January 5, 2005 
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >
January 5, 2005 [00:00:31] *** Potn has quit IRC[00:01:00] *** Potn has joined #haiku[00:06:38] *** Koki has quit IRC[00:12:42] *** illissius is now known as illissius[sleep][00:18:39] *** Potn has quit IRC[00:19:20] *** Potn has joined #haiku[00:26:37] *** Potn has quit IRC[00:27:17] *** Potn has joined #haiku[00:27:26] <[Beta]> hmm, haiku-os.pl have an impressive logo[00:35:52] <Fanskapet> yeah sure have[00:36:00] <Fanskapet> if only yellowtab could get a decent logo :)[00:37:25] *** Potn has quit IRC[00:37:41] *** Potn has joined #haiku[00:39:12] *** BGA has quit IRC[00:39:17] *** geist has joined #haiku[00:40:22] <Dr_Evil> hi geist[00:40:47] <andrew_working> hello geist[00:40:53] <andrew_working> Dr_Evil you're up late recently[00:41:01] <[Beta]> I hope the first haiku 'distrib' comes with scummvm & Beneath a Steel Sky - more people should play that game.[00:41:10] <slaad> Hehe.[00:41:24] <[Beta]> evenin'[00:41:29] <slaad> [Beta].[00:42:15] <Dr_Evil> andrew_working holidy until 10th january[00:42:23] <fyysik> now with log enabled crashed in driver text in log_printf. nonsense[00:42:36] <[Beta]> Fanskapet: maybe they should rename zeta to delta, thats a cool symbol ;)[00:42:43] <AnEvilYak> [Beta]: I never did beat that game[00:42:53] <andrew_working> oic[00:42:53] <AnEvilYak> [Beta]: back in the days when it came out I got close to the end but got stuck[00:43:00] <[Beta]> it is a tricky one.. i'm currently replaying Sam n Max on my mobile[00:43:07] <[Beta]> full talkie version :>[00:43:15] <andrew_working> say is there anyone here who intends to do java hacking on beos[00:43:16] <AnEvilYak> yeah, I have the CDs for most of the lucasarts games[00:43:19] <AnEvilYak> bought the archives long ago.[00:43:31] *** ConneX has quit IRC[00:43:32] <AnEvilYak> andrew_working: I thought it was mostly BryanV doing that on his own at this point?[00:43:40] <Dr_Evil> fyysik as I said, the crashes seem to happen at perfecty valid locations doing perfectly valid things, and stack esp as well as code eip are correct and point to valid memory[00:43:49] <andrew_working> no, I mean writing java programs on beos, not porting java[00:43:50] <AnEvilYak> [Beta]: iirc I got stuck when you're at the bottom floor and you find that hologram of the girl telling you there might be a way in[00:43:51] <[Beta]> I have.. all of their adventures. most of the star wars. none of the crap. ;)[00:43:53] <AnEvilYak> andrew_working: oh.[00:44:06] <[Beta]> Andrew : i'd like to.[00:44:37] <[Beta]> AnEvilYak: such a long time ago.. i'll wait for haiku to put it on my desktop :)[00:45:01] * tic pets Joey[00:45:10] <AnEvilYak> [Beta]: hehe.[00:45:15] <tic> (the robot, not the Friend.)[00:45:23] <[Beta]> ooh. is Joey taken as a Be app ?[00:45:36] <tic> don't think so..[00:45:43] <[Beta]> sweet. andrew_working why do you ask ?[00:46:18] <andrew_working> wondering who the customers are[00:46:36] <tic> mix one http://www.icd.se/default.aspx?aSa=Vp&Pi=982 with two of these: http://www.icd.se/default.aspx?aSa=Vp&Pi=912[00:46:57] <tic> presto - fanless 200W ATX PSU.[00:47:10] *** gipfex has joined #haiku[00:49:19] *** Potn has quit IRC[00:49:49] *** Potn has joined #haiku[00:51:23] <Dr_Evil> AndrewBachmann I'm making DVB progress[00:51:49] *** BetaMax has joined #haiku[00:51:53] <Dr_Evil> the hardware does lock to the incoming stream, it just needs 530 ms, and thats 430 more than I expected[00:56:16] *** MikeW has joined #haiku[00:57:24] <andrew_working> what is DVB?[00:57:54] <Dr_Evil> DVB-T digital video broadcasting using tv antenna[01:00:50] <andrew_working> oic[01:00:56] <andrew_working> do you know who is working on the 855 driver?[01:01:01] <andrew_working> graphics[01:01:13] <fyysik> hm, suspicious thing. interrupt installer uses parameters from card->streams in if() statements, but i don't see that those streams members get reasonable values before installing interrupt. only zeroing pointers and memory[01:01:50] <Dr_Evil> andrew_working no i don't[01:02:12] <andrew_working> ok[01:08:08] <tic> andrew_working, allen reese or some such.[01:08:17] <tic> the network guy at driversoft[01:08:32] <tic> haven't heard from him in a very long while, though.[01:08:48] *** mwilber has joined #haiku[01:13:07] <AnEvilYak> tic: I was under the impression he got another job[01:13:44] <tic> AnEvilYak, prolly. it's was just so people'd know who he is.[01:13:58] <tic> and now, if you excuse me, it's 01:16 and I'm getting up in 4.5 hours. :P[01:14:00] <tic> night![01:14:41] *** gipfex has left #haiku[01:15:18] *** Potn has quit IRC[01:15:48] *** Potn has joined #haiku[01:16:09] <slaad> You're such a pansy, tic.[01:16:16] <slaad> "Oh, I've got to get up in 4.5 hours." ;)[01:16:25] <tic> bah![01:16:29] * tic just checked the bus tomorrow[01:16:32] <tic> and now, sleep again.[01:17:04] * tic snorts at slaad[01:17:26] <tic> (sniffs?)[01:17:28] <tic> aaaaaanyway. *poff*[01:18:17] * slaad grins[01:18:30] *** Karina``` has quit IRC[01:19:22] *** BetaMax has quit IRC[01:22:47] <mwilber> AndrewBachmann, andrew_working: Are there any special libraries that I need to have in order for Haiku StyledEdit to run?[01:22:57] *** Potn has quit IRC[01:23:19] *** Potn has joined #haiku[01:24:37] <andrew_working> libtextencoding[01:24:50] <slaad> libpants![01:25:42] <Dr_Evil> libspechfilter[01:25:51] <AnEvilYak> libwedgie[01:25:56] <Dr_Evil> libspeechfilter[01:26:02] <slaad> libasshat![01:26:06] <AnEvilYak> rofl[01:26:17] <slaad> Must... resist... changing... WebIM library...names...[01:26:51] <AnEvilYak> rofl[01:26:54] <AnEvilYak> why resist?[01:27:00] <AnEvilYak> geek humor rules :)[01:27:02] <slaad> "To use WebIM you must install libasshat and libpants. You can obtain these from beos.assh.at" ;)[01:27:09] <mwilber> andrew_working: Ah right, that would explain why it gets built when I do a "jam -a StyledEdit"[01:27:11] <AnEvilYak> hahaha.[01:27:36] <AnEvilYak> slaad: im.assh.at[01:27:41] <slaad> Hahah[01:27:43] *** Karina``` has joined #haiku[01:27:48] <AnEvilYak> slaad: yay double entendres ;p[01:28:03] <slaad> Common pay cheque.[01:28:22] <slaad> s/Common/Come ooon/[01:28:42] <AnEvilYak> ;p[01:30:28] *** Potn has quit IRC[01:30:53] <slaad> Hrm. BeMail seems to generate invalid message IDs. Which could cause it to get lost in common spam filters.[01:31:19] *** ShackaN has quit IRC[01:31:30] *** Potn has joined #haiku[01:32:01] *** MikeW has quit IRC[01:34:19] *** Potn has quit IRC[01:34:28] *** Potn has joined #haiku[01:35:26] *** DaaT has quit IRC[01:37:36] *** ChanServ sets mode: +o geist[01:38:35] <fyysik> Dr_Evil[01:38:43] *** Potn has quit IRC[01:39:24] <fyysik> Dr_Evil - i created printout for return value of install_interrupt_handler. It seems retruning B_UNHANDLED_INTERRUPT in my case[01:40:12] <Dr_Evil> sorry, but that doesn't matter[01:40:26] <fyysik> ok[01:40:35] *** Potn has joined #haiku[01:42:49] *** ShackaN has joined #haiku[01:43:44] <Dr_Evil> perhaps geist has an idea what is going on, I however really don't know what kind of bug you are expirencing. it is likely that an interrupt has happened and that this causes your computer to crash[01:43:52] <fyysik> heh, no crash. rare occasion[01:44:10] <Dr_Evil> but the data we find in KDL looks perfectly valid, so it should not have crashed at thast location where it is crashing[01:44:28] <fyysik> seems so[01:45:26] <fyysik> ik> i saw at least one more person at bebits talkback who had same problem[01:52:29] *** oco_2 has quit IRC[01:52:37] <fyysik> again no crash. but cannot switch to channel 1. Only 0/1 is possible. Mystics[01:55:07] <fyysik> ok, restored haiku media prefs, switching[01:56:27] *** brennanOS has joined #haiku[01:58:14] * fyysik wonders if SB Live still has Covox/parallel port emulation[01:58:45] <fyysik> now reasonable crash and sc[01:59:10] <fyysik> happened in emuxki_inte_enable[01:59:35] *** Potn has quit IRC[01:59:50] *** Potn has joined #haiku[02:06:30] <fyysik> Dr_Evil - pls look if this is more reasonable http://beos.spb.ru/fyysik/emuxki_KDL.JPG[02:06:45] <fyysik> sc, dis eip, regs there[02:08:24] *** Potn has quit IRC[02:09:01] <Dr_Evil> fyysik yes, esp is invalid, to ret is crashing[02:09:27] * fyysik charged camera batteries a bit[02:09:47] *** Potn has joined #haiku[02:10:16] <Dr_Evil> fyysik what does "address esp" give you?[02:10:25] <fyysik> minute[02:10:50] *** tatsujin has joined #haiku[02:10:52] <Dr_Evil> fyysik and what does "address eip" give you?[02:11:12] <fyysik> area 05bc7c48 addr fc005000 size 01001000[02:11:37] <fyysik> that was address esp[02:11:48] *** brennanOS has quit IRC[02:11:56] <Dr_Evil> what are the names?[02:12:14] <fyysik> address fd0059cc:[02:12:43] <fyysik> that was esp[02:13:06] <fyysik> esp name was Stacks of medai addon server[02:13:31] <fyysik> address eip: address 00159fe3:[02:14:07] *** Potn has quit IRC[02:14:17] <fyysik> area 05b1b768 kernel_intel_text addr 001 00000 size 00087000[02:14:27] <Dr_Evil> that esp is a user stack and is invalid in kernel I think, geist are you here?