VIF recompilers: full size
EE/MVU/IOP: 1/4th of the size
Lazy allocation is still enabled but it will less triggered.
Next step is to always commit the first block
Let's the kernel manage the memory either with builtin lazy allocation or
swapped memory.
Avoid to handle SIGSEGV manually (nicer for debug) and removes 250 lines of code.
The window is redrawn when Layout() is used (I think), so the text is
repainted with the correct colour.
This fixes the gray/ungray behaviour in the Emulation Settings dialog.
(Specifically, it affected the GS, Speedhacks, and Game Fixes Panel).
From the comment, it used to be bad to gcc alignment issue.
Recent gcc align the stack on 16B boundary. Maybe it solved the issue.
I play severals minutes without any crashes but it requires more tests.
The current redirection code doesn't work on VS2015:
- It relies on undefined behaviour (*stdout/*stderr = *fp)
- stdin/stdout/stderr are not pre-opened when compiled with the v140*
toolkits
- It seems to be the reason PCSX2 fails to terminate properly.
Use a combination of named pipes and freopen to redirect stdout/stderr.
Note: The redirect stuff doesn't seem to work in debug builds, both
before and after the change.
CID 147025 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)8. uninit_member: Non-static class member m_current_count is not initialized in this constructor nor in any functions that it calls.
CID 147005 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)2. uninit_member: Non-static class member PrePrefix is not initialized in this constructor nor in any functions that it calls.
CID 146995 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)member_not_init_in_gen_ctor: The compiler-generated constructor for this class does not initialize cpu
CID 146990 (#1 of 1): Uninitialized scalar field (UNINIT_CTOR)6. uninit_member: Non-static class member PluginId is not initialized in this constructor nor in any functions that it calls.
CID 146984 (#1 of 1): Uninitialized scalar field (UNINIT_CTOR)6. uninit_member: Non-static class member m_performFileWrites is not initialized in this constructor nor in any functions that it calls.
CID 146980 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)2. uninit_member: Non-static class member m_panel is not initialized in this constructor nor in any functions that it calls.
CID 146979 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)2. uninit_member: Non-static class member m_miscGroup is not initialized in this constructor nor in any functions that it calls.
CID 146978 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)4. uninit_member: Non-static class member s_rightside_buttons is not initialized in this constructor nor in any functions that it calls.
CID 146977 (#1 of 1): Uninitialized scalar field (UNINIT_CTOR)6. uninit_member: Non-static class member SizeInMB is not initialized in this constructor nor in any functions that it calls.
-Fixes other Chessmaster hang.
-having it say it has data really quickly if the data incoming was huge seemed to throw it off. I suspect it may actually be the time it takes to fill 16qw (or whatever is incoming if lower) is the wait time but it seems rather happy with the incoming data size.
On linux, it breaks the 16B stack alignment requirement.
2 dispatchers were added to handle the call to the function. It avoid
any performance impact and remove the extra inlined asm
Fix#506
It changes nothing but it is more consistent
CID 146916 (#1 of 1): Arguments in wrong order i
(SWAPPED_ARGUMENTS)swapped_arguments: The positions of arguments in the call to disBranch do not match the ordering of the parameters:
rt is passed to rs
rs is passed to rt
Bind doesn't work on wx2.8. Replace with event tables and Connect.
wxString::Format also requires the format string to be L"something".
const char* requires an explicit conversion to wxString.
The show 128/64/32 bit menu column stuff doesn't have a direct wx2.8
replacement so just omit it - I'm not sure it's useful functionality
anyway.
-Coverity CID 152835 & 152834: The expression's value does not depend on the operands; often, this represents an inadvertent logic error.
In psxRcntWcount32(int, unsigned int): An operation with non-constant operands that computes a result with constant value (CWE-569)
CID 146889 (#1 of 1): Dereference null return value (NULL_RETURNS)4. dereference: Dereferencing a pointer that might be null wxGetApp()->GetDisassemblyPtr() when calling update
FindWindowById returns NULL if the Window with the given id is not found, no need to do a extra check for validation of the id. also do a check for wxGetApp()->GetDisassemblyPtr() to prevent a null deference.
-Fix up Vsync again, broken back in google code days, broke Fatal Fury, matches PS2 test again.
-Corrected the Vsync Gate 0, which was the wrong way around, Causing Legendz Gekitou not to work unless Vsync was wrong
Coverity seems to not like the assert trick and only considers the macro as a single statement, so let's rework branchAddr and branchAddrN to satisfy coverity and improve the code quality. The following patch fixes 8 coverity issues with the same problem , CID 151736 - 151743.
