RPCS3 Forums

(12-18-2014, 10:06 AM)derpf Wrote: [ -> ]Does this not work for anyone else now?

Relevant log (after intro and sometimes at loading screen I believe):

Code:

HLE: W {PPU[703] Thread (thread_pimple created thread)[0x003bb03c]} sys_fs warning: cellFsStat(path="/dev_hdd0/game/NPUB30041/USRDIR/Image/Assets/Spells/SWHI.lua", sb_addr=0xd00be9d0)
HLE: W {PPU[582] Thread (MultiStream PU Thread)[0x0022e5e8]} sys_spu warning: sys_spu_thread_write_spu_mb(id=577, value=0x4b4d80)
HLE: W {PPU[582] Thread (MultiStream PU Thread)[0x0022e624]} sys_spu warning: sys_spu_thread_write_spu_mb(id=577, value=0x4b4d80)
HLE: W {PPU[582] Thread (MultiStream PU Thread)[0x0022e5e8]} sys_spu warning: sys_spu_thread_write_spu_mb(id=577, value=0x4b4d80)
G: E {PPU[1] Thread (main_thread)[0x001cdd88]} Exception: Access violation: addr = 0x1 (is_alive=1, last_syscall=0x0 (Unknown func id: 0x00000000))
RSX: W {VBlank thread} VBlank thread aborted
RSX: W {RSXThread} RSX thread aborted
HLE: W {Audio Thread} cellAudio warning: Audio thread aborted
HLE: W {PPU[579] Thread (Multistream_printf_server)[0x0022e188]} sys_event warning: sys_event_queue_receive(equeue=578) aborted
HLE: W {PPU[659] Thread (_sys_mixerSurBusReq)[0x0023e1ec]} sys_cond warning: sys_cond_wait(id=654) aborted
HLE: W {PPU[651] Thread (_sys_MixerChStripMain)[0x002447e8]} sys_cond warning: sys_cond_wait(id=650) aborted
HLE: W {PPU[651] Thread (_sys_MixerChStripMain)[0x002447e8]} sys_mutex warning: sys_mutex_lock(id=649) aborted
HLE: W {PPU[651] Thread (_sys_MixerChStripMain)[0x002447e8]} sys_cond warning: sys_cond_wait(cond_id=650) failed (EPERM)
HLE: W {PPU[651] Thread (_sys_MixerChStripMain)[0x002447e8]} sys_mutex warning: sys_mutex_lock(id=649) aborted
HLE: W {PPU[651] Thread (_sys_MixerChStripMain)[0x002447e8]} sys_cond warning: sys_cond_wait(cond_id=650) failed (EPERM)
...

The last, of course, being repeated ad infinitum, and before that a segfault.

EDIT: That's weird. Doesn't segfault with interpreter, but when you press start to start the game, it appears to stop all of the threads and what I can only imagine is loading another binary, and then just quits.

Relevant:

Code:

RSX: W GLVertexProgram::Delete(): glDeleteShader(2) avoided
RSX: W GLShaderProgram::Delete(): glDeleteShader(1) avoided
MEM: ! Closing memory...
HLE: W No pause.bin found, Auto Pause will not work.
LDR: ! Loading 'c:/users/bak/desktop/rpcs3/bin/dev_hdd0/game/npub30041/USRDIR/BOOT.BIN'...
LDR: !  
LDR: ! Mount info:
LDR: ! dev_bdvd/ps3_game/ -> c:/users/bak/desktop/rpcs3/bin/dev_hdd0/game/blus30727/
LDR: ! dev_usb000/ -> c:/users/bak/desktop/rpcs3/bin/dev_usb000/
LDR: ! dev_flash/ -> c:/users/bak/desktop/rpcs3/bin/dev_flash/
LDR: ! host_root/ ->
LDR: ! dev_hdd0/ -> c:/users/bak/desktop/rpcs3/bin/dev_hdd0/
LDR: ! dev_hdd1/ -> c:/users/bak/desktop/rpcs3/bin/dev_hdd1/
LDR: ! app_home/ -> c:/users/bak/desktop/rpcs3/bin/dev_hdd0/game/npub30041/usrdir/
LDR: ! dev_bdvd/ -> c:/users/bak/desktop/rpcs3/bin/dev_hdd0/game/npub30041/
LDR: ! dev_usb/ -> c:/users/bak/desktop/rpcs3/bin/dev_usb000/
LDR: !  -> c:/users/bak/desktop/rpcs3/bin/dev_hdd0/game/npub30041/usrdir/
LDR: !  
LDR: ! Title: Puzzle Quest: Challenge of the Warlords - Revenge of the Plague Lord
LDR: ! Serial: NPUB30041
LDR: !  
LDR: E loader::init() failed: Broken file
MEM: ! Initializing memory: base_addr = 0x7c7adc0000
MEM: ! Memory initialized.
... unimplemented function ...

Does it start a second process with sys_game_process_exitspawn?

