cb4e72058a
Report usage if callback "injection."
...
These are generally giving incorrect results, but it will help to know
what games are triggering it.
2014-01-26 18:41:42 -08:00
efa6640807
Report any missing wait type handling funcs.
2014-01-26 17:14:11 -08:00
2c522cda9a
[core] fix some endian bugs
2014-01-26 14:26:27 -08:00
ab49de0331
Eat some cycles when getting the current thread id.
2014-01-19 12:44:55 -08:00
73d6accafc
Eat a few cycles when checking callbacks.
...
Some games really spam this function.
2014-01-19 12:44:54 -08:00
553990d5a0
Eat cycles when messing with the dispatch thread.
2014-01-19 12:44:53 -08:00
1a38f6fa7d
Properly remove terminated thread from ready queue.
...
If we change the priority first, we'll remove from the wrong priority
level, oops. Gotta change afterward.
Should improve Darkstalkers Chronicle: The Chaos Tower.
2014-01-09 22:05:04 -08:00
183b4cb45c
Add a static method for creating PSPPointers.
...
This is still kinda ugly, but the assignment thing was just too ugly, I
couldn't do it anymore.
2013-12-16 23:50:52 -08:00
e73d43c434
Correct sceKernelTlspl func names, etc.
...
This is mostly pedantic but it's nice to have the right names.
2013-12-02 23:40:16 -08:00
7dedf54bc3
Always resched in thread queue rotation.
...
It might be time to reschedule anyway, some games call this in a loop with
an empty queue.
May help Harukanaru Toki No Naka de Hachiyoushou (#4669.)
2013-11-29 08:59:47 -08:00
f9f6e9492d
Reorder vfpu data in saved kernel contexts when loading in a new version.
2013-11-28 13:27:51 +01:00
55500d4bb6
Reorder VFPU registers in memory so that we can flush and reload them in bulk more often.
...
Doesn't actually do that yet, that's for the NEON branch.
2013-11-28 13:27:51 +01:00
41a7157aa6
Eat some cycles when changing thread priority.
...
Some games call this a lot.
2013-11-28 01:01:22 -08:00
21c9577801
Initialize fcr31 correctly (even though unsup.)
2013-11-14 23:44:49 -08:00
26f5922174
Return the correct value for fcr0/fir.
...
This is what the PSP actually returns, it's read only.
2013-11-14 23:39:08 -08:00
baa82e0a9d
Keep syscalls the same in the interpreter.
...
Rather than having different bugs.
2013-11-04 07:59:36 -08:00
b2298a9e86
Merge pull request #4420 from unknownbrackets/savestates
...
Enable rewind feature in UI, fix a savestate bug
2013-11-03 03:21:51 -08:00
d400b1a67b
Keep track of the current thread obj.
...
To reduce overhead/improve perf. Actually, it used to be only the current
thread object, but now it's just a cache with accessors.
2013-11-02 22:12:44 -07:00
738dfc84e4
Speed up __GetCurrentThread(), it's called a lot.
...
This only needs to be an assertion.
2013-11-02 22:12:43 -07:00
f716d781b6
Retain threadend waits in savestates.
...
Oops, been broken since 3fb2c8f654
2013-11-02 17:40:23 -07:00
9ebaf7b68b
Fill registers with 0xDEADBEEF after most syscalls.
...
Some, like sceMpegRingbufferPut(), we don't, since... it's complicated.
But most we do, like a real PSP does.
2013-10-30 00:47:04 -07:00
04ec987448
Initialize thread regs with 0xDEADBEEF.
...
Also initialize the float regs to NaN.
2013-10-29 21:39:24 -07:00
95c68ae1e7
Assert some unlikely buffer overflows.
2013-10-26 18:30:55 -07:00
bf89c01215
When atrac3+ decoding fails, stop decoding silently-ish instead of decoding garbage, spamming the log
2013-10-21 23:10:30 +02:00
f5e0d80ead
Pass arguments to root threads same as regular.
