ScummVM IRC Log - 18 Sep 2017

[Back to Index]

[00:35] --> GitHub189 joined #scummvm.
[00:35] <GitHub189> [scummvm] dreammaster pushed 1 new commit to master: https://git.io/v5Fr0
[00:35] <GitHub189> scummvm/master d48caed Paul Gilbert: TITANIC: Add Deskbot German specific logic
[00:35] GitHub189 (GitHub189@192.30.252.34) left #scummvm.
[01:04] <-- Dominus left irc: Ping timeout: 260 seconds
[01:04] --> Dominus joined #scummvm.
[01:40] <-- Lightkey left irc: Ping timeout: 264 seconds
[01:48] --> dafioram joined #scummvm.
[01:52] --> Lightkey joined #scummvm.
[01:59] <dafioram> I did some playing through titanic and it looks like there may be some cases where memory isn't being freed.
[02:00] <-- Dominus left irc: Ping timeout: 260 seconds
[02:00] <dafioram> I didn't encounter any issues, but when I repeated the same actions it would add memory usage for each repeat
[02:00] <snover> dafioram: are you on windows?
[02:01] <dafioram> I saw it on windows and my ubuntu laptop
[02:01] <snover> valgrind should expose them pretty quickly
[02:02] --> Dominus joined #scummvm.
[02:03] <snover> there is also some thing you can do with Visual Studio, though i dont know anything about it other than that heres a link: https://docs.microsoft.com/en-us/visualstudio/debugger/finding-memory-leaks-using-the-crt-library
[02:03] <dafioram> it may be easier for me to use my linux vm
[02:07] <snover> and theres also Application Verifier from Microsoft, and some tool called Dr Memory
[02:10] <dafioram> does ASan do the same thing as valgrind?
[02:11] Nick change: Stormkeeper -> Storm-AFK
[02:12] <snover> unfortunately not, it detects bad memory usage but not leaks
[02:12] <snover> MemorySanitizer detects leaks but is only available on Linux, IIRC
[02:12] <snover> or, sorry, LeakSanitizer.
[02:14] <snover> i guess i should see LeakSanitizer actually works with apple clang.
[02:14] <snover> i thought that it was added in xcode 8 but it turns out that it was actually ThreadSanitizer.
[02:14] --> GitHub174 joined #scummvm.
[02:14] <GitHub174> [scummvm] dreammaster pushed 1 new commit to master: https://git.io/v5F6z
[02:14] <GitHub174> scummvm/master e6a1c5a Paul Gilbert: TITANIC: Added German DeskbotScript process additions
[02:14] GitHub174 (GitHub174@192.30.252.41) left #scummvm.
[02:15] <snover> answer: nope, not yet. AddressSanitizer: detect_leaks is not supported on this platform.
[02:15] <Lightkey> "an upgrade to a better platform is available"
[02:34] <-- dreammaster left irc:
[02:38] <dafioram> snover: I've made some progress on #9977, but I may need a hint to make faster progress from there
[02:39] <snover> well that wasnt very fun. i spent a bunch of time trying to figure out what was going on with this view i had created in sci companion; it turns out that sci companion does not create valid sci32 views.
[02:39] <snover> although it opens them.

[02:39] <snover> dafioram: ok, how can i help?
[02:39] <dafioram> check comment2 in #9977
[02:42] <dafioram> its not clear to my the handle event call isn't being made when I mouse over the button
[02:43] <snover> what is the normal backtrace for that handleEvent call?
[02:44] <snover> when it works, where does it work from?
[02:44] <dafioram> User::doit()
[02:45] <snover> and is that because User is in doits, or does some other function in the main loop call it explicitly?
[02:47] <dafioram> I saw scarey::play then it was in User::doit twice then User::handleevent
[02:47] <snover> break on User::handleEvent and then run `bt`.
[02:55] <dafioram> here was the most severe titanic memory loss https://pastebin.com/dE2q1h4u and the full memory loss is here https://pastebin.com/f7APVKJr I got this from playing the intro for about 10 secs
[02:56] <snover> whee
[02:57] <snover> so there is a leak checker, its just not as good as asan
[02:58] <snover> but yes, the engine seems to be leaking quite a lot of resources
[03:00] <dafioram> The code section involving bot dialogue is quite large (TTscripts, CTrueTalkManager,..) it may be hard to determine what can be freed and what cannot by just looking at a small portion of it.