This works pretty much fine now. Everything is good, speed is even (mostly) in the right order of magnitude on the interpreter. Game is pretty playable regardless though, sound is perfect, graphics are almost perfect, just some tiny glitches here and there. The only thing is that saving and loading doesn't work, and after some amount of time the RSX thread will seemingly randomly die with a "bad flip!" message. Homebrew Doom and Heretic do the same thing it seems.

Also, I decided to try LAN multiplayer, and I think hosting a game works, but searching for servers crashes with something I have never seen before:

(12-25-2014, 12:52 AM)ssshadow Wrote: [ -> ]This works pretty much fine now. Everything is good, speed is even (mostly) in the right order of magnitude on the interpreter. Game is pretty playable regardless though, sound is perfect, graphics are almost perfect, just some tiny glitches here and there. The only thing is that saving and loading doesn't work, and after some amount of time the RSX thread will seemingly randomly die with a "bad flip!" message. Homebrew Doom and Heretic do the same thing it seems.

Also, I decided to try LAN multiplayer, and I think hosting a game works, but searching for servers crashes with something I have never seen before:

Code:

HLE: W {PPU[2568] Thread (ListGamesSystemLink)[0x00504064]} sys_net warning: sendto(s=-1, buf_addr=0xd00debf0, len=11, flags=0x0, addr=0xd00dea80, addrlen=16)
HLE: W {PPU[2568] Thread (ListGamesSystemLink)[0x00504034]} sys_net warning: _sys_net_errno_loc()
PPU: E {PPU[2568] Thread (ListGamesSystemLink)[0x001e4574]} PPUInterpreter::LWZ(): invalid address (0xffffffff30016000)

Huh, cool. I don't think I got around to checking sendto, I've been waiting on it to run at a somewhat decent speed to test it. Does it work with the PPU JIT? Is it any faster?

The LWZ failure I think means the address cannot fit into 32 bits (because of the leading 0xffffffff). I thought the high 32 bits didn't matter with addresses. Someone should check that out.

(12-25-2014, 02:01 PM)derpf Wrote: [ -> ]

(12-25-2014, 12:52 AM)ssshadow Wrote: [ -> ]This works pretty much fine now. Everything is good, speed is even (mostly) in the right order of magnitude on the interpreter. Game is pretty playable regardless though, sound is perfect, graphics are almost perfect, just some tiny glitches here and there. The only thing is that saving and loading doesn't work, and after some amount of time the RSX thread will seemingly randomly die with a "bad flip!" message. Homebrew Doom and Heretic do the same thing it seems.

Also, I decided to try LAN multiplayer, and I think hosting a game works, but searching for servers crashes with something I have never seen before:

Code:

HLE: W {PPU[2568] Thread (ListGamesSystemLink)[0x00504064]} sys_net warning: sendto(s=-1, buf_addr=0xd00debf0, len=11, flags=0x0, addr=0xd00dea80, addrlen=16)
HLE: W {PPU[2568] Thread (ListGamesSystemLink)[0x00504034]} sys_net warning: _sys_net_errno_loc()
PPU: E {PPU[2568] Thread (ListGamesSystemLink)[0x001e4574]} PPUInterpreter::LWZ(): invalid address (0xffffffff30016000)

Huh, cool. I don't think I got around to checking sendto, I've been waiting on it to run at a somewhat decent speed to test it. Does it work with the PPU JIT? Is it any faster?

The LWZ failure I think means the address cannot fit into 32 bits (because of the leading 0xffffffff). I thought the high 32 bits didn't matter with addresses. Someone should check that out.

The recompiler segfaults after the initial logo screens, but it isn't significantly faster before that either. If you have a 8 core/thread CPU like an i7 you can easily run two instances of the game simultaneously now on the interpreter with acceptable performance. It only takes about a minute or two to get to the menu, create a character (remember, saving doesn't work) and select LAN multiplayer.

My Gameplay with Interpreter below

This used to work fine and be pretty much playable, but at some unknown point it stopped working completely, it crashes in under a second now. Log attached.


Ok completely off topic but I looked at the video posted above and then checked out the posters channel and apparently Xenia now runs full "AAA" 3D games at seemingly full speed.... gj Xenia devs I suppose.

https://www.youtube.com/watch?v=L0rsY8XMGBs

Several games broke recently, like the guided fate paradox.
Can you point to a commit that works for puzzle quest?

(08-31-2015, 08:38 PM)vlj Wrote: [ -> ]Several games broke recently, like the guided fate paradox.
Can you point to a commit that works for puzzle quest?

This is ridiculous... I confirmed that it worked in a build from January, but that it in a build from March crashes earlier than it does now. That particular crash doesn't happen any more since one or two months back, but that leaves a window of several months where the current crash was introduced but unable to be found by simple regression testing. (although I did some very quick testing, just a random build from each month)

Also this must be unrelated to the guided fate paradox because it worked recently IIRC.

We can't test every possible game with every commit, users need to post commit when a game is in a "correct enough" state and test regulary.

There are still many hidden problems in emu, it could be that some random commit is not wrong itself but discloses some other bug or race condition.