...
Before it would have issues or crash if the args were > 256 bytes.
Fixes Ys 1 & 2 Chronicles.
2013-10-16 01:31:58 -07:00
e06889069c
Don't use the destructor to free thread stacks.
...
If we do, kernelObjects.clear() will trash perfectly good memory on
loading a state. I guess it also makes shutdown slower.
Fixes #4186 .
2013-10-14 22:09:16 -07:00
a194006aad
Fix crash deleting a never started thread.
...
Fixes #4069 .
2013-10-08 22:39:49 -07:00
5be30fb857
Add support for the clear stack thread attr.
2013-10-06 12:16:58 -07:00
b928f50353
Cleanup sceKernelRegisterExitCallback() per tests.
2013-09-29 01:10:07 -07:00
a7ec139b5a
Savestate the exit callback, reporting.
...
Seems like LoadExecForUser_362A956B() uses it, so we need to save it.
I have no confidence that function works correctly, so let's report it.
2013-09-29 00:19:54 -07:00
50e9e45d65
Check version in each DoState() func.
...
They bail on PointerWrap error or bad version.
2013-09-14 20:23:03 -07:00
ca6491ea22
Auto resume dispatch at thread start/create.
2013-09-12 08:02:43 -07:00
b898e397aa
Correct sceKernelGetThreadStackFreeSize().
...
It was always returning 0.
2013-09-10 01:31:03 -07:00
0f7ac377cd
Offset the root thread's stack by arg size.
...
Otherwise it'll just overwrite, right?
2013-09-10 01:12:51 -07:00
ada492febe
Fix really stupid heap corruption mistake.
2013-09-09 22:33:17 -07:00
ccf021cef0
Return a negative error code in __KernelGetWaitID.
...
This way we can use 0 as a proper wait id.
2013-09-09 21:39:56 -07:00
2577f65b80
Reset waits when terminating a thread.
...
Terminated threads can't be waiting.
2013-09-09 21:39:38 -07:00
e19fdef0fd
Reset thread priority on terminate.
2013-09-09 21:31:04 -07:00
a3637a1cad
Set the low mystery bits of the thread attr.
...
Tests on PSP show this, and I can't find a way to not have them set.
But I don't really know what they mean.
Anyway, since you can refer the status, you can see them, may improve
compatibility.
2013-09-09 20:51:52 -07:00
0af3c50924
Eat cycles after creating a thread.
...
Improves things on some tests, should be more accurate.
Creating a thread sure takes a long time.
2013-09-09 20:49:30 -07:00
1481bbdd3b
Validate and report some thread attributes.
2013-09-09 20:48:23 -07:00
ce4c0d85bb
Support some thread stack attr options.
...
May improve memory management in some games.
2013-09-09 20:46:19 -07:00
1040123a97
Delete callbacks when their thread is deleted.
2013-09-09 00:22:00 -07:00
4bba276696
Run callbacks in creation order, not notify order.
2013-09-09 00:21:59 -07:00
87bd8d59e2
Clean up umd bad callback id errors, oops.
2013-09-09 00:21:53 -07:00
64b75ed3ff
Decentralize callback registration.
...
Callbacks are always tied to THEIR thread, and the registration was all
based on the current thread which is just not the right way.
2013-09-09 00:21:15 -07:00
94e88943d8
Keep a single list of pending callbacks.
...
Doing it per type only meant that we'd incorrectly call a callback twice
if it was notified by e.g. scePowerRegisterCallback() and
sceKernelNotifyThread(), which isn't what's supposed to happen.
2013-09-09 00:21:14 -07:00
f1b62983b5
Add callback wait handlers for threadend as well.
2013-09-08 12:14:07 -07:00
6d8213b0c5
Always specify retval when resuming a thread.
2013-09-08 12:02:18 -07:00
4c828cd9a0
Remove __KernelTriggerWait(), no longer used.
2013-09-08 11:58:40 -07:00
Unknown W. Brackets