[02:14:44] *** Potn has joined #haiku[02:15:18] <fyysik> sometimes i was getting in sc: sysinit after blah blah (context?) switch[02:15:44] <Dr_Evil> fyysik I *think* that the interrupt handler of the driver somehow clobbers the stack[02:16:23] <Dr_Evil> so that shortly after the interrupts are enabled, the first interrupts overwrites soemthing on the stack, and then the esp or the return address from stack get invalid[02:16:33] <fyysik> if you gimme some more commands i can do next photoshot[02:17:37] <slaad> "Oh, yeah, that's it baby. Show me a little more stack. Ahuh, now give me a corrupted pointer. That's it, perfect! You'll be on the cover of all the magazines"[02:18:01] <[Beta]> random :)[02:18:06] *** Master199 has quit IRC[02:18:11] <Dr_Evil> address[02:18:13] <fyysik> slaad - are you also assembler hacker?P)[02:18:16] <Dr_Evil> oops[02:18:30] <Dr_Evil> "address 603cdd8f" would be interesting[02:18:52] <@geist> HELLO SIRS[02:18:55] <slaad> Hah, nope. Only ASM I do is for microcontrollers.[02:18:57] <fyysik> just in that form, without 0x prefix?[02:19:05] <Dr_Evil> along with those other two address commands frmo earliert[02:19:11] <Dr_Evil> and then a photo[02:19:19] <Dr_Evil> fyysik shouldn't matter[02:19:25] <@geist> hi Dr_Evil how are you today im fine[02:19:52] <[Beta]> lo geist.[02:20:05] <Dr_Evil> hi geist, I'm fine, too[02:20:24] <Dr_Evil> even my DVB-T driver seems to be fine, tuning does now work :)[02:20:40] <Dr_Evil> we are trying to locate a crash in anther driver[02:20:41] *** Potn has quit IRC[02:20:51] <@geist> hahah thats grate i got bak from vaycatesion[02:20:56] <@geist> it was fun haha[02:21:03] <Dr_Evil> can you have a look at <fyysik> Dr_Evil - pls look if this is more reasonable http://beos.spb.ru/fyysik/emuxki_KDL.JPG[02:21:10] <AnEvilYak> geist: I'm afraid to ask if you're sober or not ;p[02:21:14] <Dr_Evil> and the backlog here?[02:21:22] <@geist> AnEvilYak: DONT TALK TO ME LIEK THAT[02:21:27] <AnEvilYak> <g>[02:21:28] <slaad> Haha.[02:21:40] <@geist> YUR NOT MY MOM[02:21:48] * AnEvilYak grins[02:22:13] <Dr_Evil> fyysik after you took a photo of the three address commands, you should reboot and disassemble the driver[02:22:30] <Dr_Evil> using objdump --disassemble thedriver >thedriver.asm[02:22:39] <slaad> This just in; going to the grog store to buy booze does not count as a vacation ;)[02:23:01] * AnEvilYak pours grog over slaad's head[02:23:02] <@geist> yeah, I guess you're right[02:23:10] *** Potn has joined #haiku[02:23:13] * slaad decides not to take up smoking...[02:23:16] <@geist> oh wow I haven't seen a real KDL in such a long time[02:23:19] <@geist> that brings back memories[02:23:27] <slaad> Burning head does not make for a happy Mikey.[02:23:35] <@geist> like the damn machine I had at work that sporadically generated NMIs[02:23:46] <slaad> NMI?[02:23:46] <@geist> and would spontaneously KDL. most of the time you could continue[02:23:52] <AnEvilYak> non-maskable interrupt[02:24:11] <@geist> it was a super fast machine though. it was a prototype P3 in a world of P2s[02:24:33] <slaad> Ah.[02:24:37] *** tatsujin has quit IRC[02:24:40] <fyysik> Dr_Evil - http://beos.spb.ru/fyysik/emuxki_KDL2.JPG[02:27:10] <fyysik> Dr_Evil objdump done[02:27:40] <Dr_Evil> fyysik well, the most likely source for this problem is:[02:28:05] <Dr_Evil> interrupt handler is installed without setting up every variable, that is used in interrupt handler, first.[02:28:20] <Dr_Evil> because of interrupt sharing, the handler can be executed before card interrupts are enabled[02:28:23] <fyysik> that;s very possible[02:28:34] <Dr_Evil> second possible source:[02:28:35] <[Beta]> nn[02:28:56] <fyysik> crash happens without visible sharing too[02:28:58] <Dr_Evil> interrupts are enabled before every variable, that is used in interrupt handler, is initialized.[02:29:11] <Dr_Evil> in case of interrupt sharing, both problems can coexist[02:30:37] * geist curses interrupt sharing[02:30:46] <fyysik> Dr_Evil - http://beos.spb.ru/fyysik/emuxki.asm[02:31:16] <Dr_Evil> using the output of address 603cdd8f you can calculate wich line in the disassembled driver is crashing (but as it's crashing in the kernel, you can't do that now)[02:31:26] <fyysik> maybe interrupts are shared at very early state, but when all is loaded, each device has its own here[02:31:27] *** andrew_working has quit IRC[02:31:40] <Dr_Evil> however, you can have a look what's at emuxki_int_enable + 0x17[02:32:09] <Dr_Evil> fyysik no, that doesn't happen. interrupts are always shared, or not at all[02:32:26] <Dr_Evil> geist do you think I'm correct?[02:33:32] <Dr_Evil> btw there is some dirt on your screen fyysik ;)[02:33:39] <fyysik> 0x17 seem ret instruction[02:33:40] <AnEvilYak> I was under the impression BeOS just uses the BIOS-assigned interrupt configuration anyhow...so if it's shared at the beginning it always will be[02:33:48] <@geist> right[02:34:05] *** brennanOS has joined #haiku[02:34:23] <fyysik> or pop %ebp, depends on how to count shift[02:34:42] <Dr_Evil> yes, BeOS is dump. doesn't do IRQ reassignment, or PCI memory/io resource assugnment at all[02:35:17] <AnEvilYak> I thought most OSes just leave that to the hardware anyhow[02:35:23] <AnEvilYak> Was under the impression only windows tries to reassign stuff.[02:35:32] <Dr_Evil> fyysik I'm afraid, there should be a call at emuxki_int_enable + 0x17, but that isn't importnant anymore, as we now know the problem[02:37:09] *** nPHYN1T3 has quit IRC[02:37:29] <fyysik> interesting that in emuxki_inte_disable there is lot of nop after ret - for alignment?[02:37:55] <Dr_Evil> yes[02:38:23] <Dr_Evil> it could be anything, I guess gcc uses nop to not confuse debuggers[02:38:29] *** mwilber has quit IRC[02:40:12] *** mwilber has joined #haiku[02:40:20] *** Potn has quit IRC[02:40:24] <fyysik> ok, ll reboot and look at C code again[02:40:39] <fyysik> hope Korli will look at screenshots too[02:40:48] <Dr_Evil> geist so you are on holiday?[02:41:17] <@geist> just got back today[02:41:22] <@geist> still reading my mail and unpacking[02:41:26] <slaad> Where'd you go?[02:42:31] <@geist> to texas to visit my parents[02:42:35] <@geist> been there for a few weeks[02:42:57] <Dr_Evil> I'm not even sure if my driver initialize *everything* before installing interrupt handler, because ony assums that they are disabled. due to interrupt sharing, they can happen however[02:43:04] <Dr_Evil> that makes a nice magazin article[02:44:20] *** Potn has joined #haiku[02:44:44] <fyysik> fine[02:44:52] <Dr_Evil> voidref that would make a nice blog article ;)[02:44:54] *** liquidboy has joined #haiku[02:45:15] <fyysik> emuxki_inte_enable is called BEFORE install_io_interrupt - is it OK?[02:45:57] *** Potn has quit IRC[02:46:12] <@voidref> Dr_Evil. i am not driver l33t[02:46:39] <Dr_Evil> fyysik no thats dirty[02:46:39] *** CIA-5 has quit IRC[02:46:57] <Dr_Evil> but that might not be the reason for this crash[02:47:17] <slaad> Is all your time spent on user level stuff, voidref?[02:47:26] <Dr_Evil> voidref I might write one for haiku[02:47:44] *** Potn has joined #haiku[02:47:46] <fyysik> moved, compiled, rebooting[02:47:50] <@geist> voidref is an english pigdog[02:48:10] <Dr_Evil> fyysik it's likely that emuxki_inte_enable is used for disabling interrupts, I won't look at the source[02:49:25] <Dr_Evil> fyysik I'm finished with this issue now, as my DVB-T driver finally gets a viterbi decoder lock, I'm continuing that one[02:50:09] <fyysik> heh, quite exotic for our country project - DVB TV[02:50:44] <fyysik> it is very nice to get in BeOS something so modern[02:53:28] <Dr_Evil> an 05 01:53:28 <Dr_Evil> a well, at least my DVB driver first maps the registers, than disables interrupts, clears pending interrupt bits, and then installs the handler[02:54:03] <Dr_Evil> so it's unlikely that anything unitialized is executed in the handler[02:54:38] <Dr_Evil> geist did you ever think about this issue of interrupt sharing drivers?