[03:01] <snover> before the game starts scummvm is consuming 22MiB, after playing for about 90 seconds and returning to the launcher we seem to have leaked another 22MiB
[03:02] <dafioram> its not unrealistic to see the game using 100mb
[03:03] <snover> the valgrind output looks to match the stack traces i am seeing
[03:03] <snover> so thats good, i actually have a usable memory leak checker now&
[03:04] --> GitHub160 joined #scummvm.
[03:04] <GitHub160> [scummvm] criezy pushed 1 new commit to master: https://git.io/v5FPU
[03:04] <GitHub160> scummvm/master 1d4e2c6 Thierry Crozat: I18N: Regenerate translations data file
[03:04] GitHub160 (GitHub160@192.30.252.35) left #scummvm.
[03:05] <dafioram> User::handleEvent is called too much
[03:05] <dafioram> here is what I got when I breaked on SetCursor https://pastebin.com/wJgFj5ee
[03:06] <dafioram> the bt when in the bug state looks the same except only shows the 0-2 functions in the bt
[03:10] <snover> so that User::doit call is conditional
[03:12] <dafioram> I'm actually not sure what I last say
[03:12] <dafioram> was valid
[03:12] <snover> but the condition is a room state condition so it should always be true
[03:13] <snover> and User::handleEvent receives all events and is called constantly
[03:13] <snover> even during the failure state
[03:13] <dafioram> yes
[03:17] <dafioram> when it first load the room into the bug state it does stop on saveRest::init and doDialog::changeState
[03:17] <dafioram> which is the script for room 91 (main menu screen)
[03:18] <dafioram> then it goes somewhere else from there (after doDialog::changeState)
[03:21] <dafioram> this valgrind report is probably more representative since it is me moving around rather than the prologue https://pastebin.com/JkiadpaJ
[03:24] <snover> dafioram: for phant1, the event is being sent to Feature::handleEvent (which is i.e. oldButton::handleEvent for the load old game button) so its not some issue in the event dispatching but a problem in the event handling of the feature
[03:28] <snover> it seems a bit odd that the SetCursor call is happening due to a arrowCursor::setCel(0) call since that call is *always* arrowCursor::setCel(0) regardless of whether the cursor is highlighting or not
[03:29] <dafioram> where is this?
[03:31] <snover> in the backtrace for every kSetCursor call
[03:32] <snover> er, hm. actually, never mind, i screwed up my breakpoints list
[03:32] <snover> that setCel call is what happens when the cursor changes to red
[03:33] <snover> dafioram: https://gist.github.com/csnover/1d4501a378bcb43632a03cf52b6649f4
[03:34] <dafioram> were did u break 2 get the second bt?
[03:34] <snover> i just used `bpk SetCursor`
[03:35] <dafioram> hmmm my bt doesn't trip until I leave the button, I think it should trip as soon as I am ontop of the button
[03:35] <snover> the first call happens on hover-in of a button when the cursor changes colour successfully, the second call happens on hover-out (which happens immediately after typing `go` since the mouse cursor moved when i went into the debugger)
[03:36] <snover> so with that first backtrace, then the next step is to just bpx on each of the functions in the stack trace to see which one is the broken one (the broken one will be the only one that actually triggers a breakpoint)
[03:37] <snover> and then look at the call site where you expected the next call to happen, but it didnt, and work backwards from there
[03:37] <dafioram> got it
[03:37] <dafioram> thnx
[03:38] <snover> (so, get the game into the broken state, bpx Cursor::init, see if it triggers, if not, reset back to the broken state and try arrowCursor::init, etc.)
[03:40] <snover> dafioram: i am going to take your valgrind report and create tickets out of it
[03:41] <dafioram> as long as I'm not the owner
[03:42] <dafioram> Most of that code I know little about so I'd rather not replace memory leaks for crashes
[03:46] <dafioram> maybe it wouldn't be so bad to add smart pointers in that part
[03:46] <snover> s/in that part/everywhere all the time :)
[03:47] <snover> dafioram: could you please tell me which scummvm build/commit you used?