(#1 of 1): Misused comma operator (NO_EFFECT)extra_comma: Part !!((mVU.prog.IRinfo.curPC & 1U) == 0U) of statement (!!((mVU.prog.IRinfo.curPC & 1U) == 0U)) , ((mVU.prog.IRinfo.curPC + 2U + (s32)((mVU.code & 0x400U) ? 0xfffffc00U | (mVU.code & 0x3ffU) : (mVU.code & 0x3ffU)) * 2 & mVU.progMemMask) * 4U) has no effect due to the comma.
CID 146848 (#1 of 1): Using invalid iterator (INVALIDATE_ITERATOR)
8. use_iterator: Using invalid iterator existing.
Directly edit the value through the iterator.
There's only one plugin api for this, but GSdx also checks whether shift/ctrl
are held down, so PCSX2 needs to map those too to the same API.
Make this more systematic by only mapping one shortcut to this API, and then
deriving the other two from it amd mapping them too automatically.
This also makes it possible to override it at PCSX2_keys.ini since now it
doesn't need to handle different shortcuts for the same function (which it still
can't handle, but now it also doesn't need to for this function).
Apparently I logged the values incorrectly previously. I said
that shift-q gets parsed as Q without shift, but actually the shift flag
is set correctly on such case (though the letter is still upper case for shift-q
and it's still lower case for plain q)
So I retested all the 8 modifiers combination, and noticed that the modifiers
are always parsed and set correctly, but the letter ends up upper case if any
modifiers are used, but lower case if no modifiers are used.
We still need to make lower if it's upper, but don't need to add the shift flag.
This patch fixes 3 different issues for wx event key codes and the ini parser.
- The ini parser (for PCSX2_keys.ini) parses symbols and letters correctly,
but parsed `shift-q` as `Q` without shift. Now it's parsed as `q` with shift.
- the wx event keycode has the shift flag set correctly, but always has the
upper case code for letters (e.g. `Q` when `q` was pressed, and `Q` with shift
flag when shift-q is pressed). Now uses lower case for letters.
- For symbol-only keys (e.g. `-` or `=`), the keycode is the MS VK_.. thingy,
e.g. for `=` it passed 187 instead of 61, or for `.` it had 190 instead of 46.
Now returns the ascii code where possible (assuming US KB layout).
Also, when the DevCon is enabled, all key presses are printed to the console in
a format which could be copied to pcsx2_keys.ini, so this should allow also
non-US KBs to be set up for arbitrary symbols.
It should now be possible to use letters and symbols at PCSX2_keys.ini or as
default accelerators definitions within the code.
Note: this should have supposedly been fixed at r4918 ( 851bfba ), but neither
that build from the buildbot archive, nor the official 1.2.1 release (r5875),
nor the current builds have this working, which is really weird (tested on win8)
This allows to choose which items appear at the title and at which order, as
well as the text and some of the textual values (at PCSX2_ui.ini).
The template's variables (include % if relevant but not the surrounding text):
${slot}, ${limiter}, ${speed}, ${vfps}, ${cpuusage}, ${omodef}, ${omodei}, ${gsdx}
The system treats the GSdx info as a single unit, as well as the CPU usage info.
The UI section (at the CPU usage section) is only visible in devel/debug builds.
The current template values which also demonstrate all the configurable values:
[UiTemplates]
LimiterUnlimited=-unlimited
LimiterTurbo=-turbo
LimiterSlowmo=-slowmo
LimiterNormal=
OutputFrame=frame
OutputField=field
OutputProgressive=p
OutputInterlaced=i
TitleTemplate=Slot: ${slot} | Speed${limiter}: ${speed} (${vfps}) | ${cpuusage} | ${omodef}-${omodei} | ${gsdx}
The previous longer template values:
[UiTemplates]
LimiterUnlimited=None
LimiterTurbo=Turbo
LimiterSlowmo=Slomo
LimiterNormal=Normal
OutputFrame=Frame
OutputField=Field
OutputProgressive=Progressive
OutputInterlaced=Interlaced
TitleTemplate=${gsdx} | Limiter: ${limiter} | ${omodei} (${omodef}) | Speed: ${speed} (${vfps}) | ${cpuusage} | State: ${slot}
wxDataObjectSimple derived classes do not use a wxDataFormatId
parameter (only the wxDF_INVALID, wxDF_TEXT, wxDF_BITMAP, wxDF_METAFILE,
wxDF_FILENAME, and wxDF_HTML wxDataFormatIds are allowed because they
are implicitly converted). So drag and drop probably should have been
broken on Windows, but wasn't for whatever reason.
