summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2013-09-09driver: rename a few utility functionsWladimir J. van der Laan
Add etna_ prefix to utility functions to make it clearer where they come from.
2013-09-09driver: make viewport affect clipping rectangle as wellWladimir J. van der Laan
I'm not sure this is defined by the GL spec, but makes it behave more like other GL drivers and fixes missile cam in d2x.
2013-09-09driver: fix cousins of MSAA hang bugWladimir J. van der Laan
- Change the order in `reset_context` as well - Clobber PS INPUTS state when MSAA samples changed to make sure it's always written, as we don't know the current value anymore.
2013-09-09include_v4_uapi: Added hardware performance counters structMaarten ter Huurne
It is needed when profiling support is enabled.
2013-09-09include_v4_uapi: Removed debug optionsMaarten ter Huurne
These are now internal to the kernel driver, no longer part of the ABI.
2013-09-09etnaviv: Fixed typo in comment in MakefileMaarten ter Huurne
2013-09-08driver: fix MSAA hang issueWladimir J. van der Laan
Changing the MSAA mode clobbers PS.INPUT_COUNT. Hence make sure that it is set before all other state.
2013-09-08driver: add additional debug flags `flush_all` and `finish_all`Wladimir J. van der Laan
2013-09-08tools: fix word size in etnaviv_gdbWladimir J. van der Laan
There was an error when using gpu-trace on 64 bit host systems
2013-09-08replay: update remaining etna_vidmem_unlock call in cube_etna2_gc2000 to new APIPhilipp Zabel
Commit 2fcaa58c3e0e763c78bb1d28a3d63380f5e65ce2 introduced an additional (optional) queue parameter to etna_vidmem_unlock. Update the remaining callsite in cube_etna2_gc2000.
2013-09-08documentation updates, dial gcc -g flag up to 11Wladimir J. van der Laan
2013-09-06driver: flush TS after flushing depth and stencilWladimir J. van der Laan
Flushing TS while still rendering to color/depth can sometimes hang, so flush TS after flushing color and depth instead of before.
2013-09-06driver: compiler improvements and fixesWladimir J. van der Laan
- Fix number of uniforms limit check - Fix over-eager passthrough optimization in `etna_compile_pass_optimize_outputs` - Comment updates and clarifications
2013-09-06etnaviv: use two-stage unlock, to avoid memory leak with v4 driversWladimir J. van der Laan
2013-09-05include_v4_uapi: Added stripped down version of the kernel headersMaarten ter Huurne
These headers contain 2384 lines, while the original include_v4 contains 20937 lines worth of headers. The compile time options that affect the ABI have been reduced to four: VIVANTE_NO_3D, VIVANTE_PROFILER, gcdSECURE_USER and gcdDEBUG. See gc_hal_options.h for details. The individual steps in stripping the headers can be found in the jz-3.10 branch of the GCW Zero kernel repository: https://github.com/gcwnow/linux/commits/jz-3.10
2013-09-04driver: return more informative nameWladimir J. van der Laan
2013-09-04driver: implement MSAAWladimir J. van der Laan
For now, use ETNA_DEBUG="msaa4x" or ETNA_DEBUG="msaa2x" to force usage of MSAA for screen surface (MESA may have its own way for this...).
2013-09-04tools: don't show INST_MEMs in normal state dumpWladimir J. van der Laan
Makes state dumps somewhat more readable.
2013-09-04fb_old: cube_companion → cube_companion_msaaWladimir J. van der Laan
Update old replay demo to run on cubox, to check how to get MSAA to work.
2013-09-03driver: uncomment set up TS before blitWladimir J. van der Laan
The code to set up the TS to the source surface before blit should be safe to use now. Only change TS if it differs from the desired state (so if necessary). Some flushing may still be needed when changing TS value we'll see that soon enough.
2013-09-03driver: parse ETNA_DEBUG environment variable for debug flagsWladimir J. van der Laan
This includes `no_ts` to disable tile status, and `cflush_all` to flush before every state update and draw call.
