T2 IRC Log: 2005-01-08
This is the log as captured by an IRC bot in the channel. The statements are those of the individual people and might not neccessarily reflect the policy and legal rules as set forth by the T2 SDE Project.
--- Log opened Sat Jan 08 00:00:46 2005
00:02 < mnemoc> me?
00:02 < mnemoc> or we*
00:02 < rxr> no - people that have to use it or more in general ...
00:02 < mnemoc> *g*
00:03 < rxr> and you probably because you have to listen to my nonsense ...
00:04 < rxr> 1370 builds total, 666 completed fine, 3 with errors.
00:04 < mnemoc> your nonsense makes my days shorter
00:09 < rxr> ouhm
00:10 < rxr> then I better stop posting it ...
00:10 < rxr> (leaving time to commit - e.g. xmule fixes ;-)
00:11 < mnemoc> it was a 'good thing', bored days feel looooooong
00:12 < mnemoc> thunderbird crashed again :(
00:12 < rxr> oh
00:12 < rxr> I thought you mean I take away time to do real work
00:13 < mnemoc> :)
00:14 < rxr> so - tux.org reply send
00:14 < rxr> I hope we can mirror some files they
00:14 < rxr> but it seems they have not much space to spare ...
00:14 < mnemoc> :(
00:14 < rxr> http://www.tux.org to make my log parser happy and google rate us a slightly higher
00:15 -!- valentin [~valentin@port-212-202-170-162.dynamic.qsc.de] has quit [Read error: 104 (Connection reset by peer)]
00:15 < rxr> huh????
00:15 < rxr> .oO
00:16 -!- valentin [~valentin@port-212-202-170-162.dynamic.qsc.de] has joined #t2
00:17 < mnemoc> huh what?
00:18 < rxr> that valentin ping-ponged ...
00:19 < rxr> but it was not our server
00:19 < rxr> just his client ...
00:19 < rxr> hm - I do not believe this
00:19 < valentin> what did i do ?
00:19 < rxr> hung your irssi ...
00:19 < valentin> yes - but what is 'huh what' ?
00:20 < rxr> 00:15 -!- valentin [~valentin@port-212-202-170-162.dynamic.qsc.de] has quit
00:20 < rxr> [Read error: 104 (Connection reset by peer)]
00:20 < rxr> 00:15 < rxr> huh????
00:20 < rxr> 00:15 < rxr> .oO
00:20 < valentin> ok
00:20 < rxr> this code change:
00:21 < rxr> - // retrieved = -666;
00:21 < valentin> maybe irssi does not like to run for so long
00:21 < rxr> + retrieved = -666;
00:21 < rxr> causes some illegal instruction somewhere - a place the debugger is not willing to yield ...
00:21 < rxr> rw_completed: ret_io_size: -666, io_size: 96
00:21 < rxr> Illegal instruction
00:22 < valentin> what is the next use of retrieved after that initialisation ?
00:22 < rxr> the pointer to that value is passed to WritePipeAsync
00:22 < valentin> is retrieved an instantce variable of some object ?
00:22 < rxr> ReadPipeAsync even
00:23 < rxr> this is the plain C USB wrapper - nothing with objects ...
00:23 < rxr> no this "objects where one never knows where they are" ;-)
00:23 < rxr> it is thown into the library function like this:
00:24 < rxr> (void *)&retrieved);
00:24 < rxr> and arrives there:
00:24 < rxr> rw_completed: ret_io_size: -666, io_size: 96
00:24 < valentin> void* ???
00:24 < rxr> yeah - custom data passed to the callback ...
00:24 < valentin> ok
00:24 < rxr> fprintf(stderr, "rw_completed: ret_io_size: %d, io_size: %d\n", *(u_int32_t *)ret_io_size, (u_int32_t)io_size);
00:24 < rxr> static void rw_completed(void *ret_io_size, io_return_t result, void *io_size)
00:25 < rxr> I only see one place that could crash - if it is ther (I just try to find out) the compiler seems to be generating invalid code ...)
00:26 < rxr> hm - no
00:26 < rxr> it is even thereafter ...
00:26 < rxr> strange
00:26 * rxr puzzled
00:27 * rxr can not imagine what some system library would do to the custom data pointer that could illegal instruction that way ...
00:27 < valentin> try retrieved = @opcode of some legal instruction
00:27 < rxr> hm - or - wait
00:27 < rxr> it is 0x0002
00:27 < rxr> looks quite random ..
00:27 < rxr> well - better the crash dialog claims it is 0002
00:27 < valentin> and it is realy the one line ?
00:27 < rxr> neither the crash helper wants to show me a backtrace - neither the debugger some memory address
00:28 < rxr> yes
00:28 < rxr> when I remove the assignment it works - as before - with our sporadic errors ...
00:28 < valentin> and it crashes imediatly after the '=' ?
