be a backwards step.