Let's fix and simplify the code by just using the built in
wxCustomDataObject. Also make move the default drag behaviour on both
Windows and Linux (previously default on Windows only).
It now works on Windows and Linux wx2.8/wx3.0 (previously only working
on Windows).
The goal here is to reduce clutter and to make it more useful by putting the
important values at the begining, and the less used values at the end. The GSdx
string is unmodified, and maybe we could simplify it too in the future.
Previously, the tile displayed the following items in this order:
<GSdx string> - (e.g. GSdx OGL HW | 514x224 | None)
<output mode> - Progressive/Interlaced (Frame/Field)
<limiter mode> - Limiter: None/Turbo/Slomo/Normal
<speed info> - Speed: <100-based-percentage> (<vfps>)
<CPU usage> - EE: <%> | GS: <%> | VU <%> | UI: <%>
<save state> - State: <slot-number>
This patch changes the following to this order and format:
- Move the save state value to the front since it's actually useful.
- The speed value, which now includes the limiter value if not "Normal".
- Then the CPU usage parts, but the UI value only shows on debug/devel builds.
- The output mode, shortened to Frame/Field-i/p
- the GSdx string same as before only at the end
So all the info is still there as before (except UI), but more readable format.
E.g. before (normal limiter and turbo limiter):
GSdx D3D11 HW | 640x512 | None | Interlaced (frame) | Limiter: Normal | Speed: 100% (50.01) | EE: 43% | GS: 4% | UI: 0% | State 0
GSdx D3D11 HW | 640x512 | None | Interlaced (frame) | Limiter: Turbo | Speed: 443% (221.31) | EE: 99% | GS: 16% | UI: 0% | State 0
E.g. after (normal limiter and turbo limiter) for the same values as above:
State 0 | Speed: 100% (50.01) | EE: 43% | GS: 4% | frame-i | GSdx D3D11 HW | 640x512 | None
State 0 | Speed (Turbo): 443% (221.31) | EE: 99% | GS: 16% | frame-i | GSdx D3D11 HW | 640x512 | None
Make sure we exit fullscreen only if the GS window is visible and in fullscreen.
While we don't currently have such cases, it's possible that the Sys_Suspend
command is called while it's already suspended (and the GS window is hidden)
but the code without this patch would have incorrectly made it visible again.
This tries to emulate LilyPad's "Safe fullscreen exit on escape" hack, and
hopefully also makes this LilyPad hack redundant, such that it would fix the
issue also with other input plugins.
The issue that fullscreen was not restored if the --noguiprompt confirmation
was aborted was already fixed implicitly via e4d8af0, therefore the comment
ended up incorrect. Also explain why it was broken when emulation was not
suspended before this commit, and how it could have been fixed.
When using --nogui and --noguiprompt and the user presses ESC, a prompt is
displayed asking the user to confirm exiting/closing PCSX2. This patch suspends
the emulation while this prompt is displayed.
Also updated the comment that if LilyPad has "safe fullscreen exit on ESC", then
full screen is not restored automatically if this prompt is canceled.
Specifying a minimum size for the filename text controls seems to mess
up the size calculations at higher DPIs and causes usability issues.
Use sizers and proportions instead.
Also scale the dialog widths with the DPI.
Retry and abort don't have default labels, so you end up with blank
buttons on both Windows and Linux. Fix it.
Retry is used by the portable mode settings dialog. I'm not sure if
abort is used anywhere.
A very slight refactor of commit 9eadf3 to also handle a future case where the
audio has a dialog but the GS doesn't (currently both conditions are false).
The dialog was already modal for the GS window, but the main window wasn't
blocked and pcsx2 could crash if, e.g. the user tried to close the main window
while the recording dialog was visible.
Fixes#879
CID 146870 (#1 of 1): Negative array index write (NEGATIVE_RETURNS)
5. negative_returns: Using variable xmmreg as an index to array ...
Open discussion: how to handle correctly bad register allocation?
Currently negative index is returned and a message printed. It means
we need to propagate the index check everywhere in order to not use it.