[02:55:45] <@geist> what about it[02:56:26] <Dr_Evil> that an interrupt handler is executed despite hardware interrupts being disabled[02:56:49] <@geist> yes, the handler needs to deal with that[02:56:50] <Dr_Evil> because the interrupt is generated by another device[02:57:02] <@geist> generally speaking it's pretty obvious what's going on[02:57:13] <Dr_Evil> yes, I was thinking of a sequence like:[02:57:55] <fyysik> btw, this crash didn;t happen, IIRC, at Duron 800 system[02:58:00] <Dr_Evil> map_hardware_registers(); reset_hardware(); disable_hardware_interrupts(); install_int_handler(); initialize_other_stuff();[02:58:06] <fyysik> with same BeOS install[02:58:42] <Dr_Evil> but the initialize_other_stuff(); is done too late, because another device can generate the interrupt and the handler will be executed[02:59:24] <@geist> worse than that, it might happen on another cpu immediately[02:59:31] <@geist> before the initialize_other_stuff[03:00:07] <@geist> so you should either keep a flag that says if it's initialized, or make sure it's initialized before you install the handler[03:00:17] <Dr_Evil> yes, that waht I ment[03:00:27] <@geist> pretty standard stuff[03:00:38] <@geist> one of the tricky things about driver development that's not real obvious up front[03:02:36] *** CIA-6 has joined #haiku[03:03:08] <Dr_Evil> well most drivers don't seem to do this[03:03:19] <Dr_Evil> at least those that I looked at[03:04:09] <@geist> yay[03:04:53] <Dr_Evil> for example, the linux bttv driver installs irq handler pretty early, and doesn't do checks for init done in the irq handler[03:05:13] <@geist> nice[03:05:25] <@geist> ist> is it holding a spinlock when it does it though?[03:05:34] <@geist> some sort of per-device context spinlock?[03:06:57] <Dr_Evil> no[03:07:26] <Dr_Evil> . /* disable irqs, register irq handler */[03:07:29] <Dr_Evil> btwrite(0, BT848_INT_MASK);[03:07:30] <Dr_Evil> result = request_irq(btv->c.pci->irq, bttv_irq, SA_SHIRQ | SA_INTERRUPT,btv->c.name,(void *)btv);[03:07:40] <Dr_Evil> . /* initialize hardware */[03:07:46] <Dr_Evil> init_bt848(btv);[03:07:50] <Dr_Evil> [...][03:08:41] <Dr_Evil> ahh well, I'm going to sleep[03:08:45] <fyysik> me too[03:08:59] <fyysik> was nice to get some more wisdom[03:09:42] <Dr_Evil> btw fyysik[03:09:49] <Dr_Evil> is this a SMP system?[03:09:58] <fyysik> no[03:10:07] <fyysik> but at SMP it crashed always[03:10:27] *** Potn has quit IRC[03:11:08] *** Potn has joined #haiku[03:13:00] *** Potn has quit IRC[03:13:56] *** Potn has joined #haiku[03:14:25] <CIA-6> mwilber * current/src/kits/translation/TranslatorRoster.cpp: Removed silly PPMTranslator hack because it caused issues with the Haiku RTF-Translator and it really never should have been added in the first place.[03:18:50] *** BetaMax has joined #haiku[03:22:32] *** Potn has quit IRC[03:23:58] *** Potn has joined #haiku[03:28:13] *** fyysik has quit IRC[03:29:56] *** Potn has quit IRC[03:35:29] *** mwilber has quit IRC[03:38:56] *** liquidboy has quit IRC[03:51:58] *** ShackaN has quit IRC[03:53:07] *** Kernel86 has quit IRC[04:02:55] *** mmadia has joined #haiku[04:20:04] *** BetaMax has quit IRC[04:51:14] *** mmadia has quit IRC[04:54:49] *** BetaMax has joined #haiku[04:57:25] *** BetaMax has quit IRC[04:57:28] *** Betabel has joined #haiku[04:57:42] *** Betabel is now known as BetaMax[05:00:54] *** Begasus has joined #haiku[05:15:56] * AndrewBachmann pings voidref[05:29:50] *** Potn has joined #haiku[05:45:35] <@AndrewBachmann> hrm this RTF translator is usable[05:53:17] <slaad> You sound surprised.[05:54:11] <@AndrewBachmann> well it's not up to the level of gobe's yet[05:54:22] <@AndrewBachmann> which also makes me curious[06:01:31] * slaad nods[06:03:33] <@AndrewBachmann> seems like no changes were required to stylededit to get it to use the rtf translator which is pretty cool[06:05:58] <@AndrewBachmann> Fanskapet[06:08:31] <@AndrewBachmann> bbiab[06:08:32] *** AndrewBachmann has quit IRC[06:26:16] *** Potn has quit IRC[06:26:27] *** Potn has joined #haiku[06:27:15] *** RageMax has quit IRC[06:37:54] *** Potn_ has joined #haiku[06:39:22] *** Potn has quit IRC[06:49:55] *** AndrewBachmann has joined #haiku[06:49:55] *** ChanServ sets mode: +o AndrewBachmann[07:09:28] *** Bernd has joined #haiku[07:12:05] <@AndrewBachmann> hello Bernd[07:12:16] <Bernd> hi[07:13:01] <@AndrewBachmann> is there someone still working on the i855 graphics driver[07:13:31] <Bernd> ask me?[07:13:49] <@AndrewBachmann> yes, I thought they were working for yt?[07:13:58] <Bernd> they=you? :)[07:14:10] <Bernd> yes one is working on an i855 driver[07:14:31] <@AndrewBachmann> I'm looking at getting a laptop and several of the nice ones have the i855 chipset[07:15:10] <Bernd> i prefer nvidia[07:15:22] <Bernd> just bought me some weeks ago a Samsung M40[07:15:26] <@AndrewBachmann> nvidia is pretty rare in the laptop world[07:15:27] <Bernd> very good notebook[07:15:34] <Bernd> 17"[07:15:37] <Bernd> 16:10[07:15:51] <Bernd> 1,6GHz Centrino[07:15:54] <Bernd> and amazing fast[07:16:09] <slaad> Where's mine?[07:16:18] <Bernd> left side?[07:16:19] <Bernd> :)[07:16:26] <slaad> Nuts![07:16:32] <Bernd> uh#[07:16:45] <Bernd> lost?[07:16:59] <Bernd> :)[07:18:37] *** hUMUNGUs has joined #haiku[07:19:40] <slaad> Oh well, back to the grindstone.[07:35:53] *** BetaMax has quit IRC[07:36:54] <brennanOS> you guys chatting at work?[07:39:05] <slaad> Me? Nope.[07:39:44] <@geist> it's 22:39 here[07:39:53] <@AndrewBachmann> ditto[07:40:25] <slaad> 17:40.[07:40:37] <brennanOS> it is 00:39 here[07:41:16] <brennanOS> I need to go to bed[07:41:24] <Bernd> hm funny, fligh from Hamburg to Cologne is 79 Euro and one hour. By train it is 5 hours and 80 Euro[07:42:01] *** brennanOS has quit IRC[07:42:08] <@AndrewBachmann> it's always like that in the US[07:42:14] <@AndrewBachmann> you pay a real premium to take the train[07:42:18] <Bernd> makes no sense for me[07:42:21] <@AndrewBachmann> it's worse, even[07:42:25] <Bernd> In Germany trains are different then in USA[07:42:31] <Bernd> it is one of the main traffic ways[07:42:34] <@AndrewBachmann> yeah, they are better[07:42:34] <@AndrewBachmann> :-)[07:42:36] <Bernd> and really comfortable[07:42:53] <@AndrewBachmann> well, ours are pretty comfortable actually[07:42:56] <Bernd> you can go from Hamburg to Mannheim (500mls) in around 4 hours by train[07:43:05] <slaad> Other advantage of a train is that you can take more luggage, I imagine.[07:43:13] <Bernd> slaad true yes[07:43:28] <Bernd> but who need more then 20KG if you go inside your country[07:43:32] <@AndrewBachmann> it's a lot easier to step on to a train than to walk up and step on a plane[07:43:49] <Bernd> have a 2 hour meeting there and travel ~11 hours by train then[07:43:51] <Bernd> haha[07:44:01] <Bernd> i will take my car i think[07:44:07] <tic> AndrewBachmann, but that's offset by the 4 hour diff, don't you think? :)[07:44:08] <Bernd> then i need 3 hours[07:44:18] <Bernd> :)[07:44:22] <tic> horray for autobahn, eh? :)[07:44:23] <slaad> Also, you can take your car on a train.[07:44:28] <Bernd> tic yeah![07:44:30] <Bernd> and fast car :))[07:44:38] <Bernd> i love that in germany[07:44:43] <slaad> And > 20kg luggage could apply. If you're moving across country. Or going on a holiday to a shopping "city".[07:44:45] <tic> was the car badly wrecked in the crash?[07:44:46] <Bernd> mostly you can drive as fast as you want and can[07:45:02] <Bernd> tic this is "dead"[07:45:08] <Bernd> nothing fixable on it[07:45:12] <Bernd> i bought me a new one[07:45:23] <tic> just like that? You're pretty rich :)[07:45:30] <Bernd> no, need a car[07:45:46] <Bernd> you do not need money to buy a car :)[07:45:48] <Bernd> haha[07:46:03] <tic> true; I suppose you could take a loan[07:46:08] <Bernd> yep[07:46:12] <Bernd> which makes sense for me[07:46:21] <tic> mhm[07:46:22] <Bernd> you get cars for 0.5% - 3% as loan[07:46:31] <tic> that's a pretty low rate[07:46:35] <Bernd> you do not get more if you pay it in one time[07:46:36] <tic> you don't fear it'll go up soon?[07:46:41] <Bernd> nope[07:46:46] <tic> okay :)[07:46:47] <Bernd> i bought my house 12 years ago[07:46:55] <Bernd> and the loan was in the middle 4.5% since then[07:47:00] <tic> nifty.[07:47:02] <Bernd> and this is really cheap for a house credit[07:47:09] <Bernd> makes no sense to pay cash imho[07:47:11] <tic> you don't consider moving down to Mannheim altogether?[07:47:16] <Bernd> i will[07:47:23] <tic> ah[07:47:33] <Bernd> i am searching for a house there and will "for rent" my one here[07:47:41] <Bernd> i am 3 days/week in Mannheim[07:47:44] <Bernd> and the rest here in Hamburg[07:48:04] <Bernd> or anywhere else i have to work[07:48:14] <tic> yap.[07:48:59] <Bernd> it is not easy to find a place to stay with your family. It should not be worth then what i have. So i have to look intensive for one[07:49:07] <Bernd> and my son is 8 and goes to school[07:49:16] <Bernd> also not that easy just to move, the timing must work[07:49:49] <slaad> Wow. Little Berndlettes.[07:49:57] <Bernd> yep[07:49:58] <Bernd> :)[07:50:02] <Bernd> my other son is 3 months[07:50:08] <Bernd> even more little *LOL[07:52:06] <slaad> 3months and 8? That's a pretty big gap.[07:52:11] <Bernd> yes[07:52:15] <Bernd> wasnt other possible[07:52:26] * slaad nods[07:54:03] <tic> slaad, at Siemens, where I work, 6/8 are directly from the CS class I'm from, three from my year, and three from one year above or below, rest are from the IT university. fun. :)[07:54:14] <slaad> Scary.[07:54:29] <slaad> That reminds me... I should email about this job.[07:55:23] <tic> eek, gotta catch a bus.