00:28 < rxr> later
00:28 < rxr> and I just verified somewhere after the callback returns
00:29 < valentin> i would guess that parameter causes something in that function to overwrite memory
00:29 < valentin> can you read out the address at startup ?
00:29 < rxr> I posted a line above where it is read out ...
00:30 < rxr> it reads out - and is properly assigned:
00:30 < rxr> rw_completed: ret_io_size: -666, io_size: 96
00:30 < rxr> rw_completed: ret_io_size: 96, io_size: 96
00:30 < rxr> rw_completed: RETURNINGIllegal instruction
00:30 < valentin> the address where the instruction is
00:30 < valentin> the crash pointer
00:30 < rxr> come over and show me how to get osx to yield that info ...
00:30 < rxr> I failed so far ...
00:31 < rxr> the codex debugger is not interested in tracing the callback ...
00:31 < rxr> xcode even ...
00:31 < valentin> isn't it in the crashlog ?
00:31 < rxr> no
00:31 < rxr> it does not want to show the backtrace for the affecte thread
00:31 < rxr> I ranted about this 30 minutes or so ago
00:32 < rxr> you are aware that we dump this into #t2 ?
00:33 < valentin> i get logs for each backtrace
00:33 < valentin> yes i am
00:33 < valentin> will raise our google hits
00:34 < rxr> I get a log
00:34 < valentin> but you are not allowed to paste explicit code :)
00:34 < valentin> Exception: EXC_BAD_ACCESS (0x0001)
00:34 < valentin> Codes: KERN_INVALID_ADDRESS (0x0001) at 0x0117ed4b
00:34 < valentin> that is what i get in my crashlogs
00:34 < rxr> but onyl the backtrace of the thread that are not the one with the illegal instruction with this callback thing
00:34 < rxr> click on the t2 html IRC log and read it there
00:34 < valentin> i get backtraces of all threads
00:34 < rxr> I ranted about 20 minutes while I found that out ...
00:34 < rxr> I doo
00:34 < valentin> which log did you mean ?
00:34 < rxr> too even
00:35 < rxr> but not for this case where this has s.th. to to with the callbck
00:35 < valentin> ok, i belive you
00:35 < rxr> this colorful send bug to apple thing ..
00:35 < valentin> no
00:35 < valentin> i mean the crashlogs
00:35 < valentin> in ~/Library/Logs/CrashReporter/
00:35 < rxr> hread: 0
00:35 < rxr> Exception: EXC_BAD_INSTRUCTION (0x0002)
00:35 < rxr> Code[0]: 0x00000002
00:35 < rxr> Code[1]: 0x03020100
00:35 < rxr> Thread 0 Crashed:
00:35 < rxr> Thread 1:
00:35 < rxr> 0 libSystem.B.dylib 0x900074c8 mach_msg_trap + 0x8
00:35 < rxr> 1 libSystem.B.dylib 0x90007018 mach_msg + 0x38
00:35 < rxr> 2 com.unsanity.ape 0xc0002544 __ape_internal + 0xca4
00:35 < rxr> 3 com.unsanity.ape 0xc0001330 __ape_agent + 0x40
00:35 < rxr> 4 libSystem.B.dylib 0x900246e8 _pthread_body + 0x28
00:35 < rxr> do you see a backtrace for the crashed thread 0 ?
00:36 < valentin> no
00:36 * rxr neither
00:36 < rxr> it is really s.th. with this callback or with the RunLoop around it
00:36 < valentin> did you paste that from the reporter or out of your vi ?
00:37 < rxr> I hacked some more printf's into it (hail to the apple debugger and and/or gdb port) and I get nothing after the callback is called
00:37 < rxr> out of the colorful dialog thing ...
00:37 < rxr> why ?
00:37 < valentin> because this dialog may filter some things out
00:37 < valentin> don't you think ?
00:38 < rxr> no idea - so far I hod no problem with it and thought they just place the text into the scroll view ...
