Hi, I have seen that ED has announced that there is a new set of TILER patches:
https://pyra-handheld.com/boards/threads/wrapping-everything-up.83465/
"zmatt has reimplemented his rotation hack (so the Pyra is finally fully usable again!)"
Is this reimplementaion available anywhere for integration into the letux tree?
What I would like to suggest to better work in concert: please also announce such developments here on this official Pyra kernel development mailing list.
And it would be useful to provide some technical background information (what has changed any why, what was the bug?).
BR and thanks, Nikolaus
On 12/8/18 1:33 PM, H. Nikolaus Schaller wrote:
Hi, I have seen that ED has announced that there is a new set of TILER patches:
https://pyra-handheld.com/boards/threads/wrapping-everything-up.83465/
"zmatt has reimplemented his rotation hack (so the Pyra is finally fully usable again!)"
Is this reimplementaion available anywhere for integration into the letux tree?
https://github.com/mvduin/linux/commits/letux-4.19.3/hack
I applied it to the 4.19.4 kernel, 4.19.3 is very unstable.
It's a quick hack though, not the cleaned up version zmatt mentioned here before.
Am 08.12.2018 um 15:57 schrieb aTc atc@k-n-p.org:
On 12/8/18 1:33 PM, H. Nikolaus Schaller wrote:
Hi, I have seen that ED has announced that there is a new set of TILER patches: https://pyra-handheld.com/boards/threads/wrapping-everything-up.83465/ "zmatt has reimplemented his rotation hack (so the Pyra is finally fully usable again!)" Is this reimplementaion available anywhere for integration into the letux tree?
Many thanks!
I applied it to the 4.19.4 kernel, 4.19.3 is very unstable.
It's a quick hack though, not the cleaned up version zmatt mentioned here before.
No problem.
I just want to make sure that future letux kernels include the latest fixes we know. (even if hacks) and do not make a regression if we switch to 4.19.8 (bug fix for ext4 data loss issue - which might not affect us, but who knows?) or 4.20.0.
This should make life easier for everyone.
The best of course would be if we could get at least some of our TILER patches upstream so that they melt away our private patch sets... That is the standard methodology of Letux kernel development: hack something so that the Pyra (and other devices) work. Then keep it working with upstream updates. And try to clean things up and submit upstream.
BR and a nice weekend, Nikolaus
Hi all,
Am 08.12.2018 um 15:57 schrieb aTc atc@k-n-p.org:
On 12/8/18 1:33 PM, H. Nikolaus Schaller wrote:
Hi, I have seen that ED has announced that there is a new set of TILER patches: https://pyra-handheld.com/boards/threads/wrapping-everything-up.83465/ "zmatt has reimplemented his rotation hack (so the Pyra is finally fully usable again!)" Is this reimplementaion available anywhere for integration into the letux tree?
I applied it to the 4.19.4 kernel, 4.19.3 is very unstable.
It's a quick hack though, not the cleaned up version zmatt mentioned here before.
I have looked through them and they work if applied. But one patch breaks on other platforms (omap3).
I remember that we had already discussed that and I found that
http://git.goldelico.com/?p=letux-kernel.git;a=commit;h=4f7cefb1416dca96d2a4...
tries to fix it. But makes it a platform dependent compile.
So we had discussed that a while ago, worked on a solution and then forgot about it because it worked with X11 on all platforms.
Therefore I'd suggest a better fix is to detect OMAP5 in
drivers/video/fbdev/core/fbmem.c
and make the #if dynamic.
I remember there was some API to check for the exact platform. Something where patterns could be provided and it was used to decide between AM/DM37xx and AM33xx. Maybe in the PVR/SGX driver stuff?
So something like
if (platform_is("OMAP543[02]")) vma->vm_page_prot = pgprot_device(vma->vm_page_prot); else // all others incl. OMAP3 w/o TILER fb_pgprotect(file, vma, start);
could be the better hack...
Does anyone remember how this "platform_is" function is called?
BR, Nikolaus
Am 08.12.2018 um 23:07 schrieb H. Nikolaus Schaller hns@goldelico.com:
Hi all,
Am 08.12.2018 um 15:57 schrieb aTc atc@k-n-p.org:
On 12/8/18 1:33 PM, H. Nikolaus Schaller wrote:
Hi, I have seen that ED has announced that there is a new set of TILER patches: https://pyra-handheld.com/boards/threads/wrapping-everything-up.83465/ "zmatt has reimplemented his rotation hack (so the Pyra is finally fully usable again!)" Is this reimplementaion available anywhere for integration into the letux tree?
I applied it to the 4.19.4 kernel, 4.19.3 is very unstable.
It's a quick hack though, not the cleaned up version zmatt mentioned here before.
I have looked through them and they work if applied. But one patch breaks on other platforms (omap3).
I remember that we had already discussed that and I found that
http://git.goldelico.com/?p=letux-kernel.git;a=commit;h=4f7cefb1416dca96d2a4...
tries to fix it. But makes it a platform dependent compile.
So we had discussed that a while ago, worked on a solution and then forgot about it because it worked with X11 on all platforms.
Therefore I'd suggest a better fix is to detect OMAP5 in
drivers/video/fbdev/core/fbmem.c
and make the #if dynamic.
I remember there was some API to check for the exact platform. Something where patterns could be provided and it was used to decide between AM/DM37xx and AM33xx. Maybe in the PVR/SGX driver stuff?
So something like
if (platform_is("OMAP543[02]")) vma->vm_page_prot = pgprot_device(vma->vm_page_prot); else // all others incl. OMAP3 w/o TILER fb_pgprotect(file, vma, start);
could be the better hack...
Does anyone remember how this "platform_is" function is called?
I think I found it: soc_device_match()
https://elixir.bootlin.com/linux/v4.20-rc5/source/drivers/gpu/drm/omapdrm/ds...
BR, Nikolaus
http://projects.goldelico.com/p/gta04-kernel/ Letux-kernel mailing list Letux-kernel@openphoenux.org http://lists.goldelico.com/mailman/listinfo.cgi/letux-kernel
-
aTc -
H. Nikolaus Schaller