libera/#maemo-leste/ Saturday, 2021-10-23

tmlindfreemangordon: heh finally got the ocp reg fix done.. https://lists.goldelico.com/pipermail/openpvrsgx-devgroup/2021-October/000465.html12:13
tmlindas an extra bonus we now have basic register read/write support in pvr-drv :)12:13
tmlindi also did patches to add interrupt handling to prv-drv, but the patch kept sprawling to several thousands of lines of pvr code to remove.. needs more time12:14
tmlindand testing. i only tried the fix above to init sgx on few omaps i have in my rack12:15
Wizzupgreat work :)12:16
tmlindthanks, it's a pleasure (TM)12:16
uvosWizzup: ERROR: Permission to maemo-leste/clown-boot-kexec.git denied15:45
uvosso the sphone with greatly redisgned backend is on the way to devel15:47
uvosneeds testing15:47
uvosalso its in need of a address book plugin and other backends etc15:48
uvosalso its stable again now to the point that patches are welcome15:48
uvosinterface wise15:48
Wizzupuvos: added15:49
uvosWizzup: i think that pusing the modules there is gona fail15:51
uvosWizzup: because it contains the linux kernels history15:52
Wizzupok, so you want multiple repos, one per?15:53
uvosno its fine15:53
uvoslooks like github is not rejecting the push15:53
Wizzupyou can do diff branches too15:53
Wizzupoh right15:53
Wizzupyou just mean size wise15:53
lelIMbackK edited a repository: https://github.com/maemo-leste/clown-boot-kexec15:56
Wizzupuvos: I am thinking of making the droid3 dts now use motorola,droid3 as compatible17:14
WizzupI suppose that shouldn't break things?17:14
Wizzupalso added the pstore patch now17:14
uvosdid you add  the compatible to the cpcap regulator quirk?17:28
uvosor rather the assignment17:28
uvoscpcap driver needs to know the d3 is wired like xt87517:28
uvosarch/arm/mach-omap2/pmic-cpcap.c of_machine_is_compatible("motorola,droid-bionic")17:31
Wizzupok17:53
Wizzupuvos: so I'll make that || droid3 then18:22
Wizzupsince it seems to work (?)18:22
uvosyeah sure19:02
Wizzupuvos: hm, not sure if this relates to pstore patch or my dts changes, but running into this atm, will try to figure out what triggers it https://dpaste.com/D4885QHTC19:22
Wizzupsmells like pstore though?19:23
Wizzupmaybe the ramoops entry in the dts is not ok for the droid319:23
uvosyes thats pstore crashing19:24
uvosmaybe the ram area it allocates is out of range for d319:25
Wizzupmhm19:25
Wizzupyeah with the reserved-memory stuff gone in the dts it boots fine19:27
uvoswell where is the reserved memory?19:27
Wizzupno idea :) I haven't been able to decode the dtb properly19:28
Wizzupnot in android dmesg either it seems19:43
Wizzupthe default android shell truly is a crime against humanity19:50
uvosxD19:50
uvosthe shell or the teminal emulator?19:50
uvosor both19:50
Wizzupboth I guess19:57
Wizzuphttps://github.com/MerlijnWajer/bionic-clown-boot/tree/solana19:57
Wizzupthis works now apart from the install.sh assuming /sdcard is writable19:57
uvoswhat about /emstorage/0/ or something like that19:58
WizzupI did some magic as root19:58
uvosaccuatlly i think that was introduced in android 4.019:58
Wizzupmkdir /preinstall/foo ; chmod 777 /preinstall/foo19:58
Wizzupand then just use that19:58
Wizzupthat works19:58
Wizzupin any case this is without pstore19:58
WizzupI think for me to properly fix brightness and modem I'll need to look at getting dtc to read the old motorola dtb things19:59
Wizzupbasically figure out what tmlind did at the time and recreate the code :p19:59
Wizzup(also for pstore / reserved memory)19:59
uvosso do you plan on leaving it like that?20:00