I suspect that Instruction Generation is more or less corrupted so
potentially we could just fire an exception.
There were bugs in the UI handling that allowed you to resume emulation
when the memory card dialog was opened, which could potentially cause
data loss/corruption.
Make the memory card dialog modal to prevent this. Although it's
possible to do with a modeless dialog, the solution will be much more
complicated and less future proof.
This also fixes the emulation resume bug after closing the memory card
dialog.
-This was actually a bug, may improve some games that were buggy in superVU, but these functions aren't often used.
-Coverity CID 146865 & 146864: In recVUMI_ESIN(VURegs *, int): Missing break statement between cases in switch statement (CWE-484)
-Coverity CID 146863 & 146862: In recVUMI_EEXP(VURegs *, int): Missing break statement between cases in switch statement (CWE-484)
-Coverity CID 146855 & 146854: In recVUMI_EATAN(VURegs *, int): Missing break statement between cases in switch statement (CWE-484)
-Left in so future checks can see what's going on.
-Coverity CID 146818: In ISOreadSector(unsigned char *, unsigned int, int): Code can never be reached because of a logical contradiction (CWE-561)
-Coverity CID 146868: In R5900::Dynarec::OpcodeImpl::recWritebackHILOMMX(int, int, int, int): Negative value used to index an array in a write operation (CWE-129)
-Coverity CID 146817: In R5900::Interpreter::OpcodeImpl::MMI::PMFHL(): Code can never be reached because of a logical contradiction (CWE-561)
-This code is used by both the Interpreter and Recompiler, however it was probably never checked because nothing much uses it. Out of 248 games, it was called 0 times.
-Coverity CID 146821: In _mVUflagPass(microVU &, unsigned int, unsigned int, unsigned int, std::vector<unsigned int, std::allocator<unsigned int>> &): Code can never be reached because of a logical contradiction (CWE-561)
-Coverity CID 146822: In mVUcompileSingleInstruction(microVU &, unsigned int, unsigned int, microFlagCycles &): Code can never be reached because of a logical contradiction (CWE-561)
CID 146904 (#1 of 1): Out-of-bounds write (OVERRUN)
10. overrun-local: Overrunning array VU->ialu of 8 16-byte elements at element index 8 (byte offset 128) using index i (which evaluates to 8).
CID 146903 (#1 of 1): Out-of-bounds write (OVERRUN)
10. overrun-local: Overrunning array VU->fmac of 8 32-byte elements at element index 8 (byte offset 256) using index i (which evaluates to 8).
Coverity:
CID 146846 (#1 of 1): Dereference after null check (FORWARD_NULL)
5. var_deref_model: Passing null pointer fp to fclose, which dereferences it
Without this patch, if a user initiates a recording and then cancels at the GSdx
dialog, the audio was recording anyway, which is probably highly unexpected.
However, while probably highly unexpected, it could still be useful to record
only audio, but with this patch it's now impossible.
We can reconsider if it turns out that people are actually using this "feature",
though one might as well set the video setting to be very unobtrusive (very low
resolution/bitrate) such that it uses very little CPU.
In order to flush data with some delay after save-to-memory-card, the code was
calling sioNextFrame() on each frames (60 times/s) continuously as long as the
emulation is active, regardless if FMC is in use or if memory cards in general
are in use.
Further to that, each call to sioNextFrame calls 8 times into the memory card
plugin even if FMC is not in use, resulting in overall ~500 calls/sec with the
sole goal of facilitating delayed flush after MCD save with folder memory cards.
This patch is a hack to call it once per second instead of 60 per second, but we
need to remove this call completely.
The delayed flush will now happen after 60-120 frames instead of exactly 60.
I'm not sure if it breaks the state of the folderMemoryCard (FMC) or not, but
at least it doesn't hang PCSX2.
Might need followup to make sure the state ends up valid.
When a custom PAL/NTSC frame rate is used, PCSX2 respects it for PAL and NTSC,
but disrespected it for NTSC-progressive mode (used hardcoded 30/60). Fixed.
Also, when a custom rate was used, the console message displayed half the rate
which was configured. Now print the actual rate.
The FPS has been Indicated in Paranthesis whereas Speed would provide the Integer percentage value based on the FPS limit for NTSC, PAL games. (59.94, 50)
Thanks to avih for the suggestions.
Out-of-bounds memory is no longer accessed if the realloc size is larger.
If reallocation fails, the old memory will not be freed and a memcpy
will not take place.