[03:47] <snover> for the valgrind
[03:48] <dafioram> 1.10.0git-4914-g294503f516
[03:48] <snover> thanks!
[03:53] <dafioram> I could probably look into the leaks in star control
[03:59] <snover> i really despise the DisposeAfterUse flag
[04:00] <dafioram> DespiseAfterUse?
[04:00] <snover> :)
[04:00] <dafioram> low tech shared ptr
[04:19] <snover> this ManagedSurface code is just confusing me
[04:28] <snover> looks like im just confused generally
[04:29] <dafioram> snover: whats the command to do "do the next 50 assembly calls"
[04:30] <snover> trace / t / s - Executes one operation (no parameters) or several operations (specified as a parameter)
[04:46] <dafioram> https://gist.github.com/dafioram/7e1ef043ebaa0fc74210ca3561e3fae2/revisions?diff=split
[04:46] <dafioram> the kernel call is when they are first different
[04:46] <dafioram> this is within hotspot::doit that they diverge
[04:47] <snover> so HotSpot::doit is the first function in the backtrace that triggers on bpx when it is broken?
[04:48] <dafioram> most recent
[04:49] <-- noobineer left irc: Remote host closed the connection
[04:49] <snover> i guess global 193 is the problem then
[04:49] <dafioram> that was my guess
[04:50] --> noobineer joined #scummvm.
[04:50] <snover> it looks like it should be 0 but it is 2 the first time
[04:54] <snover> so it looks like what happens is that when you hover over the in-game quit button that global gets set to 2, but then when you activate the quit button it doesnt get reset to 0
[04:55] <snover> which, actually, can simply be reproduced by clicking the quit button on the main menu and then choosing no in the dialogue that pops up
[04:57] <dafioram> yeah
[04:58] <dafioram> not just hover tho you have to click
[04:59] <snover> when some modal dialogue takes over the game loop User::doit stops receiving the events and User::doit is the one responsible for zeroing out that global
[04:59] <snover> er, sorry, User::handleEvent
[05:00] <snover> looks like sierra were having some troubles with this in room 24100 and hacked in a couple fixes :)
[05:04] <snover> anyway, i need to get to bed, but it seems like resetting global 193 to 0 in quitButton::doVerb(0000:0004) would take care of at least this problem. i dont know if there are any other buttons that cause the same trouble or not.
[05:04] <dafioram> ill test tomorrow
[05:05] <snover> of course theres not any space in that function to do that, unfortunately.
[05:05] <snover> but maybe there is somewhere nearby that works
[05:05] <dafioram> easter egg code
[05:07] <-- dafioram left irc: Quit: Leaving
[05:40] --> K4T joined #scummvm.
[05:43] --> waltervn joined #scummvm.
[05:43] #scummvm: mode change '+o waltervn' by ChanServ!ChanServ@services.
[05:44] <-- waltervn left irc: Client Quit
[05:44] --> waltervn joined #scummvm.
[05:44] #scummvm: mode change '+o waltervn' by ChanServ!ChanServ@services.
[06:36] <-- K4T left irc:
[06:59] --> girafe joined #scummvm.
[07:04] <-- _sev_ left irc: Quit: This computer has gone to sleep
[07:29] <-- TMM left irc: Quit: Ex-Chat
[08:12] --> _sev|work joined #scummvm.
[08:12] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[08:13] <-- girafe left irc: Read error: Connection reset by peer
[08:34] --> GitHub82 joined #scummvm.
[08:34] <GitHub82> [scummvm] angstsmurf opened pull request #1022: SCI: Play MIDI version of SCI0 sound resource if user prefers it (updated) (master...sci0-no-prefer-digital-sounds) https://git.io/v5FAI
[08:34] GitHub82 (GitHub82@192.30.252.40) left #scummvm.
[08:39] --> GitHub112 joined #scummvm.
[08:39] <GitHub112> [scummvm] angstsmurf opened pull request #1023: SCI: Set default MT-32 reverb before each sound (updated) (master...sci0-set-default-reverb) https://git.io/v5FAX
[08:39] GitHub112 (GitHub112@192.30.252.35) left #scummvm.
[08:44] --> TMM joined #scummvm.
[08:44] #scummvm: mode change '+o TMM' by ChanServ!ChanServ@services.
