Emulation/ppsspp
1
0
Fork 0
mirror of https://github.com/hrydgard/ppsspp.git synced 2025-04-02 11:01:50 -04:00
Code Issues Projects Releases Packages Wiki Activity

Point ra to the correct place on a new thread.

Browse source 
Turns out some games read the ra, maybe long jmps or something?
This commit is contained in:
Unknown W. Brackets 2016-09-05 08:56:27 -07:00
parent f39c603acb
commit 1e51a03d5a
1 changed files with 11 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
Core/HLE/sceKernelThread.cpp
View file

@ -1828,7 +1828,8 @@ void __KernelResetThread(Thread *t, int lowestPriority)
t->currentMipscallId = 0;
t->pendingMipsCalls.clear();
t->context.r[MIPS_REG_RA] = threadReturnHackAddr; //hack! TODO fix
// This will be overwritten when starting the thread, but let's point it somewhere useful.
t->context.r[MIPS_REG_RA] = threadReturnHackAddr;
// TODO: Not sure if it's reset here, but this makes sense.
t->context.r[MIPS_REG_GP] = t->nt.gpreg;
t->FillStack();
@ -2009,6 +2010,15 @@ int __KernelStartThread(SceUID threadToStartID, int argSize, u32 argBlockPtr, bo
// This could be stack overflow safety, or just stack eaten by the kernel entry func.
sp -= 64;
// At the bottom of those 64 bytes, the return syscall and ra is written.
// Test Drive Unlimited actually depends on it being in the correct place.
WriteSyscall("FakeSysCalls", NID_THREADRETURN, sp);
Memory::Write_U32(MIPS_MAKE_B(-1), sp + 8);
Memory::Write_U32(MIPS_MAKE_NOP(), sp + 12);
// Point ra at our return stub.
startThread->context.r[MIPS_REG_RA] = sp;
// Smaller is better for priority. Only switch if the new thread is better.
if (cur && cur->nt.currentPriority > startThread->nt.currentPriority) {
__KernelChangeReadyState(cur, currentThread, true);