This should match the Windows _aligned_realloc behaviour, except that an
extra parameter is used.
Both fields are 10 bits but the ~ operation apply on the 32 bits.
Therefore msb bits were always '1' and the condition always false.
Code is only used on the interpreter but it could explain some hangs
CID 146917 (#1 of 1): Untrusted array index read (TAINTED_SCALAR)
tainted_data: Using tainted variable cdvd.Result[1] as an index into an array mg_zones
CID 146912 (#1 of 1): Sizeof not portable (SIZEOF_MISMATCH)
suspicious_sizeof: Passing argument recConstBuf of type u32 * and argument 131072U /* RECCONSTBUF_SIZE * sizeof (recConstBuf) */ to function memset is suspicious. In this case, sizeof (u32 *) is equal to sizeof (u32), but this is not a portable assumption.
CID 146728 (#1 of 1): Buffer not null terminated (BUFFER_SIZE_WARNING)
buffer_size_warning: Calling strncpy with a maximum size argument of 128 bytes on destination array mod.name of size 128 bytes might leave the destination string unterminated
There are a few odd things with Windows and the background logo.
1. The windows containing the logo is smaller than the logo.
2. The fitting algorithm doesn't work at 200% DPI.
Both make no sense, but can apparently be worked around.
This should reproduce the old non-DPI aware pre-wxWidgets 3.0 behaviour
for the main dialogs, except the font rendering should be better.
And add comments that I should have already added but didn't.
The wxImage Scale/Rescale methods with wxIMAGE_QUALITY_HIGH upsamples
and downsamples using the bicubic and box resampling methods
respectively, so let's just use that instead of the ResampleBox method,
which is undocumentated and probably shouldn't be used.
The Rescale function also modified the image when it was not supposed
to. Fix it, and rename it scale to avoid confusion with the wxWidgets
class.
Have listbook icons (Emulation Settings, Plugin BIOS selector) on the left for all platforms.
Don't set minimum widths for Framelimiter and Frameskip panels.
On Windows, this fixes the minimise/restore weird dialog bug on the
Emulation Settings and Plugin/BIOS selector dialogs. It also fixes a bug
at high DPI, where it becomes difficult to click on GUI elements at the
top of the dialog.
It also makes it easier for me to work on the Windows DPI stuff, I don't
have a 4k monitor.
This is the same problem that the Plugin Open and BIOS load exceptions
had - The message box pops up, intercepts the hide event for GSFrame and
leaves you with no alternative but to terminate PCSX2 via the Task
Manager.
The same solution applies - Close the GSFrame first.
Expand "Disable Framelimiting" and "Synchronised MTGS". Fixes message
truncation on Linux, and Windows at higher DPI.
Don't use pxFitDigits on Spin Controls - the sizing algorithm fails
horribly on GTK3, which means you can't see anything.
And don't use a row of spacers - it causes a minor display issue with
GTK3.
Fix Windows Lilypad local volume control GUI checkbox.
Make MicroVU E-bit delay slot warning a Dev warning.
Don't minimise GS Window if main window is minimised.
previously the GS window (rendering window) also gets minimized with the Main window (GUI) whenever it gets minimized. many users didn't like this behavior, hence make it so that GS window doesn't get minimized along with the main one.
The parent window pointer parameter is no longer needed, since the parent parameter at WX Frame has been replaced with Null which doesn't force minimize it due to no parent relationship.
The following E-bit delay slot warning initially used a console error message which has a sort of bright red which would cause users to provide a lot of attention towards to it, whereas it isn't much to be bothered since it doesn't cause any issues on the game according to refraction it was already dealt with on the code.
So, use a proper warning indicator instead of a error log message.
The PCSX2 log system isn't in place yet, so on Windows it's possible for
a language error message to pop up, which might cause alarm. Suppress
the error - it's harmless and it's done everywhere else for locales
(though perhaps the code should be reworked to avoid this).
No official game stores files there. You can easily check this by putting any file in the root dir and loading the PS2 BIOS, which will show it as corrupted data with an unknown filesize.
For the install version on Linux, PCSX2 now searches for cheats in $HOME
instead of /usr. For the install version on Windows, PCSX2 now searches
for cheats in Documents/PCSX2 instead of the cheats directory where
PCSX2 is installed.
Also make a few small consistency changes.
wxWidgets 3.0 seems to have a different default device context alpha
channel handling on Windows, which makes the screenshot function
unusable.