[07:55:26] <tic> *runs*[07:57:05] <Bernd> haha[08:05:15] *** Bernd has left #haiku[08:13:48] *** Dr_Evil has quit IRC[08:19:45] *** Koki has joined #haiku[08:22:36] *** kr1stof has joined #haiku[08:38:51] *** Koki has quit IRC[08:46:40] *** Begasus has quit IRC[08:47:44] *** Potn_ has quit IRC[08:48:23] *** Potn_ has joined #haiku[09:01:53] *** emwe has joined #haiku[09:07:06] *** JBurton has joined #haiku[09:07:09] <JBurton> hi[09:14:40] *** slaad has quit IRC[09:18:11] *** lizdeika_ is now known as lizdeika[09:38:09] *** Potn_ has quit IRC[09:39:30] *** Potn_ has joined #haiku[09:50:19] *** Korli has joined #haiku[09:50:23] *** ChanServ sets mode: +o Korli[09:50:39] <@Korli> hi geist JBurton[09:50:45] <JBurton> heya Korli[09:50:49] <JBurton> it's been a while :P[09:51:08] <@Korli> back at work :p[09:54:40] <JBurton> heh, about time :P[09:55:07] <@Korli> JBurton happy new year[09:55:40] <JBurton> thanks Korli and to you too[09:57:15] *** Dr_Evil has joined #haiku[10:00:20] <@Korli> hi Dr_Evil[10:01:32] <JBurton> hi Dr_Evil[10:12:11] *** lizdeika has quit IRC[10:24:41] *** BGA has joined #haiku[10:24:41] *** ChanServ sets mode: +o BGA[10:37:16] *** RoC|Vortex has quit IRC[10:40:27] *** BGA has quit IRC[10:46:17] *** emwe has quit IRC[10:50:57] *** Potn_ has quit IRC[10:51:39] *** Potn_ has joined #haiku[10:53:28] *** kr1stof has quit IRC[10:53:45] *** Trevor1 has quit IRC[10:56:01] *** thaflo has joined #haiku[10:56:55] *** thaflo has quit IRC[10:57:29] *** thaflo has joined #haiku[10:59:15] *** liquidboy has joined #haiku[11:01:11] *** liquidboy has left #haiku[11:01:39] *** BGA has joined #haiku[11:01:40] *** ChanServ sets mode: +o BGA[11:05:33] *** BGA has quit IRC[11:07:16] *** BGA has joined #haiku[11:07:16] *** ChanServ sets mode: +o BGA[11:11:44] *** emwe has joined #haiku[11:24:16] *** Begasus has joined #haiku[11:26:16] *** emwe has quit IRC[11:34:09] *** Trevor1 has joined #haiku[11:37:02] *** stippi has joined #haiku[11:37:13] *** slaad has joined #haiku[11:39:29] *** BGA has quit IRC[11:53:35] *** Potn_ has quit IRC[11:54:52] *** Potn_ has joined #haiku[11:59:49] *** fyysik has joined #haiku[12:24:21] <fyysik> Korli - did Dr_Evil send two screenshots link to you?[12:26:03] *** ShackaN has joined #haiku[12:32:32] *** sys2 has joined #haiku[12:36:03] *** Teknomancer has joined #haiku[12:39:05] <fyysik> Korli - http://beos.spb.ru/fyysik/emuxki_KDL.JPG and[12:39:18] <fyysik> http://beos.spb.ru/fyysik/emuxki_KDL2.JPG[12:45:05] *** Racer__X has joined #haiku[13:03:47] *** fyysik has quit IRC[13:16:15] *** tqh has joined #haiku[13:21:03] *** illissius[sleep] has quit IRC[13:21:11] *** illissius[sleep] has joined #haiku[13:26:02] *** emwe has joined #haiku[13:30:58] <ShackaN> is there a rss feed with the haiku mailing lists?[13:31:32] <JBurton> don't think so[13:32:38] *** slaad has quit IRC[13:39:16] <Dr_Evil> *gähn*[13:39:26] <Dr_Evil> Hallo stippi[13:39:39] <stippi> Dr_Evil: Yo[13:39:48] *** mes_ has joined #haiku[13:39:53] <JBurton> hey stippi[13:40:04] <stippi> What's up?[13:40:18] * Dr_Evil hat heute bis 13:30 geschlafen und ist jetzt erst richtig müde[13:40:20] <Teknomancer> stippi are u the WonderBrush programmer ?[13:40:22] <JBurton> nothing to see here, move on[13:40:23] <JBurton> :P[13:40:29] <JBurton> unlike from your side, stippi :P[13:40:54] <stippi> Teknomancer: Yes.[13:41:15] <Teknomancer> ah, ok[13:41:22] <stippi> JBurton: You're puzzling me...[13:41:33] <JBurton> stippi Painter[13:41:40] <stippi> JBurton: Yes.[13:41:41] <JBurton> you have news, unlike me[13:41:47] <stippi> JBurton: Ah[13:41:50] <JBurton> that's all :P[13:41:58] <stippi> JBurton: I'm working on it right now.[13:42:10] <JBurton> cool :P[13:42:39] <JBurton> are you working on it independently from the app server ?[13:42:40] <stippi> JBurton: Implementing BPolygon and BShape support.[13:42:46] <stippi> JBurton: Yes.[13:42:48] <JBurton> oh, neat[13:42:49] <JBurton> ok[13:43:06] <JBurton> so integrating it won't be your work[13:43:10] <stippi> JBurton: I want it to be easily usable as a drawing back end.[13:43:22] <JBurton> ok[13:43:26] <stippi> JBurton: I don't know that yet.[13:43:52] <stippi> JBurton: app_server is not my beast, I don't know all about it, and I think it will take me some time getting to know all the details.[13:44:10] <JBurton> makes sense[13:44:29] <stippi> JBurton: But DarkWyrm has been pretty quite lately, even though he asked me for my status before I posted to Haiku app_server list.[13:44:43] <stippi> JBurton: So I might like to help out in integrating it.[13:44:46] <JBurton> yeah, that's true. dead quiet[13:45:09] <stippi> JBurton: At least I will clean up my code and start commiting.[13:45:22] <JBurton> stippi I'd like to ask you... how does Agg do the drawing in WonderBrush ? does it draw to a bbitmap ?[13:45:31] <JBurton> stippi ok[13:45:43] <JBurton> you have write access, do you ?[13:45:49] <stippi> JBurton: yes, yes[13:45:56] <JBurton> :=)[13:46:09] <JBurton> I ask because it's amazingly fast and smooth[13:46:34] <stippi> JBurton: But AGG can draw to just anything in memory, supporting padding also.[13:46:48] <Teknomancer> whats AGG ? :)[13:46:55] <stippi> JBurton: I tend to disagree about it being smooth and fast.[13:47:01] <JBurton> stippi oh,why ?[13:47:09] <JBurton> that's how it looked[13:47:21] <stippi> Teknomancer: It's a high quality 2D drawing engine in C++ with a BSD style license.[13:47:22] <mes_> Teknomancer: anti grain geometry 2-d engine[13:47:47] <stippi> JBurton: Maybe with only a few objects...[13:48:02] <stippi> JBurton: But there is really lots of room for speed improvements.[13:48:04] <Teknomancer> stippi u use AGG instead of StrokeLine() etc of BView ??[13:48:07] <JBurton> well, good to know[13:48:51] <stippi> Teknomancer: You mean in WonderBrush? No, I use both. AGG for actual painting on the canvas, BView drawing for controls and manipulators.[13:49:40] <Teknomancer> stippi I asked for the canvas... ah so u use AGG? how come AGG works with BViews?? is it beos library ?[13:50:19] <JBurton> er, read a couple of lines above, Teknomancer :P[13:50:28] <Teknomancer> memory ?[13:50:29] <stippi> Teknomancer: Nope. It's a cross platform C++ lib, without any dependencies. I had to work hard to integrate it into WonderBrush and use it as a drawing back end.[13:50:39] <Teknomancer> oh[13:51:03] *** Fanskapet has quit IRC[13:51:08] <stippi> Teknomancer: Before I did that, I tested a couple of other 2D engines, Zodius, then libart.[13:51:43] <Teknomancer> btw, stippi u're toolbars are nice :) I used your style for a to-be-released app of mine ;-P[13:51:44] <stippi> Teknomancer: When trying to query for information about some problem I had with libart, I found out about agg, which is far superiour and C++.[13:51:48] *** [Beta] has quit IRC[15:15:26] *** [Beta] has joined #haiku[15:15:35] <Dr3w> Korli, whats cannaIM ?[15:18:09] *** Koki has joined #haiku[15:19:14] *** lizdeika_ has joined #haiku[15:19:21] <Dr3w> Korli, are you around?[15:23:07] *** lizdeika has quit IRC[15:23:47] <@Korli> Dr3w it's an input method addon[15:26:34] *** Potn has quit IRC[15:27:10] *** Potn has joined #haiku[15:30:42] *** LupusMichaelis has quit IRC[15:32:37] *** LupusMichaelis has joined #haiku[15:33:46] *** lechu_mys has joined #haiku[15:33:55] <lechu_mys> ;)[15:33:59] <lechu_mys> hi all '[15:34:00] <lechu_mys> ;][15:44:34] *** lizdeika_ has quit IRC[15:45:20] *** lizdeika has joined #haiku[15:50:10] <lechu_mys> :| :|[15:50:16] *** lechu_mys has quit IRC[15:50:28] *** ShackaN has quit IRC[15:54:52] *** Koki has quit IRC[16:11:23] *** lechu_mys has joined #haiku[16:11:56] *** stippi has quit IRC[16:12:12] *** lechu_mys has left #haiku[16:13:47] *** Fanskapet has joined #haiku[16:15:51] *** Kernel86 has joined #haiku[16:20:16] *** ShackaN has joined #haiku[16:46:22] *** tqh has quit IRC[16:46:53] *** mes_ has quit IRC[16:50:00] *** lizdeika has quit IRC[17:06:27] <Begasus> plop[17:07:36] <tic> plop[17:08:24] <Dr3w> splosh[17:08:53] <Begasus> still connected then :D[17:10:45] *** brennanOS has joined #haiku[17:16:09] *** TuneTracker has quit IRC[17:17:06] *** fyysik has joined #haiku[17:22:44] *** emwe has quit IRC[17:33:57] <fyysik> Jan 05 16:33:57 <fyysik>[17:34:41] <Dr3w> Jan 05 16:34:41 <Dr3w>[17:38:12] <w-ber> Jan 05 16:38:12 <w-ber>[17:40:37] <fyysik> Korli ?[17:41:28] <Dr3w> fyysik ?[17:42:12] <brennanOS> Dr3w?[17:42:32] <fyysik> are you Korli, Dr3w?[17:42:51] <Dr3w> no :) I just love the funny chat that goes on in here :)[17:42:59] <Dr3w> fyysik, he was here about 2 hours ago.[17:43:32] <@Korli> fyysik[17:43:41] <Dr3w> there you go :)[17:47:55] <Dr_Evil> Dr3w this is a international chat[17:49:32] <Dr3w> Dr_Evil, but of course it is![17:50:28] <Master199> i thought we are all from the kdl :P[17:51:51] <Dr_Evil> I don't stay in KDL longer than necessary[17:52:06] <Master199> hehe[17:54:05] *** mmadia has joined #haiku[17:54:14] <Master199> hi mmadia :-)[17:54:23] <mmadia> hi Master199[17:54:31] <mmadia> and all[17:54:38] <tic> hey matt[17:54:46] <fyysik> Korli - did you get those two screenshots ?