uvosim not terribly happy20:00
Wizzupuvos: what, the clown boot tree?20:01
uvosyeah20:01
Wizzupno20:01
uvosok20:01
WizzupI think at this point we can figure out what the best way would be to merge it, if you want it merged20:01
uvossure mostly i dont want to maintain the kernel image in 2 palces20:02
uvosand most of the other sutff is common as well20:02
uvosbb and su and so on20:02
WizzupI am not sure if it's the same, so I just went with what the root exploit provided20:02
uvosalso you did try root404.sh on solana right?20:02
uvosthe stuff in root is built agains 2.3 sdk20:03
uvosso should work fine20:03
Wizzuppretty sure I did and I don't think it worked20:03
Wizzup(404)20:03
uvosok20:03
Wizzupjust squashed some commits20:03
uvosboot.img isent needed at all20:03
Wizzupok, I wasn't sure why you flashed it in the README20:03
WizzupI've just taken the assumption that whatever you had in there was necessary20:04
uvoson xt875 we use the 3.0.8 kernel to kexec from20:04
Wizzupaha20:04
uvosthe one that shipped with 4.0 and is identical to the d4 one but android 4.1 comes with a slightly different kernel20:05
uvosso i have everyone downgrade the kernel20:05
Wizzupin any case this will just do nothing at worst, at best it prevents folks from flashing a bad boot.img20:05
Wizzupbut sure I can remove it, as long as boot.img is also gone20:05
Wizzupso regarding maintaining kernel images20:05
uvosalso allow-mbmloader-flashing-mbm.bin is probubly unessecary20:06
WizzupI think the first step is  for you or me to verify that the 5.10.74 that I made also works for you20:06
Wizzupno, that is necessary I think20:06
uvosits not needed to flash bspw on old versions of mbm20:06
WizzupI was not able to flash bwsp otherwise20:06
WizzupIIRC20:06
uvoson d4 /targa at least20:06
uvosok20:06
uvosmaybe d3 mbm is to new20:06
Wizzup(maybe it was just for boot, cannot remember)20:06
WizzupI was thinking of testing new kernels for kexecboot purposes simply from kexecboot20:07
uvosidealy i think we should just install the kernel and have kexecbootstart.sh figure out what dts to use at runtime imo20:07
Wizzupshould work, I think20:07
uvossure20:08
Wizzupwe will still need different directories or branches for bionic and d420:08
Wizzuper d320:08
uvosonly a different allow-mbmloader-flashing-mbm.bin, root234.sh the motofail binary should be different20:09
Wizzupparazyd: idk if you're going to be around one of these days but I have a few changes for the mapphone kernel to add d3 support20:09
uvosand dtb that gets installed on both20:09
Wizzupuvos: what about the kexec modules20:09
uvosoh right20:09
uvosyeah20:10
Wizzupbasically everything in kexecboot/*20:10
Wizzupapart from kernel20:10
Wizzupand maybe kexec binary20:10
uvosyes20:11
Wizzupupdated this a bit https://leste.maemo.org/Motorola_Droid_320:16
Wizzupparazyd: also I think the wiki icons are still broken20:16
uvosmaybe make the page title recognize that the device is sold as Milestone 3 outside of US20:19
Wizzupwith various codenames you mean20:20
Wizzup?20:20
uvosits nota  code name20:20
uvoscode name is solana everywhere20:20
Wizzupok20:20
uvosthe marketing name for droids released outside the us is Milestone20:21
uvosbecause lucas arts has the exlusive rights to the name "droid"20:21
Wizzupok20:21
Wizzuplol20:21
Wizzupshall I rename the wiki page, or just make it a clear note at the top?20:21
uvoswhatever you prefer20:21
Wizzupdone20:22
Wizzupinteresting it's also suffering from the screen redraw problems the d4 had before you fixed it20:25