2013-09-02driver: make etna_compile_rs_state more flexibleWladimir J. van der Laan
Allow writing the RS command to another place except inside the etna_surface structure (mainly useful for debugging, and preparing to use the RS in other places such as blitting).
2013-09-02etnaviv: Avoid including gc_hal_kernel_context.hMaarten ter Huurne
Fixes the build when GCABI=v4. There is no gc_hal_kernel_context.h in include_v4/, but the only thing from that header that was actually required was the include of gc_hal_kernel_buffer.h, so include that directly.
2013-09-01driver: disable depth compressionWladimir J. van der Laan
Causes corruption (blocking) issues in glquake when TS is enabled.
2013-09-01driver: set up TS before blit to framebufferWladimir J. van der Laan
This appears to fix most of the TS problems. Re-enable TS again for color and depth framebuffer surfaces. Depth still seems to have some problems (visible in quake as see-through-walls).
2013-08-31fb: fix out-of-bounds bug in alpha blend sampleWladimir J. van der Laan
2013-08-29driver: comment update, small refactor of sync_contextWladimir J. van der Laan
2013-08-28driver: disable TS for nowWladimir J. van der Laan
We're not managing it correctly resulting in corruption. Better to trade some performance (not even sure how much) for correctness, so disable it.
2013-08-27driver: flush before some state changesWladimir J. van der Laan
There seems to be some corruption caused by changing certain states without a cache flush. This is especially visible in quake.
2013-08-26driver: fix polygon depth biasWladimir J. van der Laan
Number of units was wrong. This fixes shadows in quake.
2013-08-26etnaviv: allow tiling from/to non multiple of four coordinatesWladimir J. van der Laan
Should fix lightmap corruption in quake.
2013-08-26docs: add gc1000 in marvell to gpus comparisonWladimir J. van der Laan
2013-08-26Merge pull request #12 from tchebb/patch-1Wladimir J. van der Laan
Fix typo in driver version check
2013-08-26Fix typo in driver version checkThomas Hebb
2013-08-26Merge pull request #11 from tchebb/eurekaWladimir J. van der Laan
Add GPL drivers for the Chromecast (eureka) board (GC1000)
2013-08-26Add eureka GPL driverThomas Hebb
2013-08-26Use /dev/gal3d if availableThomas Hebb
2013-08-26include_v4: add GCABI_CHIPIDENTITY_EXTWladimir J. van der Laan
For some reason this was not defined, even though the v4 kernel structure has the concerning fields
2013-08-25Merge pull request #10 from tchebb/patch-1Wladimir J. van der Laan
Fix incorrect strcmp() return value check
2013-08-25Fix incorrect strcmp() return value checkThomas Hebb
2013-08-24fb: add tests for A8 and L8 texture formatsWladimir J. van der Laan
Try to find glquake lightmap bug...
2013-08-24driver: reject invalid vertex element specificationWladimir J. van der Laan
Return NULL in this case instead of a mock object to signal error. No rendering will be done with a NULL vertex element specification.
2013-08-24driver: show format name in debug messagesWladimir J. van der Laan
2013-08-23remove lots of trailing spacesWladimir J. van der Laan
whitespace only changes
2013-08-23driver: zero-initialize TS memoryWladimir J. van der Laan
It is important to initialize the TS to zero, as random pattern can result in crashes. Do this on the CPU as this only happens once per surface anyway and it's a small area, so it may not be worth queuing this to the GPU.
2013-08-23driver: fix scissor state in explicit contextWladimir J. van der Laan
This commit changes scissor state in reset to _FIXP.
2013-08-23fb: in viv_gpu_top add compile time option to use clock_Wladimir J. van der Laan
monotonous time should be more appropriate for a profiling tool, however it is not always available
2013-08-23driver: fix initialisation orderWladimir J. van der Laan
2013-08-23comments/documentation updateWladimir J. van der Laan
2013-08-23driver: move transfer pool creation/destruction to etna_transferWladimir J. van der Laan
While changing this I noticed that the transfer pool was not getting freed on pipe destroy, this commit fixes this small memory leak as well.
generated by cgit (git 2.34.1) at 2025-07-16 20:04:40 +0000