Wrap some SysMemForKernel's nids, fixing #7960 #7965
Conversation
|
@hrydgard this commit have problem ? |
|
OMG,If I include sceKernelMemory.cpp |
As SysMemForUser in there
|
I move SysMemForKernel to sceKernelMemory |
| std::string name; | ||
| BlockAllocator alloc; | ||
| static int GetStaticIDType() { return SCE_KERNEL_TMID_Fpl; }// wrong | ||
| int GetIDType() const override { return SCE_KERNEL_TMID_Fpl; }// wrong |
There was a problem hiding this comment.
I take a search but no SCE_KERNEL_TMID_HEAP
There was a problem hiding this comment.
partitionId, name,flags,name not used
Not sure really need save status these variable
There was a problem hiding this comment.
We don't use the "XyzInformation" convention in PPSSPP, even if JPCSP does. I really think we should have testing to understand how this differs from sceHeap. Also, testing can identify which TMID is uses - there's a syscall that gives you the type id for a uid, iirc.
I can't remember for sure, but I thought we used the TMID for savestates, though. I don't think it's good if they're wrong.
-[Unknown]
There was a problem hiding this comment.
So this pull request is not ready for v1.1
There was a problem hiding this comment.
Also yes,I copy most of code from sceHeap.cpp
There was a problem hiding this comment.
yeah, it won't make it. we can start adding some kernel stuff after 1.1 but for now I want to keep things the way they are to be safe.
1.1 will be soon though, I'm planning to release next weekend.
There was a problem hiding this comment.
See also PPSSPP_KERNEL_TMID_Module and similar for using fake TMID values. For savestates to work correctly, we need unique values right now.
-[Unknown]
|
Unfortunately,the screen is still black after HLE\sceKernelThread.cpp:2851 sceKernelSuspendThread(276): cannot suspend current thread modify log: |
|
In the soft gpu,the menu appear now. |
|
The crash is from sceKernelDeleteHeap... |
As this commit is work (tested on Tales of Phantasia with Chinese patch)
|
Just a note that my change is based on jpcsp/jpcsp@3c25717 |
| @@ -2301,3 +2301,115 @@ void Register_SysMemUserForUser() | |||
| { | |||
| RegisterModule("SysMemUserForUser", ARRAY_SIZE(SysMemUserForUser), SysMemUserForUser); | |||
| } | |||
|
|
|||
|
|
|||
| struct HeapInformation : public KernelObject { | |||
There was a problem hiding this comment.
I think we should:
- Put these functions and structs in
sceKernelHeap.cpp. - Keep
SysMemForKernelhere and export the functions insceKernelHeap.h(kinda like sceKernelMutex.h, etc.)
-[Unknown]
|
This will need some rebasing and cleanup work but seems worthwhile if it gets Chinese language patches going, with not too much trouble. @sum2012 do you intend to improve this some more anytime soon? Otherwise I might clean it up a bit soon for you, but you'll have to test it. |
hrydgard
changed the title
|
@hrydgard I remember this code not more work , maybe I transate something wrong |
|
Hm. Either way it needs a rebase if we should keep it open. Close or not? |
|
I would rebase it in Sunday in Hong Kong time |
|
Oh, there's absolutely no rush. Whenever you have time. 1.5 is still far away.. |
|
|
|
Closing this in favor of #12225. Sorry for the long delay ;) |
#7960
The log is better
https://gist.github.com/sum2012/2ab67196dc4a97ff60e7