Wizzupso probably some pvr clock rate problem20:25
uvoswell its not really fixed, it happens whenever sgx takes over a certain amount of ms to render the frame. due to some timeing issue somewhere tiler or dss fails to pickup the frame.20:32
uvosuping the clock rate just made it happen less20:33
uvosit sill happens just the same if you load sgx enough20:33
Wizzupah20:33
Wizzupwell it might be gone with 1.17720:33
Wizzup1.17(20:33
uvosspoiler allert: its not20:33
Wizzuphehe20:33
Wizzupso I made notes of all the things currently broken https://leste.maemo.org/Motorola_Droid_3#Notes20:33
WizzupI think it's about time I'm going to call it done for now and pick it up later20:33
Wizzupuvos: btw, how are you sure it's not gone, it might be with modesetting ?20:35
uvosi happens in wayland20:36
Wizzupaha20:36
uvos*it20:36
Wizzupwell, wayland is not the gold standard20:43
Wizzupbut yeah20:43
uvosyou can see it happen in plain drm too20:43
freemangordonYAY, fast path works :)20:46
uvos;)20:46
uvosfreemangordon: sweeeeet20:46
uvosfreemangordon: on n900 i presume20:46
freemangordonon d420:46
freemangordonwill try on n900 in couple of minutes20:47
uvosdident you have issues with sgx crashing?20:47
uvosnot that im complaining20:47
Wizzupfreemangordon: cool!20:47
Wizzupuvos: hm now I wonder how 5.11 boots since it should have pstore enabled20:58
Wizzupit's possible that the patch that I applied is ok for 5.11 but not for 5.1020:58
Wizzupas in I didn't remove that from the 5.11 mapphone dts, and it's enabled in our -devel kernel20:59
uvospretty sure we have been using exactly that patch since like 5.820:59
Wizzupright, will have to think about what it could be then20:59
uvosbut do check21:00
uvosif it wasent ajusted at some point21:00
parazydWizzup: Yeah I'm coming back from my trip in two days so I can look at everything21:09
parazydI'm at a conference in Portugal21:09
freemangordonuvos: on n900 it is way more stable than on d421:15
uvosfreemangordon: ok21:20
uvoshows perf?21:20
uvoshow soon can we get this packaged21:20
* uvos excited21:20
freemangordonhmm, not good :(21:33
freemangordonglmark2 Score: 2021:33
freemangordonsomething happens on CPU21:34
uvosn900 or d4?21:34
freemangordondespite I am issuing GL commands21:34
freemangordonn90021:34
uvossounds like how ddk1.9 performs21:34
uvosd4 also pegs the cpu doing something21:34
freemangordon"slow" path ( using mmapped memory) is a bit faster21:34
freemangordonthis is weird21:34
uvospretty sure ddk1.9 on d4 runs into the same bottle neck21:35
uvosglmark dose like 35 ish iirc21:35
uvossuggests cpu is doing a copy no?21:35
freemangordonwith chromeos dri?21:35
freemangordonyeah, it is doing copy it seems21:36
uvosfreemangordon: no with ddk1.9/xorg21:36
freemangordonah21:36
uvoschomeos dri is fast on wayland21:36
uvosand cant get glmark dri to run21:36
freemangordonseems the drivers are ok, it is just dri that misbehaves21:36
freemangordonwill try through shim tomorrow, to see what happens21:37
freemangordonuvos: what is "can't get to run" means? it just hangs?21:37
freemangordonugh21:38
freemangordonsorry, I am tired21:38
uvosyeah with cromeos mesa it hangs21:38
uvoswith full blobs it works fine21:38
freemangordonsame here21:38
uvoslast time i tried21:38
freemangordonah, ok21:38
uvos(long ago)21:38
freemangordonyeah, same here (hangs with chromeos21:38
uvosok21:39
freemangordonanyway, going to have some rest, will investigate tomorrow21:39
freemangordonttyl21:39
uvosok good night :)21:39

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!