When floats and doubles were converting to integers we weren't doing the correct transformation.
AArch64 provides direct ops for all four of the op types. So lets use them
- f32 -> int64
- f32 -> int32
- f64 -> int64
- f64 -> int32
Doesn't fully fix the case of overflow for AArch64 since overflow behaviour is different.
x86 returns 0x8000'0000 or 0x8000'0000'0000'0000 while AArch64 saturates to the maximum signed
integers. Can't work around that without checking overflow flags.
This does solve the typical case though.
Fixes audio problems in all FMod games.
If the source GPR had data that was larger than the element it would overwrite other elements
Mask it correctly. This then matches behaviour with the other CPU backends
Theres a fair number of these so I won't describe them all.
A couple highlights are MPSADBW and PHMINPOSUW.
These don't really match with AArch64 very well so their IR is a bit ugly.
We need to save the vector registers otherwise we will corrupt them.
Also in the case of printing a vector register, fall down the specialized path
Only useful when debugging
Addresses #146 a little more by providing an interface to perform
fmt-compatible logging.
No more, will people on the project be tormented by classic printf
features like:
- Accidentally passing in a non-trivial type
- PRI macros
- Not being able to add support for custom types
- Mixing up signed/unsigned printf formatting specifiers accidentally
fmt-capable versions of the logging functions are named the same as the
existing functions, just with a "Fmt" or _FMT suffix (depending on
whether or not it's a function being used or a macro, respectively).
e.g. LineDefinitions are moderately beefy, they contain
two strings and a vector of strings among other things,
so we can move instances into their containing vector to avoid some
allocation churn.
Now possible in a less messy way, since the type is now centralized in
one location.
Makes it strongly typed and prevents any potential accidental implicit
assignments to Type instead of something intended for the Data members.
Significantly shortens the amount of code necessary for testing the type
of a decoded operand.
Also relocates the type field out of all the union types to have it in a
central location.
If the guest has sent a signal and the si_code is SI_USER then
we need to pass that siginfo_t through without touching it.
User could be sticking whatever they want in to that struct
Begins the process of addressing issue #146.
This is separated off, so that others can make use of fmt for other
purposes (general localized non-sucky string formatting), while the
logging rework is being tackled.
Overriding glibc's mmap and munmap functions don't encapsulate that functions they use
for allocating stack data so it was falling down the system mmap/munmap path.
This was causing host side stacks to end up in the 32-bit VA space in 32-bit applications.
Ryan Houdek