As a incomplete fix, remove the alpha channel information from the final
image on Windows. Screenshots degrade slightly.
The "Other Hacks" and "microVU Hacks" checkbox descriptions are
truncated in both Windows and Linux. Use sizer flags so the descriptions
show up properly.
The custom scroll event handler increments/decrements the slider value
by one when a click event is detected. It also propagates the event to
the wxWidgets3.0 built in slider scroll event handler, which repeats the
same action. This causes the slider to add/subtract 2 whenever it is
clicked on Windows.
Remove the custom function.
The code was restoring the defaults with hardcoded values. This patch
restores the values however they're defined as defaults for AppConfig.
The code still uses hardcode values to set the highlights (bold) of the
default radio button text - using SetDefaultItem.
Note that other than these two panels, the speedhacks panel is the only other
which has a restore-defaults button, and it already does so programatically.
It's probably not worth trying to unify these three restore-defaults button
into a single system.
Debian has a goal to make reproducible builds therefore make it an
option instead of distro specific.
.
I added an "OR openSusE" to not "break" the old openSUSE behavior but ideally
they should just use -DDISABLE_BUILD_DATE=TRUE instead.
.
The old -DopenSUSE is not used for anything else so I removed it.
Credit for this fix goes to @poodle111
Commit 760de9915b broke the detection of
Ctrl and Shift key presses since it misses the VK_CONTROL and VK_SHIFT
virtual keys.
If pcsx2 is compiled for 32-bits and then compiled for 64-bits without
cleaning the build files, the wrong wx3.0 manifest will be included. The
same will happen vice versa.
Place the generated rcdefs.h file in a platform specific directory so
the correct manifest will always be used.
On wxWidgets3.0, using GetSymbol() on a nonexistent symbol causes error
messages to be logged. This causes all plugins except GSdx to output
error messages when an attempt to load the PS2EsetEmuVersion symbol fails.
Suppress the error message - it'll probably just cause unnecessary
confusion.
This means that we are now no longer touching files that haven't technically been written to. Some games use timestamp information to automatically highlight the save that was last written to, so this should fix a small but annoying bug where it would highlight the wrong one.
Do note that while there is a much simpler check that looks like this:
// Remove (== don't flush) all memory card pages that haven't actually changed.
for ( auto oldIt = m_oldDataCache.begin(); oldIt != m_oldDataCache.end(); ++oldIt ) {
auto newIt = m_cache.find( oldIt->first );
assert( newIt != m_cache.end() ); // if this isn't true something broke somewhere, the two maps should always contain the same pages
if ( memcmp( &oldIt->second.raw[0], &newIt->second.raw[0], PageSize ) == 0 ) {
m_cache.erase( newIt );
}
}
m_oldDataCache.clear();
It can fail in edge cases that don't actually seem too unlikely. Imagine a save being deleted, and then a new save from the same game but in a different slot being created quickly afterwards. It seems quite possible that the new save's file data then occupies the exact same pages as the old save's, and since it's from the same game it might be close enough to where a page sized section (juse 0x200 bytes!) matches the data from the old save that previously resided in that location -- which would cause this code to throw away and not flush this data! It's a shame too, since this variant would be a few ms faster as well, but I feel it's better to be safe than sorry here.
This mainly means that the superblock is now no longer written every single time the memory card is closed, but only when it's changed (which should be exactly once, when the memory card is formatted). It also means that you can format a memory card and then have the emulator crash later without having to reformat the card next time.
We're not actually deleting files though, we just rename them to prepend _pcsx2_deleted_, in case something breaks or whatever, so the user can in an emergency just restore the save by removing that part of the filename.
This eliminates prompts at the start of a game complaining about no
memory card being inserted.
I'm honestly not entirely sure if this is safe (is there some memory
card driver that could cache results between different executables?) but
if it isn't we'll see it soon enough!
* Reduce the amount of times path strings are constructed.
* Move file metadata writing to the file helper, which means it will only be written once per consecutive file access instead of on every file chunk.
This allows us to skip a bunch of accesses trying to find a matching file or memory location, presumably without actual consequences. This isn't really gonna change much in actual game use, but does speed up conversions of FileMemoryCards.
It was possible for an invalid (because never written to, so filled with 0xFF) file entry to be recognized as valid in GetFileEntryPointer(), which cascaded up to it flushing file data as fileEntryDict data and thus losing the relevant file data page.