00:39 < rxr> ouhm
00:39 < rxr> you are right
00:39 < rxr> Thread 0 Crashed:
00:39 < rxr> 0 libobjc.A.dylib 0x908311f4 objc_msgSend + 0x14
00:39 < rxr> 1 <<00000000>> 0x92f65244 0 + 0x92f65244
00:39 < rxr> 2 <<00000000>> 0x92de4a48 0 + 0x92de4a48
00:39 < rxr> 3 <<00000000>> 0x92e77e30 0 + 0x92e77e30
00:39 < rxr> 4 <<00000000>> 0x92e7eb88 0 + 0x92e7eb88
00:39 < rxr> 5 <<00000000>> 0x92eb8e3c 0 + 0x92eb8e3c
00:39 < rxr> 6 <<00000000>> 0x92e58848 0 + 0x92e58848
00:39 < rxr> 7 <<00000000>> 0x92ec9a4c 0 + 0x92ec9a4c
00:39 < rxr> 8 <<00000000>> 0x92e8ec58 0 + 0x92e8ec58
00:39 < rxr> 9 <<00000000>> 0x92e02e28 0 + 0x92e02e28
00:39 < rxr> 10 <<00000000>> 0x92df54ec 0 + 0x92df54ec
00:39 < rxr> 11 <<00000000>> 0x92dfd904 0 + 0x92dfd904
00:39 < rxr> 12 <<00000000>> 0x92eb9f00 0 + 0x92eb9f00
00:39 < rxr> 13 com.apple.TWAINClientCocoa 0x00009870 main + 0x38
00:39 < rxr> 14 com.apple.TWAINClientCocoa 0x00009568 _start + 0x188 (crt.c:267)
00:40 < rxr> 15 com.apple.TWAINClientCocoa 0x000093dc start + 0x30
00:40 < rxr> but - hm
00:41 < rxr> no that is not the log matching my latest crash
00:41 < rxr> some values and the the thread count are different ...
00:41 < valentin> scroll down
00:41 < valentin> there are multiple crashes in one log (find that quite anoying ...)
00:42 < rxr> OH - YES
00:42 < rxr> do they get anything right ?
00:42 < valentin> maybe that is a feature for dau users
00:42 < rxr> ok - there it is the same:
00:42 < rxr> Thread 0 Crashed:
00:42 < rxr> Thread 1:
00:42 < rxr> ...
00:43 < valentin> that sucks ...
00:45 < rxr> Andréé Watter - Si exceptionne
00:45 < rxr> ^- quite nice french rock
00:45 < rxr> at least s.th. good while hacking ...
00:45 < rxr> valentin: the question is: is this an indication of an more underlying bug?
00:46 < rxr> maybe the cause for other real problem?
00:46 < valentin> could be so
00:46 < rxr> or should I just discard this try to use this for my "run-time-verification" code ß
00:46 < rxr> ? even
00:46 < valentin> there are three explanation for an illegal instruction
00:46 < valentin> (0) broken cpu
00:46 < valentin> 1 broken gcc
00:46 < valentin> 2
00:47 < valentin> 2 overwritten memory
00:47 < valentin> 3 jump at wrong address
00:47 < valentin> i guess it is 2 or 3
00:47 < rxr> well - so far I never had problems with this 750FX ...
00:47 < rxr> I also never overclocked it so far ...
00:48 < rxr> altough it might have suffered due to constant T2 builds and the strange Apple fan control ....
00:48 < rxr> and got some sunburn ...
00:48 < rxr> well - it can not be overwritten memory
00:48 < rxr> if the apple patched gcc generates correct code
00:49 < rxr> the value is anyway assigned to zero at the top of the initiating function
00:49 < rxr> it is a function local variable ...
00:49 < valentin> i did not mean the variable but something that happens in between due to the value of the variable
00:49 < valentin> i do not know the code
00:49 < rxr> darwin.c ...
00:50 < rxr> btw do not do svn up
00:50 < rxr> you will get some broken instrucmentation code that will crash as discussed here ...
00:50 < rxr> better news: 1370 builds total, 685 completed fine, 3 with errors.
00:53 < rxr> valentin: throwing a positive value into the value does not crash this internal usb function
00:53 < rxr> hm - maybe the negative assignment generates incorrect code?
00:53 < valentin> what does a negative value mean ?
00:53 < rxr> e.g. writing a 64 value over the int boundaries or so ...
00:54 < valentin> sure this case does not something evil ?
00:54 < rxr> what case?
00:54 < valentin> the negative value case
00:54 < rxr> well it is applicatoin private date ...
00:55 < rxr> this self pointer thing you normally always have with callbacks
00:55 < rxr> osx should not touch it in any way ...
00:55 < valentin> i need to look into the code first, otherwise that discussion is quite pointless
00:56 < valentin> can you point me to some line in darwin.c ?
00:56 < rxr> well yours is slightly different
00:56 < rxr> browse to usb_bulk_read(
00:56 < rxr> and follow received
00:56 < rxr> retrieved even
00:57 < valentin> ok
00:57 < rxr> I'm soon in bed ...
00:58 < rxr> refCon
00:58 < rxr> Arbitrary pointer which is passed as a parameter to the callback routine.
00:59 < rxr> (void *)&retrieved);
00:59 < valentin> i think this may be a bug in ReadPipeAsync that overwrites ram
00:59 < rxr> " (void *)&retrieved);
00:59 < rxr> oops
00:59 < rxr> "Arbitrary pointer" is for me when the library and system stuff leaves it alone ...
01:00 < valentin> so it is evaluated in the callback only - now i understand
01:00 < rxr> btw. what got out of your prime investigation lately ?