[08:51] <-- LittleToonCat left irc: Remote host closed the connection
[09:52] --> jamm joined #scummvm.
[09:52] <-- jamm left irc: Changing host
[09:52] --> jamm joined #scummvm.
[10:36] <-- triple left irc: Ping timeout: 240 seconds
[10:36] --> triple joined #scummvm.
[10:40] <-- Dominus left irc: Ping timeout: 260 seconds
[10:42] --> Dominus joined #scummvm.
[10:52] --> ignalina joined #scummvm.
[10:59] <-- _sev|work left irc: Read error: Connection reset by peer
[10:59] --> _sev|work_ joined #scummvm.
[10:59] #scummvm: mode change '+o _sev|work_' by ChanServ!ChanServ@services.
[11:48] <-- ignalina left irc: Quit: ignalina
[12:02] --> ignalina joined #scummvm.
[12:25] <-- Cheeseness left irc: Quit: Leaving.
[12:28] --> _sev|work__ joined #scummvm.
[12:28] <-- _sev|work_ left irc: Read error: Connection reset by peer
[13:29] <-- _sev|work__ left irc: Quit: This computer has gone to sleep
[13:58] --> Littleboy joined #scummvm.
[13:58] #scummvm: mode change '+o Littleboy' by ChanServ!ChanServ@services.
[14:01] --> ajax16384 joined #scummvm.
[14:01] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[14:03] <-- LePhilousophe left irc: Ping timeout: 240 seconds
[14:10] --> LePhilousophe joined #scummvm.
[14:11] --> _sev|work joined #scummvm.
[14:11] <-- _sev|work left irc: Changing host
[14:11] --> _sev|work joined #scummvm.
[14:11] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[15:03] --> Henke37 joined #scummvm.
[15:12] <-- ignalina left irc: Quit: ignalina
[15:25] <-- TMM left irc: Quit: Ex-Chat
[15:54] --> GitHub83 joined #scummvm.
[15:54] <GitHub83> [scummvm] csnover closed pull request #831: SCI: Play MIDI version of SCI0 sound resource if user prefers it (master...sci0-no-prefer-digital-sounds) https://git.io/viXjm
[15:54] GitHub83 (GitHub83@192.30.252.45) left #scummvm.
[15:54] --> GitHub126 joined #scummvm.
[15:54] <GitHub126> [scummvm] csnover closed pull request #832: SCI: Set default MT-32 reverb before each sound (master...sci0-set-default-reverb) https://git.io/vi1e6
[15:54] GitHub126 (GitHub126@192.30.252.42) left #scummvm.
[15:58] --> ny00123 joined #scummvm.
[15:59] --> Strangerke_ joined #scummvm.
[16:02] <-- Strangerke left irc: Ping timeout: 264 seconds
[16:02] Nick change: Strangerke_ -> Strangerke
[16:02] <-- ajax16384 left irc: Quit: Leaving
[16:03] <-- _sev|work left irc: Quit: This computer has gone to sleep
[16:05] --> _sev|work joined #scummvm.
[16:05] <-- _sev|work left irc: Changing host
[16:05] --> _sev|work joined #scummvm.
[16:05] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[16:12] --> m_kiewitz joined #scummvm.
[16:12] #scummvm: mode change '+o m_kiewitz' by ChanServ!ChanServ@services.
[16:12] <m_kiewitz> PSA: Amazon France currently sells Super NES Mini
[16:17] <-- Deledrius left irc: Read error: Connection reset by peer
[16:21] --> Deledrius joined #scummvm.
[16:24] <-- Deledrius left irc: Read error: Connection reset by peer
[16:25] <m_kiewitz> PSA2: If anyone uses CCleaner and updated it within the last month, then your computer is probably infected now
[16:26] <m_kiewitz> https://www.engadget.com/2017/09/18/piriform-ccleaner-hacked-malware-downloads/
[16:27] --> Deledrius joined #scummvm.
[16:30] --> ajax16384 joined #scummvm.
[16:30] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[16:38] --> ignalina joined #scummvm.
[16:41] <-- jamm left irc: Ping timeout: 246 seconds
[16:48] <-- LePhilousophe left irc: Ping timeout: 255 seconds
[17:05] <-- _sev|work left irc: Quit: This computer has gone to sleep
[17:05] --> _sev|work joined #scummvm.
