Extra result handling from funcStat CB in savedata_op #5399
Conversation
|
Fixes some games(Ghost of Sparta, maybe others) having no sound on first load because savedata_op would not return an error in this particular case. |
| @@ -610,6 +610,12 @@ static NEVER_INLINE s32 savedata_op(ppu_thread& ppu, u32 operation, u32 version, | |||
| // //return CELL_OK; | |||
| //} | |||
|
|
|||
| if (operation == SAVEDATA_OP_AUTO_LOAD && !fs::is_dir(dir_path)) | |||
There was a problem hiding this comment.
SAVE vs LOAD wasn't supposed to have any effect on function behaviour except visual.
There was a problem hiding this comment.
You're right cellSaveDataAutoLoad2 has the same exact behavior as cellSaveDataAutoSave2.
I'll investigate a bit more.
There was a problem hiding this comment.
After further testing it seems it's the CB returning CELL_SAVEDATA_CBRESULT_OK_LAST_NOCONFIRM that result in an error.
Strangely enough the CB returning CELL_SAVEDATA_CBRESULT_OK_LAST terminates the function with a CELL_OK.
RipleyTom
force-pushed
the
loaderror
branch
2 times, most recently
from
f5dbb1a to
eaee477
Compare
RipleyTom
changed the title
|
Just in case I did extra checking: CELL_SAVEDATA_CBRESULT_OK_LAST_NOCONFIRM = 2 1(aka 6) leads to termination with CELL_OK and otherwise follow the same path. There is a function at seg000:0000CBF0 bl do_stuff_3 that does some extra stuff before quitting but I thing it just cleans the cxml session. |
|
Then this is a fallback for all values >= 2 or less than -5 |
RipleyTom
force-pushed
the
loaderror
branch
2 times, most recently
from
8448148 to
940a9dd
Compare
|
You're right, fixed. |
| if (result->result < CELL_SAVEDATA_CBRESULT_OK_NEXT) | ||
| { | ||
| return CELL_SAVEDATA_ERROR_CBRESULT; | ||
| } | ||
|
|
||
| //Skip and return without error | ||
| if (result->result == CELL_SAVEDATA_CBRESULT_OK_LAST) |
There was a problem hiding this comment.
Need to remove this part (see #4397)
There was a problem hiding this comment.
Hmm that's not correct, I'll take a look at the game but from what I see in disassembly it terminates without calling any other callback.
There was a problem hiding this comment.
It does terminate later, it's already handled.
There was a problem hiding this comment.
Reversed Battle Princess of Arcadias, the launch race condition has nothing to do with save functions, no reason to keep the hack in.
|
This PR fixes #4985 On latest master, I still see the MEM Access Violation as mentioned in the issue. With this PR, I get past the error and comfortably ingame! |
The funcStat callback is called even if the directory doesn't exist.
Its returning statSet->reCreateMode doesn't seem to change the resulting behaviour as per my testing on real HW.
REd library:
CELL_SAVEDATA_CBRESULT_OK_LAST_NOCONFIRM returns CELL_SAVEDATA_ERROR_PARAM
CELL_SAVEDATA_CBRESULT_OK_LAST returns CELL_OK
Both instantly terminate after calling funcStat.