[17:54:57] <mmadia> what is up mike?[17:55:18] <mmadia> http://bash.org/?433064 :)[17:55:28] <tic> mmadia, not much. coding.[17:55:43] <Dr3w> Can I see the screenshots!?![17:55:48] * Dr3w wants eye-candy...[17:56:15] <tic> screenies of what?[17:56:36] <@Korli> fyysik i got them, i'm surprised, crash happens in normal init code[17:56:42] <fyysik> Dr3w - are you developer? assembler guru? kernel hacker?[17:57:09] <fyysik> Korli - that one with crash in emuxki_inte_enable ?[17:57:31] <Dr_Evil> Korli we already analyzed it last night[17:57:47] <Dr3w> app developer, wanna-be kernel dev. Look at the commits for user mccall on the openbeos CVS.[17:57:49] *** brennanOS has quit IRC[17:57:52] <Dr3w> long time ago though.[17:58:22] <@Korli> Dr_Evil then ?[17:58:47] *** TLF has joined #haiku[17:59:31] <tic> moo[17:59:47] *** mmadia has quit IRC[18:01:34] <Dr_Evil> Korli wait[18:01:52] <fyysik> Dr_Evil suspected that not all variables/ponters used in interrupt handler are initialized for moment when interrupt happens first time, IIRC[18:03:44] <Dr_Evil> [02:19] <Dr_Evil> fyysik I *think* that the interrupt handler of the driver somehow clobbers the stack[18:03:44] <Dr_Evil> [02:20] <Dr_Evil> so that shortly after the interrupts are enabled, the first interrupts overwrites soemthing on the stack, and then the esp or the return address from stack get invalid[18:03:44] <Dr_Evil> [02:31] <Dr_Evil> fyysik well, the most likely source for this problem is:[18:03:44] <Dr_Evil> [02:32] <Dr_Evil> interrupt handler is installed without setting up every variable, that is used in interrupt handler, first.[18:03:45] <Dr_Evil> [02:32] <Dr_Evil> because of interrupt sharing, the handler can be executed before card interrupts are enabled[18:03:47] <Dr_Evil> [02:32] <Dr_Evil> second possible source:[18:03:49] <Dr_Evil> [02:32] <Dr_Evil> interrupts are enabled before every variable, that is used in interrupt handler, is initialized.[18:03:50] <fyysik> maybe SB Live CT4830 has some more interrupts to be disabled besides those 4 you disabling in emuxki_init ? Or some register with different from listed names requires to be set?[18:03:52] <Dr_Evil> [02:33] <Dr_Evil> in case of interrupt sharing, both problems can coexist[18:03:54] <Dr_Evil> [02:34] * geist curses interrupt sharing[18:03:56] <Dr_Evil> {02:46] <Dr_Evil> I'm not even sure if my driver initialize *everything* before installing interrupt handler, because ony assums that they are disabled. due to interrupt sharing, they can happen however[18:03:59] <Dr_Evil> [02:47] <Dr_Evil> that makes a nice magazin article[18:04:32] <Dr_Evil> fyysik as already said, disabling card interrupts doesn't help when the interrupt is shared (as it most likely is in a lot of systems)[18:04:49] <tic> Anyone here used the Marvell driver?[18:06:58] <@Korli> Dr_Evil the handler is installed after the crash point, no ?[18:07:17] <Dr_Evil> I didn't look at the source[18:07:59] <fyysik> Korli - we cannot be sure about real crash point, though[18:08:41] <fyysik> this screenshot represents only one of such crash cases...[18:09:04] *** thies has quit IRC[18:09:07] <@Korli> ah this is annoying[18:09:56] <@Korli> fyysik did you check with an empty interrupt handler as i suggested ?[18:10:01] <Dr_Evil> thats because an interrupt handler is corrupting something[18:10:54] <fyysik> Korli -IIRC you proposed to uncomment all TRACE statements in handler. But i can do test now with immediate return' in handler. No problem[18:11:04] <fyysik> setting laptop here to chat[18:11:13] <@Korli> have to go, bye[18:11:18] *** Korli has quit IRC[18:11:20] *** thies has joined #haiku[18:11:24] * tic listens to Rubicon[18:12:58] *** fyysik has quit IRC[18:15:45] *** fyysik has joined #haiku[18:16:40] <fyysik> Dr_Evil - crashed with empty interrupt handler returning B_HANDLED_INTERRUPT. Should i try UNHANDLED instead?[18:21:57] *** mmadia has joined #haiku[18:25:20] <Dr_Evil> yes[18:26:47] <fyysik> same s*t. changing compiler to 2.9, to see if results will be absolutely same[18:27:17] <fyysik> now it happened in log_printf in dump_hardware()[18:27:24] <Dr_Evil> fyysik can you lookup in Devices Preferences which devices share the IRQ? It's the "resource usage" (ALT+U) tab[18:27:53] <Dr_Evil> fyysik hello, wake up please. if it's causes by an interrupt, it will happen in any team at any place[18:28:16] <fyysik> Dr_Evil i said that several times. No device sharing here. So we cannot supect that reason:([18:28:58] <fyysik> Dr_Evil - yeah, but dump_hardware happened, as Korli said, before interrupt handler is installed[18:29:08] <Dr_Evil> well, than it's an hardware problem. if an empty interrupt handler still crashes, and there is no interrupt sharing, then we can stop now[18:29:12] *** AnEvilYak has quit IRC[18:29:24] <Dr_Evil> did you install any other drivers recently?[18:30:29] <fyysik> hmm, Dr_Evil - no, i didn;t and it happens in very different BeOS installs. One - my own distro from R5 and Bone7a and second - BeOS MAX based on PE3[18:31:00] <fyysik> and i don;t think that hardware is So wrong, as it don't crash with Be's drivers[18:31:29] <fyysik> PE 5.03 i meant[18:31:33] <Dr_Evil> make a picture of "resource use"[18:31:37] <Dr_Evil> or better[18:31:55] <Dr_Evil> run "poke" from command line, type "pci", press enter, and save output to a file[18:32:05] <fyysik> ok[18:32:39] *** AnEvilYak has joined #haiku[18:38:03] <fyysik> http:/beos.spb.ru/fyysik/fyysik_poke_pci.txt Dr_Evil[18:38:11] <fyysik> http://beos.spb.ru/fyysik/fyysik_poke_pci.txt Dr_Evil[18:39:22] <Dr_Evil> oh an Intel motherboard chipset :)[18:41:56] *** TLF has quit IRC[18:42:05] <Dr_Evil> ok thats the soundcard:[18:42:06] <Dr_Evil> bus 00 device 0c function 00: vendor 1102 device 0002 revision 08[18:42:38] <fyysik> yup[18:43:19] <Dr_Evil> ok no interrupt sharing[18:43:33] *** TLF has joined #haiku[18:43:41] <Dr_Evil> but there are two devices that have an uncomfigured interrupt[18:43:59] <fyysik> one is USB[18:44:03] <fyysik> probably[18:44:13] <fyysik> ik> i disabled interrupt for that in BIOS[18:44:51] <fyysik> secon is probably Radeon[18:45:06] <fyysik> disabled IRQ for VGA in BIOS[18:46:16] *** Dr3w has quit IRC[18:49:26] *** thaflo has quit IRC[18:52:45] *** Master199 has quit IRC[18:53:38] *** Begasus has quit IRC[18:54:18] <fyysik> Dr_Evil - may it be that we;re searching in wrong place and problem occurs in haiku multiaudio add-on? (Though, as notice, same addon works with Be's gamedriver)[18:59:07] *** ConneX has joined #haiku[19:11:38] *** RageMax has joined #haiku[19:13:56] *** ConneX has quit IRC[19:28:31] *** m_eiman has joined #haiku[19:36:52] *** Korli has joined #haiku[19:36:52] *** ChanServ sets mode: +o Korli[19:38:37] *** enthropy has joined #haiku[19:42:08] *** mmadia has quit IRC[19:58:12] *** dos4 has joined #haiku[20:01:02] <BeMonni> latest BeBits compiler doesn't seem to handle latest changes in input_server... gets a lot of warnings and some errors[20:03:15] <@Korli> BeMonni ??[20:03:19] *** AnEvilYak has quit IRC[20:03:53] <BeMonni> for example: warning: ANSI C++ forbids declaration `RkwCvtHan' with no type[20:04:32] <@Korli> it isn't input_server[20:04:49] <BeMonni> it's in one of the input methods...[20:05:03] *** Bernd has joined #haiku[20:05:14] <@AndrewBachmann> the asian input method?[20:05:14] <@voidref> does it have a return type?[20:05:20] *** Bernd has left #haiku[20:05:20] <@Korli> this addon is based on an external lib, hence the warnings[20:05:36] <BeMonni> voidref: don't think so... I try to check if it works with "int"[20:05:36] <@AndrewBachmann> sounds like a job for Jamfile flags[20:05:43] *** AnEvilYak has joined #haiku[20:05:46] <@AndrewBachmann> i.e. -w[20:05:49] <@voidref> you have to have return types[20:05:52] <@Korli> AndrewBachmann very likely[20:05:55] <@voidref> otherwise you are coding illegaly in c++[20:05:59] <@voidref> and stupidly in C![20:06:03] <dos4> please submit small patches to the mailinglist first BeMonni, many fixes in the large diff seems to be done in a wrong way[20:06:07] <tic> hey yakster.[20:06:19] <@Korli> voidref you can't even think how you're right[20:06:24] <@AndrewBachmann> we don't patch third party libs unless necessary[20:06:28] <@AndrewBachmann> removing warnings is not necessary[20:06:39] *** khorben has quit IRC[20:06:39] <BeMonni> dos4: I have whole tomorrow to make smaller patches... I just posted the link for big patch to get comments about how to modify the style...[20:06:48] <@voidref> warnings are the bane of competent programming[20:07:04] <@voidref> inane warnigns hide real ones[20:07:10] <@voidref> which causes hard to find bugs[20:07:14] <BeMonni> AndrewBachmann: I already found one bug because I eliminated few hundred warnings...