01:00 < valentin> i emailed djb
01:00 < rxr> retrieved = (int32_t)-666;
01:00 < valentin> he did not answer but i did not expect that anyway
01:00 < rxr> rw_completed: ret_io_size: -666, io_size: 96
01:00 < rxr> rw_completed: ret_io_size: 96, io_size: 96
01:00 < rxr> rw_completed: RETURNING
01:01 < rxr> Illegal instruction
01:01 < rxr> as before ...
01:02 < valentin> can you point me to the rw_completed implementation ?
01:02 < rxr> it is some lines about the read and write functions in the same file
01:03 < rxr> btw it does not seem to crash for writes
01:03 < rxr> only for reads ...
01:04 < valentin> the signature is
01:04 < valentin> static void rw_completed(void *ret_io_size, io_return_t result, void *io_size)
01:04 < rxr> ehrm
01:04 < rxr> yes and ?
01:04 < rxr> btw I changed the assignemnts to 666
01:04 < rxr> now i get a segmentation fault ...
01:05 < valentin> with a backtrace ?
01:05 < rxr> Thread 0 Crashed:
01:05 < rxr> 0 de.exactcode.AvisionDS 0x0445abb8 sanei_usb_find_devices + 0x70 (sanei_usb.c:506)
01:05 < rxr> too bad we have no valgrind on max osx, hm ?
01:05 < rxr> mac even ..
01:06 < valentin> back to the signature
01:06 < rxr> as far as I can see there is random memory overwritten somewhere
01:06 < valentin> the callback has 3 arguments
01:06 < rxr> otherwise one value would not cause s SIGILL and another some segmentation fault
01:06 < rxr> sepecially not when it is some "this" pointer for private use ...
01:06 < valentin> that is what i say since 30 minutes
01:07 -!- freed [~uni@pD9E77F0E.dip.t-dialin.net] has joined #t2
01:07 < freed> hi @ll
01:07 < valentin> hi freed
01:07 < freed> hi valentin
01:07 < rxr> typedef void (*IOAsyncCallback1)(
01:07 < rxr> void *refcon,
01:07 < rxr> IOReturn result,
01:07 < rxr> void *arg0);
01:07 < rxr> hi freed
01:07 < freed> hi rxr
01:08 < valentin> but ReadPipeAsync only calls the callback with one parameter ?
01:08 < rxr> refcon
01:08 < rxr> The refcon passed into the original I/O request
01:08 < rxr> result
01:08 < rxr> The result of the I/O operation
01:08 < rxr> arg0
01:08 < rxr> Extra argument
01:08 < valentin> ok
01:09 < valentin> now i understand it
01:09 < rxr> IOReturn (*ReadPipeAsync)(
01:09 < rxr> void *self,
01:09 < rxr> UInt8 pipeRef,
01:09 < rxr> void *buf,
01:09 < rxr> UInt32 size,
01:09 < rxr> IOAsyncCallback1 callback,
01:09 < rxr> void *refcon);
01:10 < rxr> freed: I'm off in bed in some seconds
01:10 < valentin> ahm rxr
01:10 < valentin> your code has to crash
01:10 < rxr> in fact I wanted to crawl into it early ...
01:10 < valentin> you overwrite stack memory
01:10 < rxr> why ?
01:10 < rxr> hm ?
01:10 < valentin> if (result == kIOReturnSuccess)
01:10 < valentin> *(u_int32_t *)ret_io_size = (u_int32_t)io_size;
01:10 < valentin> else
01:10 < valentin> *(u_int32_t *)ret_io_size = result;
01:10 < rxr> this was there ...
01:10 < valentin> but that is a bug
01:10 < rxr> AFAIR I did not tuch it ...
01:11 < rxr> oehm
01:11 < rxr> why =
01:11 < valentin> because ret_io_size points to retrieved
01:11 < valentin> which is a local variable ?
01:11 < valentin> or isn't it ?
01:11 < rxr> it is a local, and ?
01:11 < valentin> ok, it isnt
01:11 < rxr> uuhm?
01:11 < valentin> yes it is
01:12 < rxr> it is a local of the callee ...
01:12 < rxr> but this looks all right
01:12 < valentin> int usb_bulk_read(usb_dev_handle *dev, int ep, char *bytes, int size,
01:12 < valentin> int timeout)
01:12 < valentin> {
01:12 < valentin> ....
01:12 < rxr> they cast it to *int32_t and then dereference it ..
01:12 < valentin> u_int32_t retrieved = 0;
01:12 < valentin> but that is a local variable (!!!!)
01:12 < rxr> they pass the pointer to the function
01:12 < rxr> and?
01:12 < valentin> it points to a stack adress ???
01:12 < rxr> and ?
01:13 < valentin> and ReadAsync calls the callback _after_ usb_bulk read returned, doesn't it ?