Not sure if the other two entry accesses changed here are affected as well but better be safe than sorry, I suppose.
If you inserted a PSX memory card into a slot, then swapped that with a
PS2 one, the memory card manager still displayed that card as a PSX one
with "MBit" instead of "MiB" as the size unit. Fixed.
This reduces memory card initialization time.
Without this, it always indexes all files when the emulator boots, which
can take a few seconds if you have lots of files.
With this, the only file indexed is the PS2 "your system configuration"
file.
This has the side-effect of not being able to see save files in the PS2
BIOS unless you insert a game disc matching the files you want to see. I
don't think this is a big problem, but there should probably be a "don't
filter" option somewhere in case you want to manage files in the BIOS.
Essentially, I'm telling the memory card to re-index itself with a
filter based on the game's disc serial every time a new executable boots
in the emulator.
This currently works for a lot of games, but fails in edge cases where
the game disc's serial does not match the game serial that is written to
the memory card as part of the save file's directory name. This affects
mostly (only?) games that have multiple discs. We could circumvent this
by adding a "save game serial" or something into the GameDatabase which
tells us what we should filter by for those cases.
Apart from this edge case, this appears to work surprisingly well. Try
it and see if you can find other issues!
FileMemoryCard: Log reads and writes so I know what kind of commands I have to deal with.
FolderMemoryCard: Create basic class/method outline based on FileMemoryCard.
FolderMemoryCard: Add a FolderMemoryCardAggregator so I don't have to write every method in a way that has to handle more than one memory ca
Also shuffle around the location of code because C++ cares about stuff
needing to be defined before they're usable.
FolderMemoryCard: Implement Open().
FolderMemoryCard: Implement GetSizeInfo().
FolderMemoryCard: Implement some basic structure of Read()
FolderMemoryCard: Implement parts of Read() and Save().
Shouldn't it be Write() or Load()? Anyway, this doesn't work yet, but it
gets part of the formatting procedure done which is at least something!
FolderMemoryCard: Add method to calculate ECC.
FolderMemoryCard: Start implementing the FAT.
MemoryCard: More logging.
FolderMemoryCard: Formatting works now!
Formatted memory card isn't actually recognized as formatted yet because I don't store folder metadata yet, but we're getting there!
FolderMemoryCard: Recognize when it's trying to access a data cluster.
FolderMemoryCard: Add directory/file entry support.
On further inspection this might not a be a good way to handle erasing.
FolderMemoryCard: Method to get a file entry and file path from a file's data cluster.
FolderMemoryCard: wxDirName is garbage, let's just use wxFileName for the folder too...
FolderMemoryCard: Fix Erase method.
FolderMemoryCard: Start implementing file writes.
This is still quite broken but we're getting somewhere here!
FolderMemoryCard: Load the data from the host file system into the memory card on emulation start.
Also store superblock to host file system on end.
FolderMemoryCard: Fix a few warnings.
FolderMemoryCard: Implement file reads.
FolderMemoryCard: Proper ECC reads.
FolderMemoryCard: Reads to unmapped locations should return all 0xFF.
FolderMemoryCard: Some sort of working WriteToFile.
(Note: Doesn't always work depending on what order data gets written...)
FolderMemoryCard: Forgot a 'b' for reading files in binary mode. Whoops.
FolderMemoryCard: Load timestamps from the host filesystem.
FolderMemoryCard: r+b needs the file to exist so create if it doesn't.
FolderMemoryCard: Failsafe to allow non-sequential writes.
FolderMemoryCard: Use a cache for writes. Does not flush to host FS yet!
FolderMemoryCard: Flush the data written to the cache to the host file system on exit.
FolderMemoryCard: Since we have a cache now, remove code related to formatting, it's no longer needed.
FolderMemoryCard: More binary file mode mistakes...
FolderMemoryCard: Make it actually possible to disable/eject cards.
FileMemoryCard: Revert changes made for logging data.
FolderMemoryCard: Remove excessive logging.
MemoryCard: Note that the superblock struct is no longer unused.
FolderMemoryCard: A disabled card shouldn't try writing data on exit.
FolderMemoryCard: Log when flushing data.
FolderMemoryCard: Replace plain constants with const variables.
Should make it easier in the future to change the memory card size, if
needed.
FolderMemoryCard: Sort of handle the case when the total size of files in the memory card folder exceed the size of the card.
Not elegant but prevents ugly errors. The file that caused the card to
"overflow" will be seen as corrupted data by the PS2 browser.