[20:07:22] <@AndrewBachmann> if a warning bothers you, you supply a patch to the third party, and after they accept it, then we apply it[20:07:56] <dos4> instead of adding prototypes everywehre, mark functions as static if they don't need to be globally visible[20:08:11] <BeMonni> stupid Jam still builds almost whole tree even though I said to only recompile canna[20:09:01] <@Korli> BeMonni change dir to canna, then jam clean and jam[20:09:18] <BeMonni> Korli: Dr_Evil said "jam clean" is discouraged[20:09:30] <@AndrewBachmann> yeah, you don't need to jam clean[20:09:31] <@AndrewBachmann> just jam[20:09:37] *** dos4 is now known as Dr_Evil_[20:09:46] <Dr_Evil_> il_> i said it shouldn't be done that often[20:10:09] <@AndrewBachmann> if you are environment variables then you may want to jam clean[20:10:14] <@AndrewBachmann> er, changing env vars[20:10:20] <BeMonni> Dr_Evil_: It's the only way to make Jam output all warnings even from files that compile[20:10:23] * w-ber is an environment variable[20:10:50] * Procton unsets w-ber.[20:10:56] <w-ber> ...[20:11:22] <tic> del wber[20:11:26] * AnEvilYak uncoffees Procton[20:11:31] <tic> import operator; operator.del wber[20:11:35] <Procton> now that is impossible.[20:11:59] <tic> bleh, del is a keyword.[20:12:03] <tic> like print. I dislike magicness.[20:12:15] <AnEvilYak> Procton: don't bet on that ;p[20:12:40] *** MikeW has joined #haiku[20:12:46] *** Dr_Evil has quit IRC[20:13:16] <@Korli> ""jam clean" is discouraged", what's this ?[20:14:48] <tic> jam peanutbutter[20:15:04] *** khorben has joined #haiku[20:15:11] <AnEvilYak> jam tuna[20:15:20] <tic> hrm. that works, but not with jam, though?[20:15:43] <@AndrewBachmann> it's just not necessary Korli[20:16:12] <@Korli> AndrewBachmann then how do you rebuild the local dir only ?[20:16:19] <@AndrewBachmann> jam[20:16:26] <BeMonni> lol... who cares about what's necessary when building alpha/beta code... every removed bug is less pain in the a$$ later on[20:16:45] <@Korli> AndrewBachmann jam doesn't rebuild anything you know[20:17:01] <@AndrewBachmann> no, it rebuilds what you changed[20:17:15] <BeMonni> sometimes it'[20:17:19] <@Korli> changes in Jamfile aren't seen you know[20:17:37] <BeMonni> sometimes it's necessary to rebuild even unchanged files except if you set to treat warnings as errors[20:17:38] <@AndrewBachmann> ok, but it's not usual to change the Jamfile[20:17:53] <@Korli> AndrewBachmann this is the case we're talking about[20:18:04] <fyysik> Korli - tried to return from emuxki_int immediately, with both versions - B_UNHANDLED_INTERRUPT and B_HANDLED. Crashes:([20:18:09] <@Korli> orli> or if i change my gcc[20:18:28] <@AndrewBachmann> BeMonni did you change the jamfile?[20:19:17] <BeMonni> AndrewBachmann: Don't even remember what files I change... that's why I usually do big diffs... to refresh my memory what files I have changed... Don't want to wait days for individual directory diffs to finish[20:19:19] <@Korli> fyysik i don't understand what's happening, crashes at which point ?[20:19:37] <Dr_Evil_> using "jam -a target" is most times much better than jam clean[20:19:51] <fyysik> lat timed it crashed in emuxki_setup, last name in sc was read_io_16[20:19:56] <fyysik> last time[20:20:03] <@Korli> BeMonni i meant "don't do jam -a target" because it would rebuild every dependency[20:20:23] <@AndrewBachmann> lol[20:20:52] <@Korli> unless you want to rebuild libbe.so for a change in your little addon[20:20:57] <BeMonni> ok... do any of you want me to make individual directory patch for any of the changes I did near new year?[20:22:09] * Dr_Evil_ would reject most changes from the big diff[20:22:15] <BeMonni> before I tell jam to start building whole tree again...[20:22:34] <BeMonni> Dr_Evil_: That's why I asked if you want just the ones you would accept ;)[20:22:53] <BeMonni> Dr_Evil_: And which ones ;)[20:22:59] <Dr_Evil_> thats the reason why I asked early to stop and instead start sending small patches to the mailinglist[20:23:15] <BeMonni> Dr_Evil_: Everytime one of the partial diffs is commited the big one gets smaller and easier to find clean parts ;)[20:23:30] <BeMonni> Dr_Evil_: I can't send mail on this machine, don't have BeMail installed ;)[20:25:13] <BeMonni> Dr_Evil_: And I really dislike mailinglists because some of the software mangle the diff files...[20:25:50] <Dr_Evil_> your are wasting time[20:25:54] <@Korli> fyysik maybe beos crashs when no interrupt handler is installed for an interrupt[20:26:11] <@Korli> bye[20:26:26] *** Korli has quit IRC[20:26:43] <BeMonni> Dr_Evil_: Well... In the end I'm not wasting time if I save few days when you others fix issues I have already resolved and I can finally dump Windows and switch to Haiku for rest of my life...[20:28:31] <fyysik> now real handler installed, crash in emuxki_setup last sc name write_io_16. I will continue hacking, but now i seems found reason fro troubles with Bezilla/FF and gcc 2.953. Should check that too this night[20:28:37] <BeMonni> Dr_Evil_: Actually one thing that is good with "redundant" function prototypes is when parameter count is wrong... compiler detects such issues ;)[20:31:05] *** thaflo has joined #haiku[20:38:48] *** enthropy has quit IRC[20:40:29] <fyysik> first write_io_16 op in emuxki_setup is reset followed with snooze[20:41:21] <fyysik> mmadia[20:41:56] *** AtomoZero has joined #haiku[21:00:39] *** TuneTracker has joined #haiku[21:03:37] *** MikeW has quit IRC[21:05:25] *** Karina`` has quit IRC[21:13:44] <Fanskapet> http://www.grandtheftendo.com/ :)[21:13:50] <Fanskapet> GTA for NES 8bit :)[21:13:55] <Fanskapet> pretty neat :)[21:19:23] *** stippi has joined #haiku[21:19:28] <stippi> yo[21:21:53] <Fanskapet> alloah[21:23:10] <fyysik> Dr_Evil_ what do you think if joystick port at soundcard also requires separate interrupt handling?[21:23:24] *** Karina`` has joined #haiku[21:30:54] <fyysik> no crash, so got emuxki.log this time[21:30:55] *** AtomoZero has quit IRC[21:34:10] *** AtomoZero has joined #haiku[21:58:21] *** brennanOS has joined #haiku[22:02:54] *** Dr3w has joined #haiku[22:02:56] *** mmu_man has joined #haiku[22:02:57] *** ChanServ sets mode: +o mmu_man[22:03:15] <@AndrewBachmann> hello mmu_man[22:03:35] <@mmu_man> plop[22:04:32] *** oco has joined #haiku[22:05:05] *** m_eiman has quit IRC[22:05:36] *** BGA has joined #haiku[22:05:36] *** ChanServ sets mode: +o BGA[22:05:38] <@mmu_man> anyone interested in FATX support in dosfs ?[22:05:40] <@mmu_man> =)[22:05:42] <AnEvilYak> FATX?[22:05:55] <@mmu_man> the XBox version[22:05:55] * AnEvilYak troutslaps BGA[22:06:12] <AnEvilYak> mmu_man: didn't know it used a modified FAT[22:07:25] * fyysik thinks that even WinFS will use FAT[22:07:32] <fyysik> FAT8 or so[22:07:47] <AnEvilYak> you realize winfs is a layer on top of the filesystem right?[22:07:50] <AnEvilYak> it's not the FS itself.[22:07:55] <AnEvilYak> it's more like what Apple does with Spotlight[22:08:03] <AnEvilYak> where the db/indexing is separated from the fs impl.[22:08:12] <@BGA> AnEvilYak![22:08:15] <AnEvilYak> BGA![22:08:22] <tic> boing boing[22:08:31] <tic> AnEvilYak, that's suckiness[22:08:36] <AnEvilYak> tic: yes and no.[22:08:44] <AnEvilYak> tic: the BFS style approach has major issues when it comes to multiuser.[22:10:24] <tic> true.[22:11:25] <@mmu_man> actually by reading Dominics book it looks like NTFS has quite a lot of features BFS has[22:11:35] <@mmu_man> like attributes and indices[22:11:39] <@mmu_man> just noone uses them[22:11:53] <AnEvilYak> mmu_man: NTFS only has 64KB for attrs though[22:11:53] <@AndrewBachmann> yes, the NTFS file system should store the mimetype in a native attribute :-)[22:12:01] <AnEvilYak> mmu_man: just like XFS[22:12:13] <@mmu_man> AnEvilYak yeah well it's enough for many things we currently do[22:12:21] <AnEvilYak> mmu_man: true.[22:12:23] <@mmu_man> BFS currently only supports 4GB /attr anyway =)[22:12:32] <@mmu_man> (due to not supporting the offset value)[22:12:32] <AnEvilYak> mmu_man: does it?[22:12:36] <AnEvilYak> ah.[22:12:50] <fyysik> AnEvilYak - nobody knows. Till now WinFS is fake, it is MS to decide if it will be at all and how to do it:) Hype atm[22:12:54] <Dr3w> What are the BFS multiuser issues?[22:13:03] <@mmu_man> Dr3w some on indices[22:13:14] <AnEvilYak> Dr3w: indices store inode refs directly[22:13:20] <AnEvilYak> which allows you to completely bypass permissions[22:13:23] <@mmu_man> even though indices belong to a user and other users can create indices as well[22:13:29] *** konrad has joined #haiku[22:13:31] <@mmu_man> AnEvilYak not exactly[22:13:31] <Dr3w> Ooo doesn't sound good.[22:13:34] <AnEvilYak> mmu_man: close enough.