I'm testing on SCPH-50002 booting with latest wLE via USB.
I do have a saved cfg, video mode set to 1080i.
I do have a saved cfg, video mode set to 1080i.
Good news!Did some quick tests; Saving cfg in any video mode is fine now, no crashes, themes and Lang also seem fine. A couple things I experienced though.
I've tested on a slim using USB and SMB. They both seem very stable, but I must say I'm not a heavy user of OPL. I will test using FAT with HDD.First 2 times I tried to initiate HDD resulted in a crash. (Then it randomly worked every time after that).
Cover Art is not displaying. (My ART is bmp)
Scrolling through my games list too quickly results in a crash.
All these things could be related though since I'm doing it all from HDD, "may" be an issue with hdd support?
Loading themes from HDD is fine though, BMP background working also so probably not why cover art isn't working.
I'll look into this. I didn't notice it was missing. Thank you for testing!Only other thing was Notifications don't work which isn't a huge deal, I was never really happy with how it's coded anyway and wouldn't disagree if it needed to be removed.
What makes you think that? I've read somewhere that not everyone has the same issues. This makes me think not everyone is using the same (latest) version of all sources. Has anyone tried to git clone and compile everything from scratch? ps2toolchain, ps2sdk, ps2sdk-ports, gsKit, OPL?Due to issues with current versions of OPL which I 'think' are caused by a recent gsKit update..
It's good to hear the port is working great. Thank you for bringing the BMP issue to the attention. Would you be able to debug this issue? I've currently got too many things on my plate. Improving gcc9 is what's currently consuming all my free time.However there is still one problem with the newlib build of OPL and its just BMPs. I couldn't get the console to lock up at all by speeding through my games list or anything with HDD anymore which is great!
But BMP COVs definitely don't load (I noticed flashes of garbage when trying to load) PNG COVs load fine.
@Maximus32 I'm hoping this BMP issue can be resolved and then not only will it be great to have a newlib toolchain etc but it will also fix current issues with OPL.
I've recently been granted permissions on ps2dev also. So I can now pull my own PRI know the pull requests are currently open; @sp193 in order for OPL to be fixed it seems we need newlib or to revert a gsKit commit until newlib is merged.. I will try bisect tomorrow night if I have time to find out which commit is causing current problems.
Just a hunch, there seem to be 2 different issues atm.. if using r1440 compiled by elpatas; people are getting graphics glitches on default theme (likely caused by this version not being compiled with the latest gsKit so it's missing the 4 bit CLUT).. if using r1440 by akuhaks bot; people aren't getting graphics glitches but they are getting console lock ups (likely because this version is built with latest gsKit commits)..What makes you think that? I've read somewhere that not everyone has the same issues. This makes me think not everyone is using the same (latest) version of all sources. Has anyone tried to git clone and compile everything from scratch? ps2toolchain, ps2sdk, ps2sdk-ports, gsKit, OPL?
@El_Patas please make sure you keep ps2toolchain, ps2sdk, ps2sdk-ports and gsKit updated.if using r1440 compiled by elpatas; people are getting graphics glitches on default theme (likely caused by this version not being compiled with the latest gsKit so it's missing the 4 bit CLUT).. if using r1440 by akuhaks bot; people aren't getting graphics glitches but they are getting console lock ups (likely because this version is built with latest gsKit commits)..
Thank you for tracking it down to that commit. It's very strange that commit it causing the issue. It should be more safe now all UCAB allocations flush the data cache before using it. The only 'less safe' thing I see is that FlushCache(0); is replaced with SyncDCache(p, p+size-1);. The latter flushes only what needs to be flushed, the first one flushes everything. Flushing everything should not be needed, unless it accidentally flushes another bug ;-).Ok I rebuilt everything from scratch and compiled OPL and have freezing immediately at logo with an existing CFG (without it freezes in display settings). I then rolled back a couple commits in gsKit and rebuilt OPL and have it working without issues.. was able to narrow it down to this commit.
I don't like the idea of adding bugs to fix other bugs, so I'll see if I can find the cause of the lockup.For whatever reason its causing problems with the current builds but I remember when you found this issue and that its required for newlib to run OPL in HiRes without problems..so what I am hoping is that we can (temporarily) revert this commit and perhaps include it in the newlib pull request instead; that way when newlib is merged it will be included (since these lock ups aren't present in the newlib build) but it will also fix current issues people are having with OPL.
@Maximum32, as i said on Discord i update all this things always when a developer mark@El_Patas please make sure you keep ps2toolchain, ps2sdk, ps2sdk-ports and gsKit updated.
I didn't know becouse I'm not on discord, but good.@Maximum32, as i said on Discord i update all this things always when a developer mark
them or inform in the pull request of a new OPL commit if they are needed to update.
As Tupakaveli said me recently he forggot to mark in the pull request that was needed upate the gskit
for the 4bit clut.
Thanks for taking the time to find the cause, appreciated.The time function in ps2sdk has been recently changed, and the bug is there.
Sure, just finished work so I'll be able to test in a few hours.can you check if this fixes all the current freezing issues in OPL? I'll do the PR to ps2sdk when I get feedback it's all ok. You can also check the version you currently think is working (with the reverted gsKit commit). Try changing the resolution a few times, and see if you get the timeout screen for a normal amount of time.
Good one. I rebased onto the master branches shortly before sending the PR's. After the rebase I did a few simple checks, but didn't notice this problem.The inversion must've been rebased into newlib too at some point. Strange that it doesn't cause lockups there too.