[17:05] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[17:05] <-- _sev|work left irc: Client Quit
[17:06] --> LittleToonCat joined #scummvm.
[17:09] <-- ignalina left irc: Quit: ignalina
[17:12] --> ignalina joined #scummvm.
[17:16] --> TMM joined #scummvm.
[17:16] #scummvm: mode change '+o TMM' by ChanServ!ChanServ@services.
[17:23] <-- ignalina left irc: Quit: ignalina
[17:31] --> erdic joined #scummvm.
[17:35] --> Farmboy0 joined #scummvm.
[17:35] <-- Farmboy0 left irc: Changing host
[17:35] --> Farmboy0 joined #scummvm.
[17:45] --> ignalina joined #scummvm.
[18:12] --> _sev_ joined #scummvm.
[18:12] #scummvm: mode change '+o _sev_' by ChanServ!ChanServ@services.
[18:27] Nick change: Storm-AFK -> Stormkeeper
[19:22] <-- thesquib left irc: Quit: ZNC - http://znc.in
[19:29] --> thesquib joined #scummvm.
[19:34] --> girafe joined #scummvm.
[20:12] <-- thesquib left irc: Ping timeout: 246 seconds
[20:14] --> thesquib joined #scummvm.
[20:21] #scummvm: mode change '+o Strangerke' by ChanServ!ChanServ@services.
[20:25] <-- ajax16384 left irc: Read error: Connection reset by peer
[20:25] --> criezy joined #scummvm.
[20:25] #scummvm: mode change '+o criezy' by ChanServ!ChanServ@services.
[20:48] <-- Dark-Star left irc: Remote host closed the connection
[20:55] <-- Farmboy0 left irc: Remote host closed the connection
[20:56] <-- Henke37 left irc: Quit: ERR_SHUTDOWN
[21:06] --> Dark-Star joined #scummvm.
[21:10] <-- waltervn left irc: Quit: Leaving
[21:23] <-- ignalina left irc: Quit: ignalina
[21:24] <-- thesquib left irc: Ping timeout: 240 seconds
[21:27] --> thesquib joined #scummvm.
[21:36] <-- thesquib left irc: Ping timeout: 248 seconds
[21:44] --> thesquib joined #scummvm.
[21:57] <-- girafe left irc: Quit: Leaving
[22:00] --> GitHub118 joined #scummvm.
[22:00] <GitHub118> [scummvm] wjp opened pull request #1024: Clean up use of SearchMan in fallback detection (master...detection_searchman) https://git.io/v5NRq
[22:00] GitHub118 (GitHub118@192.30.252.34) left #scummvm.
[22:05] <wjp> if anybody is able to test or review the cge, cge2, sludge and director fallback detectors, please do
[22:07] <-- borosky left irc: Ping timeout: 248 seconds
[22:09] <criezy> I can take a look at cge and cge2. I have a partically translated French Soltys that should be detected by the fallback detector.
[22:09] <criezy> *partially
[22:12] --> dreammaster joined #scummvm.
[22:12] #scummvm: mode change '+o dreammaster' by ChanServ!ChanServ@services.
[22:13] <-- ny00123 left irc: Quit: Leaving
[22:16] --> GitHub131 joined #scummvm.
[22:16] <GitHub131> [scummvm] dreammaster pushed 1 new commit to master: https://git.io/v5NEf
[22:16] <GitHub131> scummvm/master 5af8ed2 Paul Gilbert: TITANIC: Add VocabMode enum, fix German vocab loading
[22:16] GitHub131 (GitHub131@192.30.252.42) left #scummvm.
[22:22] <-- exmensa left irc:
[22:23] --> ignalina joined #scummvm.
[22:29] <-- Stormkeeper left irc: Ping timeout: 240 seconds
[22:29] --> Stormkeeper|2 joined #scummvm.
[23:06] --> Cheeseness joined #scummvm.
[23:32] <-- ignalina left irc: Quit: ignalina
[23:39] <-- kurtwr left irc: Ping timeout: 248 seconds
[23:40] --> kurtwr joined #scummvm.
[23:46] --> dafioram joined #scummvm.
[00:00] --- Tue Sep 19 2017