01:13 < rxr> nope
01:13 < rxr> ReadAsync RunLoops to wait for the completion
01:13 < valentin> ah - ic
01:14 < rxr> so - still not solution ...
01:14 < valentin> what a shitty code - why do they use ReadAsync then ?
01:15 < rxr> maybe because without the Async the timeout is not controllable ?
01:15 < rxr> freed: how are you?
01:15 < rxr> freed: any AVR Atmel progress on your side ?
01:17 < valentin> rxr: that may be a reason
01:17 < valentin> i am just reading CFRunLoop documentation
01:19 -!- CIA-10 [~CIA@to.je.spocco.com] has quit [Remote closed the connection]
01:22 < rxr> but rereading the code I do not fully understand why to have this RunLoop stuff in it at all
01:24 < rxr> valentin: no the timeout is available with the *To functions
01:25 < rxr> we could try if it makes a difference to strip the Async away and try what happens
01:25 < rxr> but I really need to bed - I do not want to get into such a shifted sleep cycle again ...
01:25 < valentin> there is a function called ReadPipeTO
01:25 < rxr> yes
01:26 < rxr> that is what I meant above ...
01:26 < valentin> with timeout and without Async :)
01:26 < rxr> ack
01:26 < valentin> maybe you can try tomorow using that one
01:26 < rxr> yep - I do ..
01:26 < rxr> so /me off ...
01:26 < valentin> without runloops we do not need in that thread
01:26 < valentin> gn8 rxr
01:27 < rxr> well - but then the gui would hang even more, hm?
01:27 < rxr> so we still might like to keep our io thread ?
01:27 < valentin> yes
01:27 < rxr> what did you mean with "we do not need in that thread" then ?
01:28 < valentin> we do not need to activate the run loop because it is active in the UI thread
01:28 < rxr> yep
01:28 < rxr> hm
01:28 < valentin> i do not even know if RunLoopRun is thread save
01:28 < rxr> well - the docu list "run the current RunLoop"
01:29 < valentin> but is there one run loop for each thread ?
01:29 < rxr> I guess every thread has - it is just useless and most probably empty in our io thing ...
01:29 < valentin> or one run loop per process
01:29 < rxr> the docu speaks of thread ...
01:29 < rxr> and about defautls ..
01:29 < rxr> my conclusion is every thead has a default (most probably empty) one ..
01:30 < valentin> There is exactly one run loop per thread. You neither create nor destroy a thread?s run loop. Core Foundation automatically creates it for you as needed. You obtain the current thread?s run loop with CFRunLoopGetCurrent.
01:30 < rxr> but smart conclusions and osx implementation seems to differ occationaly ... ...
01:30 < valentin> rxr: maybe libusb isn't threadsave ?
01:30 < valentin> they get the run loop only once and use it forever ?
01:30 < valentin> and sane uses bulk_read in different threads ?
01:31 < rxr> 1370 builds total, 705 completed fine, 3 with errors.
01:32 < valentin> ok, there are many calls to CFRunLoopGetCurrent
01:32 < valentin> just remove that runloop / Async stuff tomorow and we will see ...
01:33 < valentin> gn8 rxr
01:36 < rxr> n8 valentin
01:36 < valentin> rxr: still here ?
01:36 < valentin> mom
01:36 < rxr> I hope you go to bed, too ...
01:36 < rxr> yeah - telling freed how to transfer bytes over the network
01:36 < valentin> ReadPipeAsyncTO(device->interface, pipeRef,
01:36 < valentin> bytes + retrieved, size, 0,
01:36 < rxr> .oO
01:36 < rxr> I'm blinded ...
01:36 < rxr> thansk
01:37 < valentin> what happens if that value is negative ?
01:37 < rxr> ugly things ...
01:37 < valentin> take a look tomorow :)
01:37 < valentin> gn8
01:37 < freed> gn8
01:38 < rxr> valentin: what a useless piece of code ...
01:38 < rxr> this retrieved is totally bogus ..
01:38 < rxr> since size not adapted anyway ..
01:38 < rxr> and no loop insight, too ...
01:39 < valentin> hmm
01:39 < valentin> you seem to be rigth
01:39 < valentin> right even
01:39 < rxr> I commented it out ...
01:39 < rxr> now
01:39 < valentin> but this will not fix our IO problems
01:40 < rxr> no
01:40 < rxr> but my runtime verification code ...
01:40 < rxr> thanks for the spot
01:40 < mnemoc> re
01:40 < valentin> only took us 80 minutes *g*
01:40 < valentin> hi menmoc
01:40 < mnemoc> hi valentin
01:40 < rxr> never reuse code not written in house
01:41 < rxr> so - hopefully I can father some data of this run-time check ...
01:41 < rxr> since I only added it to review if the timer ever returns without an associated async event ..
01:43 < rxr> ok - code running again ...
