Code Geass Hangyaku no Lelouch Lost Colors Chinese translate blackscreen on startup

[daniel229]

PSP1000 and PSP2000 both blackscreen since unknownbrackets@d273dee in #8296

Since #7952 only PSP2000 works fine

[unknownbrackets]

Are there any sceIo errors logged now? Perhaps we could JpcspTrace those functions to see if something different is happening.

-[Unknown]

[daniel229]

I saw this,then get the Invalid address

user_main    E[HLE]: Util\BlockAllocator.cpp:139 Block Allocator (08800000-0a000000) failed to allocate 21991424 (014f9000) bytes of contiguous memory

[unknownbrackets]

What does it allocate before that? We could log the allocation calls.

-[Unknown]

[daniel229]

debuglog here
https://gist.github.com/daniel229/534f43807e8b5cd689be

[unknownbrackets]

Hmm. Here it looks like the timing of root being deallocated is wrong.

When root first exists and creates the main_thread:

41:44:492 main_thread  I[HLE]: Util\BlockAllocator.cpp:395 Block: 08800000 - 08804000 size 00004000 taken=1 tag=usersystemlib
41:44:492 main_thread  I[HLE]: Util\BlockAllocator.cpp:395 Block: 08804000 - 08a71100 size 0026d100 taken=1 tag=ELF/simple
41:44:492 main_thread  I[HLE]: Util\BlockAllocator.cpp:395 Block: 08a71100 - 09fbf000 size 0154df00 taken=0 tag=(untitled)
41:44:492 main_thread  I[HLE]: Util\BlockAllocator.cpp:395 Block: 09fbf000 - 09fc0000 size 00001000 taken=1 tag=stack/main_thread
41:44:492 main_thread  I[HLE]: Util\BlockAllocator.cpp:395 Block: 09fc0000 - 0a000000 size 00040000 taken=1 tag=stack/root

After that, root frees itself as it should:

41:44:515 main_thread  I[MODULE]: HLE\sceKernelModule.cpp:1729 277=sceKernelLoadModule(name=disc0:/PSP_GAME/SYSDIR/BOOT.BIN,flag=00000000,(...))
41:44:515 root         D[KERNEL]: HLE\sceKernelThread.cpp:2952 Context switch: main_thread -> root (276->275, pc: 082214e4->08220c90, module loaded) +12us

And then "simple" is created:

41:44:516 simple       D[KERNEL]: HLE\sceKernelThread.cpp:2952 Context switch: main_thread -> simple (276->278, pc: 0822150c->0883c95c, started module) +164us

Next we have user_main get created:

41:44:516 simple       I[KERNEL]: HLE\sceKernelThread.cpp:1943 279=sceKernelCreateThread(user_main, 0883ca70, 00000020, 262144, 80000000, 00000000)

And then user_main frees simple:

41:44:516 user_main    D[KERNEL]: HLE\sceKernelThread.cpp:2189 sceKernelDeleteThread(278)
41:44:516 user_main    D[KERNEL]: HLE\sceKernelThread.cpp:444 Freeing thread stack simple

And then it fails to allocate:

41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:391 -----------
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:395 Block: 08800000 - 08804000 size 00004000 taken=1 tag=usersystemlib
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:395 Block: 08804000 - 08a71100 size 0026d100 taken=1 tag=ELF/simple
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:395 Block: 08a71100 - 08a72000 size 00000f00 taken=0 tag=(untitled)
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:395 Block: 08a72000 - 08a8b000 size 00019000 taken=1 tag=UserSbrk
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:395 Block: 08a8b000 - 09f7f000 size 014f4000 taken=0 tag=(untitled)
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:395 Block: 09f7f000 - 09fbf000 size 00040000 taken=1 tag=stack/user_main
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:395 Block: 09fbf000 - 09fc0000 size 00001000 taken=1 tag=stack/main_thread
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:395 Block: 09fc0000 - 0a000000 size 00040000 taken=0 tag=(untitled)
41:44:517 user_main    I[HLE]: Util\BlockAllocator.cpp:397 -----------
41:44:517 user_main    E[HLE]: Util\BlockAllocator.cpp:139 Block Allocator (08800000-0a000000) failed to allocate 21991424 (014f9000) bytes of contiguous memory

Notice that 014f9000 fails because only 014f4000 is available. It's actually missing 0x5000. If simple had not been using the top slot, user_main would go there instead. And then there would be more than enough space (0x40000 more.)

But, maybe that's actually right, and it's trying to do something else. There's a bunch of kernel syscalls earlier that aren't really supported, and they could even be doing things to fix this situation.

-[Unknown]

[sum2012]

After #12986 , the game should improve,
@daniel229 can you re-test ?

[sum2012]

I test v1.9.3-981-g288fe3ade which merged #12986
It fixed
Closing this issue