FolderMemoryCard: Some sanity checks.
FolderMemoryCard: superBlock member really should have that m_ too to be consistent.
MemoryCard: Switch back to FileMemoryCard for merging.
FolderMemoryCard: Implement GetCRC() via a timestamp of the last memory card write.
Reasoning:
Regarding auto-ejecting on save load, I see that the current
implementation checks that by comparing memory card CRC and reinserting
if it mismatches. Since it's actually just about seeing if the memory
card state of the savestate and the current state match, my GetCRC() now
returns a millisecond timestamp of the last time the card was written
to. This should work out to the intended result, though I had to use
wxGetLocalTimeMillis() instead of wxGetUTCTimeMillis() since the latter
isn't available for some reason.
Fix GCC warnings and error.
MemoryCard: Switch implementations via a #define.
FolderMemoryCard: Add a NextFrame() method that should be called once per frame. Flushes written data to the host file system after a certain amout of frames have passed without any writes (currently 60).
MemoryCard: Add the NextFrame() method to the plugin API.
Counters: If the FolderMemoryCard is selected, inform it every frame in VSyncEnd() that a frame has passed.
VSyncEnd: Probably better to inform the memory card before the frame limiting.
Fix error when using wxWidgets >= 3.0.
FolderMemoryCard: Extract into its own .h/.cpp files.
FolderMemoryCard: Change cache to a map to reduce memory usage.
FolderMemoryCard: More gracefully handle lack of space when adding files.
When the GS plugin is suspended and resumed, PCSX2 will set GS frame
title using an uninitialised array, making it prone to crashing in
those situations. Initialise the array.
This prevents a PCSX2 lock up that requires Task Manager to kill. It
seems the messagebox intercepts the GSframe's close event, hides itself
and then leaves you with no way of ending PCSX2 normally. So close the
GS frame before the message box appears.
If the user encounters a plugin/BIOS error and the plugin/BIOS dialog is
already open, make the dialog switch to the panel that corresponds with
the error.
Also, fix the open dialog detection and provide a slightly different
messagebox if the plugin/BIOS dialog is already open - for these cases it
won't say "Click Ok to go to the x configuration panel" anymore.
When the Console is closed (except through Misc -> Show Console), PCSX2
does not remember that you closed the Console and will reopen it the next
time. You also have to press Show Console twice if you want to reopen
the Console after closing it.
This is caused by the Console looking for its "parent" main emulator
window, when secretly, they're siblings :)
Make the console properly register when it is closed.
The dialog event handling is a bit messed up. An ok/cancel event sends a
close event, which sends a cancel event and repeats. This would actually
be an infinite loop if wxWidgets didn't detect a loop.
Rework the event handling to avoid the loop and to remember the
positions of modal dialogs as well.
Do not do additional processing on an event after EndModal is called.
This fixes a message box bug on Linux wx3.0 builds. Without this fix,
Release and Devel builds emit cancel signals even if 'Ok' or other
buttons are pressed, and Debug builds trip an assertion with the message
"EndModal called twice or ShowModal not called."
PCSX2 doesn't know whether it is in hardware or software mode. So this
code does not work if the user is using software rendering mode without
F9 toggling.
Don't disable vsync if the user toggles rendering mode.
When booting a game, a missing BIOS (and possibly other BIOS errors)
will crash or lockup PCSX2 (except for a Windows wxWidgets 2.8 build,
I have no idea why). The messagebox that appears is also badly
formatted.
Use the event queue to handle the error instead of dealing it with
immediately and fix the appearance of the messagebox.
This also deletes the BiosSelectorDialog code since it isn't used
anymore and fixes a condition where, on plugin load failure, the
component selector dialog may open up at the wrong page.
wxWidgets 3.0 now calls Output() instead of Printf() for wxMessageOutput
derived classes. Fix the command line help hack so it works for that.
Also use wxString instead of FastFormatUnicode, it's simply not needed
here.
wxCharCodeMSWToWX() has been removed from wxWidgets 3.0 and the
replacement does not have the same behaviour. Add a replacement function
to translate Windows VK_* keycodes to WXK_ keycodes.
On Windows, the languages displayed in the dropdown box of the language
panel selector are not in alphabetical order. This is caused by the
CBS_SORT window style, which is set by wxCB_SORT, having a ridiculous
sorting order. Remove the option to fix the language order and sort a
minor UI glitch.
Gregory Hainaut