01:43 < rxr> i/o working
01:43 < rxr> preview complete ...
01:43 < rxr> btw: we migth like to review anyway if the runloop code removal brings us speed
01:44 < rxr> maybe the permanent run loop enter for every single USB package causes the big performance impact we see compared to the 2x faster transfer times in linux ...
01:44 < valentin> yep
01:44 < rxr> ouhm
01:44 < rxr> crash
01:45 < rxr> Exception: EXC_BAD_ACCESS (0x0001)
01:45 < rxr> Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x00000028
01:45 < rxr> Thread 0 Crashed:
01:45 < rxr> 0 libobjc.A.dylib 0x908311ec objc_msgSend + 0xc
01:45 < rxr> 1 de.exactcode.AvisionDS 0x0445db2c -[AvisionDS preview:] + 0x78 (AvisionDS.M:382)
01:45 < valentin> maybe that is my code
01:45 < valentin> my scanner never came so far due to IO errors
01:46 < rxr> this was a preview IIRC ...
01:46 < rxr> not a scan
01:46 < valentin> yes i know
01:46 < rxr> hey
01:46 < rxr> valentin:
01:46 < rxr> you do not believe
01:46 < rxr> this
01:46 < valentin> and i never saw a preview using my new multithreaded io error
01:46 < valentin> what ?
01:47 < rxr> o async operation completed. result = -536870165, size = 0
01:47 < rxr> rw_completed: ret_io_size: 3789920, io_size: 0
01:47 < rxr> rw_completed: ret_io_size: -536870165, io_size: 0
01:47 < rxr> rw_completed: RETURNING
01:47 < rxr> CFRunLoop most likely returned without I/O completed!
01:47 < rxr> this is my new checker code
01:47 < rxr> and the code is currently in this always "internal error state" we experienced some times
01:48 < rxr> btw. this is not yet your multithreaded i/o facade ...
01:48 < rxr> I reverted it locally because I thought it caues the segfaults ...
01:48 < rxr> I can svn up again now ...
01:48 < valentin> ok
01:48 < valentin> i am not in OSX currently
01:48 < valentin> so i suggest you remove all this runloop crap
01:49 < valentin> and do an s/Async//g
01:49 < valentin> i guess there are some cocoa timer events getting in the way
01:50 < rxr> Skye Sweetnm - Billy s
01:50 < rxr> is also quite nice
01:50 < valentin> so you can put your mind at rest and go to sleep
01:51 < rxr> in a minute
01:51 < rxr> 1:50 or 2:00 does not matter much ...
01:51 < valentin> i need to drown my cake in chocolade
01:53 < rxr> btw
01:53 < rxr> didn't you need that change:
01:53 < rxr> Skye Sweetnm - Billy s
01:53 < rxr> ehrm
01:53 < rxr> +
01:53 < rxr> + double last_progress;
01:53 < rxr> or did you tweak it locally ?
01:54 < valentin> i just commented it out
01:54 < rxr> valentin: using your threaded wrapper I get dome date
01:55 < rxr> but then this:
01:55 < valentin> because i did not have the declaration
01:55 < valentin> dome date ?
01:55 < valentin> some data ?
01:55 < rxr> well - my i/o core runs:
01:55 < rxr> rw_completed: ret_io_size: -666, io_size: 20304
01:55 < rxr> rw_completed: ret_io_size: 20304, io_size: 20304
01:55 < rxr> rw_completed: RETURNING
01:55 < rxr> [sanei_usb] sanei_usb_read_bulk: wanted 20304 bytes, got 20304 bytes
01:55 < rxr> but your code gets:
01:55 < rxr> 2005-01-08 01:54:34.027 TWAINClientCocoa[909] stream ended because: Error during device I/O
01:55 < rxr> 2005-01-08 01:54:34.132 TWAINClientCocoa[909] scanning ....
01:55 < rxr> and then it crashes ...
01:56 < valentin> when doing what ?
01:56 < rxr> I clicked on scan IIRC
01:56 < valentin> scan does not work
01:56 < rxr> or maybe preview
01:56 < rxr> but I think scan
01:56 < rxr> 0 de.exactcode.AvisionDS 0x044614cc -[AvisionDS transformPreviewDataBytesRGB:] + 0xec (AvisionDS.M:297)
01:56 < rxr> 1 de.exactcode.AvisionDS 0x0446184c -[AvisionDS previewProcessData] + 0x11c (AvisionDS.M:328)
01:56 < rxr> ah - so it was preview ...
01:57 < rxr> yep - preview crahes
01:57 < rxr> was that known ?
01:57 < valentin> i never got any data out of the scanner as i told you 20 minutes before
01:58 < rxr> ok ok
01:58 < valentin> so i did not have a crash
01:58 < valentin> but an empty preview
01:58 < rxr> my brain starts to fall aslepp ...
