* First crack at ensuring the render thread doesn't touch GPU state while it's being serialized
* Get rid of the semaphore wait
* Add some extra fields into GPU3D's serialization
* Oops, TempVertexBuffer is already serialized
* Move vertex serialization into its own method
* Lock the GPU3D state when rendering on the render thread or serializing it
* Revert "Lock the GPU3D state when rendering on the render thread or serializing it"
This reverts commit 2f49a551c1.
* Add comments that describe the synchronization within GPU3D_Soft
- I need to understand it before I can solve my actual problem
- Now I do
* Revert "Revert "Lock the GPU3D state when rendering on the render thread or serializing it""
This reverts commit 1977566a6d.
* Let's try locking the GPU3D state throughout NDS::RunFrame
- Just to see what happens
* Slim down the lock's scope
* Narrow the lock's scope some more
* Remove the lock entirely
* Try protecting the GPU3D state with just a mutex
- I'll clean this up once I know it works
* Remove a duplicate method definition
* Add a missing `noexcept` specifier
* Remove an unused function
* Cut some non-hardware state from `GPU3D`'s savestate
* Assume that the next frame after loading a savestate won't be identical
* Actually, it _is_ worth it
* Don't serialize the clip matrix
- It's recalculated anyway
* Serialize `RenderPolygonRAM` as an array of indexes
* Clean up some comments
- I liked the dialogue style, but oh well
* Try restarting the render thread instead of using the lock
- Let's see what happens
* Put the lock back
* Fix some polygon and vertex indexes being saved incorrectly
- Taking the difference between two pointers results in the number of elements, not the number of bytes
* Remove `SoftRenderer::StateBusy` since it turns out we don't need it
- The real synchronization was the friends we made along the way
* Reorganize namespaces
- Most types are now moved into the `melonDS` namespace
- Only good chance to do this for a while, since a big refactor is next
* Fix the build
* get this started
* implement DSi RTC commands
* set up RTC clock timer. lay down basic idea of a clock.
* make the date/time registers writable
* move RTC state to its own structure, to make it easier to deal with
* more RTC work
lay base for date/time dialog
* get the bulk of the RTC functionality going
* much simpler design for RTC stuff
* aha, that is what it is
* start working on the RTC IRQ
* implement all types of RTC IRQ
* start refining sleep mode. code still kinda sucks.
* implement keypad IRQ
* refine it some more
* shut the fuck uuuuuupppppppppppppp
* Slight polish to DMA
- Default-initialize members explicitly
- Mark some methods as const noexcept
- Initialize DMA::MRAMBurstTable to DMATiming::MRAMDummy
- Use the default destructor
* Move DMA_Timings definitions to a source file
- To ensure constant and unique addresses
* Include some extra DMA members in the savestate
* Simplify serializing the DMA table
- Extend the dummy table to 256 bytes (same length as the real ones)
* Revert the type change to DMA::DoSavestate
* Keep the MRAMBurstTable inside the DMA class, instead of using a pointer
- If we use a pointer to an external table, then we can't use it in savestates (else that external table gets overwritten)
* Refactor Savestate::Var{8,16,32,64}
- They now delegate to VarArray
- They're declared in the class header so they're likely to be inlined
* First crack at refactoring Savestate to work in-memory
- Well, third, but who's counting?
* Implement Savestate::Finish
* Remove the VersionMajor and VersionMinor fields
- Instead, pull their values directly from the savestate buffer
* Mark a new constructor as explicit
* Rename Reset to Rewind
* Fix a linebreak
* Implement Savestate::Rewind
* Add ROMManager::ClearBackupState
* Refactor ROMManager to use the refactored Savestate
* Capitalize "Least"
- It was driving me nuts
* Add a log call
* Increase default Savestate buffer length to 32MB
* Use C-style file I/O instead of C++-style
- Dumping bytes to a file with C++'s standard library is a MONSTROUS PAIN IN THE ASS
* Quote the savestate's file path for clarity
* Write the savestate's length into the header
* Add some extra logging calls
* Fix section-loading
* Remove the deprecated Savestate constructor
* Convert a char* to a u32 with memcpy, not a cast
* Fix section-handling in loads
* Include <cstring> in Savestate.h
- This was causing a build error on Linux
also including:
* getting rid of shitty strings
* all new, cleaner ROM handling code
* base for DSi savestates
* GBA slot addons (for now, memory cart)
* Implement "lag frame" flag and counter, and expose flag and both frame counters. BizHawk wants these.
* Track frame count and lag frames while the system isn't running.
- savestates used to read a four bytes from a single byte value
- a few unassigned variables
- some other things
- also make the ROR macro an inline function
* bump savestate version up. will break compatibility, but the alternative would be guessing the missing matrix entries somehow, so blarg.
* avoid relocating the savefile if loading a savestate fails.
