Avoid System::GetInstance() and ppcState global in Jit64 and CachedInterpreter. #11677
Conversation
AdmiralCurtiss
force-pushed
the
deglobal-jit
branch
4 times, most recently
from
72eb8b0
to
e67f4e5
Compare
| @@ -9,11 +9,10 @@ | |||
|
|
|||
| // We offset by 0x80 because the range of one byte memory offsets is | |||
| // -0x80..0x7f. | |||
| #define PPCSTATE(x) \ | |||
| MDisp(RPPCSTATE, (int)((char*)&PowerPC::ppcState.x - (char*)&PowerPC::ppcState) - 0x80) | |||
| #define PPCSTATE(state, x) MDisp(RPPCSTATE, (int)((char*)&(state).x - (char*)&(state)) - 0x80) | |||
There was a problem hiding this comment.
Any reason why you can't use something like offsetof here? We don't actually need to know where in memory the struct is when calculating this.
There was a problem hiding this comment.
I don't see why not.
There was a problem hiding this comment.
MSVC has no problems with this, but GCC unfortunately does.
/buildbot/pr-ubu-x64/build/Source/Core/Core/PowerPC/Jit64/Jit_SystemRegisters.cpp:23:29: error: ‘field’ is not a constant expression
23 | return PPCSTATE(cr.fields[field]);
There was a problem hiding this comment.
I've reverted the change again but if you want to experiment it's here: https://github.com/AdmiralCurtiss/dolphin/tree/deglobal-jit-offsetof
There was a problem hiding this comment.
Oh, I see. Come to think of it, we made a special helper for this in JitArm64 for this reason:
There was a problem hiding this comment.
Yeah that should work... lemme try that here too.
There was a problem hiding this comment.
Seems to work.
AdmiralCurtiss
force-pushed
the
deglobal-jit
branch
4 times, most recently
from
df313c2
to
31204ee
Compare
|
Poke? |
AdmiralCurtiss
force-pushed
the
deglobal-jit
branch
from
31204ee
to
0e06364
Compare
AdmiralCurtiss
force-pushed
the
deglobal-jit
branch
from
0e06364
to
fc394bd
Compare
No description provided.