01:58 < valentin> then go to bed ,)
01:58 < rxr> yeah ..
01:58 < rxr> you just wanna eat your cake alone ...
02:00 < rxr> 1370 builds total, 713 completed fine, 3 with errors.
02:03 * rxr bed ...
02:03 < valentin> gn8
02:06 < rxr> yeah - n8, too
02:07 < mnemoc> n8 rxr
02:52 -!- rxr_ [~rene@p213.54.196.126.tisdip.tiscali.de] has joined #t2
02:52 -!- Topic for #t2: T2 | 2.1.0-beta2 RELEASED | The next generation of System Development Enviroments (SDE) | http://www.exactcode.de/t2
02:52 -!- Topic set by mnemoc [] [Tue Dec 28 14:55:44 2004]
02:52 [Users #t2]
02:52 [ _Ragnar_] [ freed] [ mnemoc] [ praenti] [ rxr_ ] [ valentin]
02:52 [ daja77 ] [ jsaw ] [ mtr_ ] [ rxr ] [ sparc-kly]
02:52 -!- Irssi: #t2: Total of 11 nicks [0 ops, 0 halfops, 0 voices, 11 normal]
02:52 -!- Channel #t2 created Sun Aug 8 21:15:33 2004
02:52 -!- [freenode-info] help freenode weed out clonebots, please register your IRC nick and auto-identify: http://freenode.net/faq.shtml#nicksetup
02:52 -!- Irssi: Join to #t2 was synced in 10 secs
03:02 < mnemoc> *yawn*
03:08 -!- rxr [~rene@p213.54.194.178.tisdip.tiscali.de] has quit [Read error: 110 (Connection timed out)]
04:20 -!- ideal [~idealm@211.100.227.107] has joined #t2
05:43 -!- Postal [~Postal@ool-43562a7f.dyn.optonline.net] has joined #t2
05:43 < Postal> Hey does anyone know the svn update command?
05:45 -!- Postal [~Postal@ool-43562a7f.dyn.optonline.net] has quit [Client Quit]
05:51 < mnemoc> uhm
06:42 < mnemoc> jsaw: does rock^Whwplug use procinfo?
07:11 -!- mnemoc_ [~amery@200.75.27.47] has joined #t2
07:14 -!- uni_ [~uni@pD95D0F8F.dip.t-dialin.net] has joined #t2
07:27 -!- mnemoc [~amery@200.75.27.37] has quit [Read error: 110 (Connection timed out)]
07:32 -!- freed [~uni@pD9E77F0E.dip.t-dialin.net] has quit [Read error: 110 (Connection timed out)]
08:33 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has joined #t2
08:42 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has quit [Read error: 54 (Connection reset by peer)]
08:42 -!- N0V4K_ [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has joined #t2
08:43 -!- N0V4K_ [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has quit [Client Quit]
08:43 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has joined #t2
09:24 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has quit ["FreeBSD N0V4K.Zero 5.3-RELEASE FreeBSD 5.3-RELEASE Chao"]
09:24 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has joined #t2
09:26 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has quit [Remote closed the connection]
09:43 -!- sparc-kly [~sparc@64.237.130.154] has quit [Read error: 104 (Connection reset by peer)]
11:59 -!- You're now known as rxr
11:59 < rxr> moin
11:59 < rxr> [5] network/xmule [5] e17/efsd
11:59 < rxr> [5] e17/entrance [5] contrib/joystick
11:59 < rxr> [5] multimedia/blender [5] kde/koffice
11:59 < rxr> [5] graphic/pixieplus [5] kde/superkaramba
11:59 < rxr> [5] scientific/qmatplot [5] audio/xmp
11:59 < rxr> [5] database/mysqlcc [5] multimedia/nvrec
11:59 < rxr> [5] base/ifhp [5] audio/vstserver
11:59 < rxr> [5] multimedia/kaxtv [5] multimedia/mythtv
11:59 < rxr> [5] multimedia/avview [5] audio/ladspa-vst
11:59 < rxr> [5] audio/audacity [5] scientific/xaos
11:59 < rxr> 1370 builds total, 910 completed fine, 20 with errors.
12:03 < mnemoc_> moin rxr
12:04 < rxr> moin
12:05 < mnemoc_> there is no CIA bot neither on #commits
12:05 < rxr> hm
12:08 < mnemoc_> rxr: are you using XMLRPC?
12:08 < mnemoc_> (me talking about cia)
12:08 < rxr> no
12:09 < rxr> xmlmail or how the nammed it ...
12:09 < rxr> I did not wanted some strange python code or so on the server for security considerations
12:09 < rxr> why ?