[22:13:44] <AnEvilYak> mmu_man: that's basically how jbq explained it anyhow[22:13:53] <@mmu_man> it's letting ppl open vnodes by entry_refs which is dangerous[22:14:14] <@mmu_man> even can get the kernel to open a bad vnode and crash[22:14:31] <@mmu_man> indices themselves are just special directories[22:15:12] <AnEvilYak> right but it was my understanding that the indices key/value pairs use inode numbers as the value[22:15:27] <tic> B_FIX ;)[22:15:31] <AnEvilYak> so as to have a persistent way of locating the file that doesn't change with rename/move ops[22:15:43] <@mmu_man> now because BFS must reconstruct the paths for queries it stores teh parent vnid in the inodes[22:15:51] <AnEvilYak> right.[22:16:00] <@mmu_man> (which is also why supporting hardlinks is not possible)[22:16:31] <@mmu_man> so _kopen_vnode() could just parse teh path up, and access() it[22:16:39] <@mmu_man> just will slow things down[22:16:52] <@mmu_man> still vnodes have unix perms for themselves in[22:17:30] <@mmu_man> the only prob is to a) make sure parents allow access (say /boot is --------- and /boot/foo is rwxrwxrwx it should still be forbidden to access /boot/foo)[22:17:55] <@mmu_man> b) make sure the vnode opened is valid (= pat of the file system)[22:18:11] <@mmu_man> and not a fake data with valid magic numbers :)[22:18:19] <@AndrewBachmann> mmu_man in my experience the parent permissions don't matter[22:18:28] <AnEvilYak> parent permissions aren't supposed to matter[22:18:31] <@mmu_man> that should be possible also by reconstructing paths (but being very careful since any data can be wrong)[22:18:43] <AnEvilYak> for example, a user can create a subdir in their home dir that you have permissions to[22:18:46] <@mmu_man> AndrewBachmann they might not in bfs but they must in POSIX[22:18:47] <AnEvilYak> and it will let you access that subdir[22:18:50] <AnEvilYak> but not the rest of home[22:18:57] <AnEvilYak> (on linux)[22:18:58] <@AndrewBachmann> I don't think that's right mmu_man, AnEvilYak has it[22:19:14] <@AndrewBachmann> this is my experience on other unix[22:19:14] <AnEvilYak> AndrewBachmann: I know it is, it's how I access my area of my brother's web server.[22:19:24] <AnEvilYak> also known as "Bypass Traverse Checking"[22:19:26] <AnEvilYak> which NT does as well[22:19:34] <@mmu_man> at least the direct parent perms must be checked before removing the vnode[22:19:44] <@mmu_man> NT has ACLs[22:19:54] *** Dr_Rock has joined #haiku[22:19:56] <AnEvilYak> mmu_man: well, yeah, for remove certainly[22:19:58] <AnEvilYak> mmu_man: but not for access[22:20:05] <AnEvilYak> mmu_man: since remove would be an operation on the parent[22:20:13] <AnEvilYak> (updating the directory)[22:20:25] <@mmu_man> still ensuring the vnode is part of teh fs imposes making sure it points back up to the root vnid[22:20:37] <konrad> mmu_man is thats why NT is so much better than Unix? (ACLs)?[22:20:41] *** Dr_Rock has quit IRC[22:20:45] <@mmu_man> AnEvilYak hmm right unlink takes the parent vnid[22:20:53] <@mmu_man> konrad certainly not[22:20:55] <@mmu_man> ACL SUXOR[22:20:56] <ShackaN> ACL r0ck![22:21:04] <konrad> mmu_man ah I guess its another feature then =)[22:21:12] <tic> bah. it isn't.[22:21:26] <ShackaN> /etc/shadow sux0r ![22:21:28] <ShackaN> :)[22:24:05] <AnEvilYak> I like ACLs :([22:24:36] <tic> two opposing wills; who will be the winner of this game![22:25:00] <@AndrewBachmann> tune in next week[22:26:00] *** Dr3w has quit IRC[22:26:02] <tic> for more Haiku action![22:27:18] *** AtomoZero has quit IRC[22:28:01] *** AtomoZero has joined #haiku[22:28:44] <@mmu_man> <Batman> ... But the worst... is just yet to come !!![22:29:09] <@mmu_man> Tune in tomorrow, same time, same channel !! </Batman>[22:29:22] *** Greyhawk_Dragon has joined #haiku[22:29:31] <ShackaN> what are we gonna see tomorrow?[22:29:48] <tic> POW! KA-BLAM![22:30:19] <konrad> ShackaN maybe we all should read http://www.microsoft.com/windowsserversystem/facts/migrate.mspx and see what we do tomorrow?[22:30:55] <ShackaN> oh[22:31:05] <ShackaN> that's what I'm going to do! SURE![22:31:41] <ShackaN> "Get a free windows vs. linux evaluation kit"!?!?[22:31:43] <ShackaN> wtf!?[22:31:49] <konrad> Hmm Iam intrested in the "outperforms Linux"[22:32:35] <ShackaN> what I'd like to see there one day "outperforms linux, but can't catch on with haiku"[22:32:36] <ShackaN> :D[22:32:41] <konrad> Hehe[22:32:48] <konrad> That would be nice[22:33:07] <tic> ugh, a server OS that can't be remotely controlled through an ssh session is useless.[22:33:21] <konrad> For you it might yes[22:33:59] <tic> fat chance of an admin doing a web server restart from my PDA through a GPRS connection on Windows in case it goes down and the dude is out of office.[22:34:37] <konrad> He never has to restart it in the first place, he is using windows, it just workes[22:34:53] <konrad> tic dont you ever watch tv?[22:35:03] <tic> hahahaha[22:35:04] <tic> Right.[22:35:15] <tic> that's why all Windows 'puters get virii and whatnot?[22:35:23] <tic> windows on a server my ass.[22:36:30] <konrad> .... is like placing the server on the desktop (linux)[22:36:46] <@mmu_man> ShackaN sure NT outperforms Linux[22:36:53] <@mmu_man> just need to see which box you run each[22:37:15] <@mmu_man> I can make a samba server on a PII[22:37:23] <@mmu_man> XP needs at least a P4[22:37:28] <ShackaN> :)[22:37:35] <@mmu_man> of course it will outperform the PII[22:37:42] <@mmu_man> except if I put linux on the P4 as well =)[22:38:40] <ShackaN> uhm, dilbert.com I so sloooow today...[22:39:00] <konrad> mmu_man http://linuxtoday.com/news_story.php3?ltsn=2003-05-08-019-26-NW-MS-RH&tbovrmode=1[22:39:50] <AnEvilYak> ShackaN: loaded fine here[22:40:37] <ShackaN> don't know there, but I have a week stripes worth to read and it's taking ages :)[22:42:23] <@mmu_man> yeah I can do a bench too[22:42:28] <@mmu_man> Linux: $0[22:42:32] <@mmu_man> Samba: $0[22:42:38] <@mmu_man> XP: too expensive[22:42:39] *** thaflo has quit IRC[22:42:52] <@mmu_man> Linux+Samba = x sec/MB[22:43:00] <@mmu_man> XP = NAN sec/MB[22:43:58] <@mmu_man> not sure I understood today's dilbert[22:44:06] <@mmu_man> EASLEEP anyway[22:46:41] <ShackaN> mmu_man, I'm not a today yet[22:46:45] <ShackaN> *at[22:47:13] <konrad> night, all[22:47:16] *** konrad has quit IRC[22:49:36] <@AndrewBachmann> mmu_man link?[22:51:33] <tic> o/~ I must, I must, I must increase my bust! o/~[22:51:49] <@AndrewBachmann> ?[22:51:59] <AnEvilYak> hahahaha.[22:52:07] <AnEvilYak> http://www.dilbert.com/ Andrew[22:52:12] <@mmu_man> ENOSYS[22:52:24] <@mmu_man> (hardlinks not supported on BFS)[22:52:26] <@mmu_man> :p[22:53:07] <@AndrewBachmann> is this the dilbert you are referring to mmu_man: http://www.dilbert.com/[22:53:14] <@mmu_man> yup[22:53:31] <@AndrewBachmann> he implies that she can't be the receptionist because she is too ugly[22:54:24] *** Korli has joined #haiku[22:54:24] *** ChanServ sets mode: +o Korli[22:54:50] <tic> Grr, I don't like broken views.[22:56:32] <ShackaN> nobody likes broken stuff[22:57:03] <AnEvilYak> ShackaN: except for people who like to fix broken stuff ;p[22:57:51] <ShackaN> you mean m$ engineers write broken code for fun ?[22:58:30] <tic> the dnd stuff is slightly broken in BColumnListView[22:58:45] <ShackaN> that could be an explanation why I'm seeing new windows exploits popping up every week[22:59:53] <tic> nope, listen to konrad. windows is good.[23:00:06] <AnEvilYak> he's no longer around to be listened to[23:00:08] <AnEvilYak> so his spell is broken ;p[23:00:49] <tic> aww, darnit.[23:00:53] * tic likey windows[23:01:18] <tic> at least, those that are instances of BWindow! :)[23:02:34] *** Greyhawk_Dragon has quit IRC[23:03:17] <ShackaN> wouldn't it be nice to have Qt in haiku instead of Beos native windowing mechanism ?[23:03:34] <@AndrewBachmann> mmu_man she says she's not the receptionist because she's was appointed "directory of initial client contact" or something like that[23:03:42] <@AndrewBachmann> director, rather[23:04:06] <@Korli> fyysik[23:04:11] <fyysik> me[23:04:20] <@Korli> what's up ?[23:04:21] <fyysik> got once emuxki.log[23:04:30] <fyysik> when it didn't crash[23:04:46] <fyysik> dunno if it may be useful for something[23:04:51] <@Korli> rather annoying[23:05:42] <fyysik> Korli - is there any importance that SB card has also joystick at it (so io range) ?[23:05:54] <fyysik> joystick port[23:06:35] <@Korli> midi port is present on most sb live i think[23:07:31] <@Korli> maybe you can try to enable interrupts after installing the handler ?[23:08:54] <fyysik> iirc i tried it..in vain. and now it crashes somewhere in emuxki_setup after write_io_32.[23:09:02] <fyysik> so before handler install[23:09:42] <Dr_Evil_> esp again invalid fyysik?[23:10:09] <fyysik> Korli - is it standard practice, to read nabmbar fro, base_registers[o] or is there another way?[23:10:52] <fyysik> Dr_Evil_ it seems that everytime i got crash in emuxki_setup or emuxki_init, esp is wrong[23:12:07] <Dr_Evil_> geist question: when interrupts happen in BeOS, what stack is used by the interrupt handler?[23:12:19] <[Beta]> mmu_man: re: FATX ... is there really any point in supporting that as of now ?