Bizhawk never would hit this because it only ever runs waterboxes in one host thread, but an application that spun up many threads and ran waterboxes in each would leak 32 bytes of heap for each native thread destroyed, which is super duper not really meaningful at all
Waterbox guest code now runs on a stack inside the guest memory space. This removes some potential opportunities for nondeterminism and makes future porting of libco-enabled cores easier.
This replaces the old managed one. The only direct effect of this is to fix some hard to reproduce crashes in bsnes.
In the long run, we'll use this new code to help build more waterbox features.
Cores that used the .invisible section to store data were saving it; this was a regression from before, so PCFX states should be back down to the previous release size, or perhaps a bit smaller.
Add the ability to dirty track libco cothreads, as used in the bsnes core. This saves a lot of space in those states and they're now quite competitive in size.
some bsnes cothreads call callbacks that hit managed threads. We shouldn't do that, but we do, and sometimes those threads run MSVC's __stkchk which can, depending on circumstances, blow up if the thread extents aren't set.
This also means that we cannot save space on a lot of cothread stacks because __stkchck will blow up any detection guards we try
The waterbox system now uses host os facilities to track whether memory has been written to, to automatically choose what thing to savestate. This results in a large size decrease for some cores, like snes9x or gpgx (when running cartridge games). Doesn't do much for cores that were already memory efficient, or for bsnes because of libco compatibility issues; but those cores don't regress either.
and who said waterbox can't thread. well, it sort of can't. but it sort of can.
the speedup isn't that great, but speed is now pretty close (5%?) to snes9x in the only game that matters (final fantasy 5)
For the external cores, I simply listed whatever the release build we have right now built them with. There are other makefiles and such hanging around...
Nyma cores have to move some big complex structs on init and it's annoying and error prone. This solution is not fast, but these are one time transfers anyway, and it does keep code size and saved size down. Architecture yay.
This was benign, because libunwind will ignore a frame header it doesn't understand (__eh_frame was still fine). But now there's no spew in the console. And over the next 50 years it will save a combined 0.3 seconds of cpu time
The goal here is to provide an unwind implementation that works. We (probably) don't have a working ldso, so reconfigure libunwind to work without that. To do that, it needs __eh_frame_* variables which it can addressof to find the locations of .eh_frame and .eh_frame_hdr. There is no way I could get gold to add these at all, so switch back to ld and add a custom linkscript. The custom linkscript gives us the opportunity to simplify the savestate memory mapping stuff inside ElfLoader, which should knock a bit of time off of state loads -- those VirtualProtect calls are not cheap. This also removes a potential source of nondeterminism with certain clever out of range pointers.
Any waterbox core that I didn't recompile for this commit will assert now until it is recompiled, because .wbxsyscall is in the wrong palace.
Trigger condition: Any 32x rom smaller than 4MB in total size that tries to read out of bounds between its actual end and the 4MB limit
I've fixed the emulator crash, but I do not know if such roms are actually mirroring correctly - they probably aren't
Create an all new waterbox build environment:
WSL2 + Ubuntu 20.04 LTS (Other linuxes may work)
Musl libc with waterbox customizations
LLVM's libclang-rt, libunwind, libcxxabi, libcxx
Static linking to elf files
Compared with the old system, this is easier to set up a dev env for and easier to update in the future. The executables are larger but produce smaller savestates due to static linking. The modern toolchain means advanced library features and language features that sometimes appear in some upstream cores will be reusable.
If you tried setting your gains to 100 in the previous patch, it worked sort of, but I decided to incorporate ekeeke/Genesis-Plus-GX@f6f4556 as well to actually fix it completely, and rescale the setting properly so that 1.0 really is neutral gain.
Fixes#1319
gpgx upstream automatically gives 64KiB sram to every game of size 2MB or lower, unless specifically overrided. That sucks and we don't do it. But some games do need it; there's not enough information to autodetect otherwise. So add a flag that we can set in gameDB for it, and set it true for two known games that have problems otherwise.
mapper_32k_w expects to be able to write to rom area, so those roms now need larger savestate space
We could be extra clever and alloc_invisible the space and then restore after loadstate, but this pokeymans garbage doesn't deserve that work
nattthebear