[23:12:38] <AnEvilYak> so you can install Haiku on X-Box? ;p[23:12:44] <[Beta]> in the future, it would be nice; Haiku on xbox \o/[23:13:15] <@Korli> fyysik did you try to call debugger() in emuxki_setup ?[23:13:48] <@AndrewBachmann> Dr_Evil_ you can alloc your own stack if you want[23:14:00] <fyysik> explicitly? dunno how. Only tried to set some panic() calls to check pointers[23:14:27] <fyysik> Korli btw, maybe my sources are too old?[23:14:31] <@Korli> debugger("emuxki_setup 1");[23:14:41] <@Korli> fyysik cvs is current[23:14:51] <fyysik> all files are dated by 32003[23:14:56] <fyysik> 2003[23:15:03] <fyysik> November[23:15:24] <@AndrewBachmann> there is set_signal_stack and sigaltstack[23:16:00] <@Korli> fyysik that's right[23:16:38] <fyysik> Korli - so, first line in emxuki_setup after variable declaration should be that debugger() ?[23:17:21] <@Korli> fyysik you can call it several times, you should just change the string passed to know where you are in the code[23:17:39] <fyysik> Korli - and there is little problem without importance, but anyway. include <midi.h> instead "midi.h" or something alike[23:17:41] <@Korli> type 'c' to continue when in debugger[23:18:07] <fyysik> ok[23:18:36] <@Korli> fyysik with our build system, headers search work this way[23:19:55] *** tqh has joined #haiku[23:22:46] <fyysik> never worked with debugger syntax in BeOS. I suspect i can look at variable values there?[23:24:16] <fyysik> should copy sources at this machine[23:24:19] <tic> hrm, I need to learn the evaluation order for boolean statements...[23:24:40] *** TLF has quit IRC[23:24:50] <ShackaN> for boolean expressions the evalutation order is not relevant..[23:24:52] <Dr_Evil_> if (a & 4 || a & 8) works ;)[23:25:11] <Dr_Evil_> but you never know how the compiler does it[23:25:43] <AnEvilYak> unless you know the operator precedence rules[23:25:50] <AnEvilYak> and assume the compiler follows them[23:26:02] <ShackaN> the compiler should follow them...[23:26:05] <AnEvilYak> iirc in C the bit ops generally have the higheest priority[23:26:09] <ShackaN> yes[23:26:09] <AnEvilYak> ShackaN: keyword: "should"[23:26:24] <ShackaN> first !, then AND then OR[23:26:32] <AnEvilYak> I thought ^ was higher[23:26:51] <Dr_Evil_> AnEvilYak i wasn't talking about operator precedence[23:26:52] <ShackaN> I didn't mention ^[23:26:58] <AnEvilYak> but I'm rusty, and I usually use parens anyhow.[23:27:03] <AnEvilYak> Dr_Evil_: what then?[23:27:08] *** oco has quit IRC[23:27:38] <AnEvilYak> Dr_Evil_: I mean, theoretically the compiler could condense that into a single & ...[23:27:41] <Dr_Evil_> how the compiler implements the test[23:27:45] <Dr_Evil_> yes[23:27:58] <tqh> fyysik should cvs-version of widget and gfx work without patches on trunk? Firefox1.1 seems to have problems launching but don't give good reason why.[23:28:00] <AnEvilYak> depending on how smart the optimizer is[23:28:10] <@geist> Dr_Evil_: should be the kernel stack for the current thread[23:28:21] <@geist> I think it used the pretty standard way of using the current ring0 stack[23:28:28] <@geist> I *think* they were 8k in size[23:28:28] <ShackaN> uhm[23:28:42] <Dr_Evil_> geist so each user thread has a kernel stack. ok[23:28:44] <ShackaN> geist, so the interrupt handler gets run in a random stack ?[23:28:52] <@geist> that's correct[23:28:59] <ShackaN> didn't know that..[23:29:05] <tic> status = acquire_sem(); if appRunning and status == OK:, as opposed to: if appRunning and acquire_sem() = OK.[23:29:06] <@geist> and since it might be in kernel space at the time, some of the stack may already be in use[23:29:12] <tic> inside a while loop, only the boolean would get evaluated.[23:29:19] <tic> (is that what's called short-circuiting?)[23:29:32] <@geist> that's a pretty standard way to do it, especially on x86. it's pretty tough to do anything else due to the way the interrupt mechanism works[23:30:05] <ShackaN> <geist> I *think* they were 8k in size << doesn't stack space grow as needed ?[23:30:09] <Dr_Evil_> geist ok, thanks. I know you already told me this once, but I didn't remember[23:30:19] <ShackaN> uhm, wait, not in an interrupt handler...[23:30:38] <AnEvilYak> ShackaN: stack size is generally fixed afaik.[23:30:41] <ShackaN> geist, if you run out of stack in a IH, what happens ?[23:31:02] <ShackaN> AnEvilYak, I read that in most OSes it is not[23:31:07] *** Lebuzzer has joined #haiku[23:31:20] <@geist> ShackaN: the system dies[23:31:28] <AnEvilYak> ShackaN: well, it's easier to have an expandable stack in userspace than in the kern[23:31:32] <@geist> the kernel stack does not grow[23:31:46] <@geist> it's fixed in size, and preallocated, which is why it should be as small as possible[23:31:53] <tic> o/~ Ich hab' getraumt ich kan fliege wie hoch wie ich will o/~[23:31:56] <ShackaN> ok[23:32:19] <ShackaN> tic, did you have a Kafkian experience ? :D[23:32:59] <tic> ShackaN, nah, I'm just listening to Blümchen.[23:33:04] <tic> I'm not a bug. :P[23:33:12] <fyysik> ok, now system stucks in find_free_pte[23:33:14] <ShackaN> :D[23:33:28] <ShackaN> Benjamin Bluemchen?[23:34:06] <ShackaN> stupid question, beos is reentrant, right?[23:34:31] <Dr_Evil_> yes[23:34:35] <ShackaN> ok[23:34:40] <ShackaN> I wasn't sure[23:34:54] <@Korli> ShackaN what does this mean ?[23:35:21] <ShackaN> basically.. that more kernel threads can be in execution "at the same time"[23:35:33] <ShackaN> (iirc)[23:35:46] <AnEvilYak> re-entrancy generally implies that the kernel itself is preemptible[23:35:47] <@Korli> i> i didn't know this meaning[23:35:48] <AnEvilYak> iirc[23:35:59] <AnEvilYak> at least in this context.[23:36:38] <@geist> yes, that's all it means[23:36:40] <AnEvilYak> in userspace it usually refers to several threads being able to make the same call at the same time[23:36:46] <AnEvilYak> i.e. the _r versions of various syscalls[23:36:49] <@geist> basically kernel code works pretty much like user code[23:36:58] <@geist> you can block on semaphores, can be interrupted by interrupts, etc[23:37:00] <ShackaN> yes, the cpu can interrupt a thread in kernel mode to execute another :) that's what I said, isn't it ?[23:37:29] <AnEvilYak> ShackaN: the interrupting thread doesn't have to be a kern thread though[23:37:33] <@geist> only very deep kernel code is non reentrant. ie, if you disbale interrupts and acquire a spinlock you're basically entering a critical section[23:37:36] <ShackaN> kaN> k[23:37:52] <@mmu_man> the kernel is mostly reentrant because it is preemptible[23:38:09] <@geist> yeah that's a better way to say it[23:38:13] <ShackaN> do you have the posix synchronization stuff in haiku ?[23:38:16] <@geist> the kernel is preemptible[23:38:20] <@geist> and thus reentrant[23:38:24] <@mmu_man> classical UNIX kernels have a big lock that makes sure only one thing at once is done in the kernel[23:38:25] <Dr_Evil_> the kernel is preemptible because it is reentrant[23:38:30] *** ConneX has joined #haiku[23:38:39] <@mmu_man> and it is never interrupted until that thing is done and teh syscall exist[23:38:43] <@mmu_man> exits[23:38:50] <@geist> right[23:38:58] <@mmu_man> or the call itself uses (un)interruptible_wait_on()[23:39:00] *** Zenton has joined #haiku[23:39:08] <@geist> well, more like the classic unix kernel just disables interrupts while in the kernel always[23:39:22] <@geist> and with multiple cpus they had to cram a big lock on it to keep multiple cpus out of it[23:39:27] <@mmu_man> yup[23:39:39] <@geist> on a single proc machine it's actually pretty efficient[23:39:43] <ShackaN> and that's fscking slooow :D[23:39:54] <@geist> since kernel code doesn't have to deal with locking, the code is simpler[23:40:03] <ShackaN> uhm[23:40:03] <@geist> however interrupt latency is pretty high[23:40:19] <@geist> on a non-interactive system that's not a problem at all[23:40:25] <@mmu_man> like ohhh I can't hear the music when moving that window in X11[23:40:31] <ShackaN> hahahha[23:40:41] <@mmu_man> Linux was like that until recently[23:40:47] <@mmu_man> *BSD too[23:40:50] <ShackaN> wow[23:40:52] <@geist> it's the classic "throughput vs responsiveness" argument[23:40:57] <@mmu_man> yeah[23:41:03] <@geist> adding locking and being reentrant is less efficient[23:41:06] <@geist> but more responsive[23:41:10] <ShackaN> I'm glad I use linux just since last year :)[23:41:14] <@mmu_man> and more complex == more bugprone[23:42:08] <@mmu_man> ok, I should go to bed as sleep isn't a preemptible call I'm afraid[23:42:16] <ShackaN> :)[23:42:19] <ShackaN> bonne nuit[23:42:58] <AnEvilYak> mmu_man: what if we make you a zombie process?[23:45:23] <w-ber> AnEvilYak: how would that work in this case? Kill his parents?[23:45:33] <AnEvilYak> w-ber: I dunno[23:45:40] <AnEvilYak> "You shot zombie flanders!" "He was a zombie?"[23:45:52] <tic> This is a job far Superdoo![23:51:00] *** Sikosis has joined #haiku[23:51:00] *** ChanServ sets mode: +o Sikosis[23:53:35] *** Sikosis has quit IRC[23:57:11] *** mmu_man has quit IRC