12:10 < mnemoc_> "Please upgrade if you're still using a client that doesn't send XML messages" <--- on #commits' topic
12:10 < mnemoc_> and http://cia.navi.cx/stats/project/t2 is outdated
12:11 < rxr> /t2/packages/kaxtv.html
12:11 < rxr> ouhm
12:11 < rxr> ^- top #7 hit on our t2 site
12:12 < mnemoc_> o_O
12:12 < rxr> but most of those people that get to that page thru google exit there, too :-(
12:12 < mnemoc_> :\
12:21 < mnemoc_> how can i ask gcc to show me the #include tree it use on a build?
12:21 < rxr> include tree as tree might not provided by gcc
12:22 < rxr> most often -E (only preprocessing is a help
12:22 < rxr> it still contains marks what got included where ...
12:22 < rxr> recently I found -dD (IIRC) helpfull it will dump the preprocessed source including(!) the macro definitions
12:24 < mnemoc_> sounds as what i want :)
12:28 < mnemoc_> brb
12:30 < rxr> well - some stuff fails due to python library dependencies - as usual ...
12:46 < mnemoc_> -E -dD or just -dD ?
12:48 < mnemoc_> what extension is used for preprocessed files?
12:55 < rxr> I always used -E -dD
12:56 < rxr> when you do not supply -o you get it on stdout ...
12:56 < rxr> when -o is present the specified file is used of course ...
12:56 < mnemoc_> i was lazy and added -e -dD to CFLAGS, now i have text .o files :)
13:14 -!- nzg [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
13:14 < nzg> moin
13:15 < rxr> moin nzg
13:15 [Users #t2]
13:15 [ _Ragnar_] [ ideal] [ mnemoc_] [ nzg ] [ rxr ] [ valentin]
13:15 [ daja77 ] [ jsaw ] [ mtr_ ] [ praenti] [ uni_]
13:15 -!- Irssi: #t2: Total of 11 nicks [0 ops, 0 halfops, 0 voices, 11 normal]
13:15 < nzg> that damn box is down again....
13:15 < rxr> ouch
13:16 < nzg> this time it will be my boss who calls ;-)
13:16 < rxr> ,)
13:31 * nzg is away: einkaufen
13:33 < nzg> sry
13:34 -!- uni_ [~uni@pD95D0F8F.dip.t-dialin.net] has quit [Remote closed the connection]
13:38 < rxr> 1370 builds total, 998 completed fine, 22 with errors.
13:46 < rxr> # ./debug.sh
13:46 < rxr> Segmentation fault (core dumped)
13:46 < rxr> I thought this was solved/work arounded ... ?
13:59 < rxr> kkk1370 builds total, 1019 completed fine, 23 with errors.
14:01 < mnemoc_> i thought it was
14:08 -!- valentin [~valentin@port-212-202-170-162.dynamic.qsc.de] has quit [Read error: 60 (Operation timed out)]
14:08 < mnemoc_> rxr: problems with ec.de server?
14:15 < jsaw> re
14:17 < mnemoc_> re jsaw
14:17 < jsaw> hi mnemoc_
14:18 < jsaw> hmm. again a segfaulting bash... o_O
14:18 < mnemoc_> yep
14:18 < jsaw> rxr: what does valgrind say?
14:18 < mnemoc_> OT: which extension is used for gcc -E output files?
14:19 < jsaw> .d iirc
14:19 < rxr> jsaw: later
14:19 * rxr breakfast ...
14:19 < jsaw> oki
14:23 < mnemoc_> svn: PROPFIND of '/t2/trunk': could not connect to server (http://svn.exactcode.de)
14:25 < jsaw> dito
14:26 < rxr> yep
14:26 < rxr> ack
14:29 < jsaw> btw, I assume, the uselib patch for 2.4 applies to 29-rc... have to check again.
14:29 -!- valentin [~valentin@port-212-202-42-228.dynamic.qsc.de] has joined #t2
14:29 < valentin> test
14:30 < valentin> so we seem to be back online
14:30 < mnemoc_> yes
14:31 < valentin> my gf was cleaning up and stumbled over the cable
14:31 < mnemoc_> hehe
14:32 < jsaw> rxr: will you push the new .cache files into trunk?
14:32 < valentin> cu later
14:32 < jsaw> cu valentin
14:48 < rxr> jsaw: sorry - this was a in-system build of a package on my athlon which had nto yet an up-to-date'ly patched bash installed ...
14:48 < rxr> 1370 builds total, 1045 completed fine, 25 with errors.
14:49 < rxr> [5] network/xmule [5] e17/efsd
14:49 < rxr> [5] e17/entrance [5] contrib/joystick
14:49 < rxr> [5] multimedia/blender [5] kde/koffice
14:49 < rxr> [5] graphic/pixieplus [5] kde/superkaramba
14:49 < rxr> [5] scientific/qmatplot [5] audio/xmp
14:49 < rxr> [5] database/mysqlcc [5] multimedia/nvrec
14:49 < rxr> [5] audio/vstserver [5] multimedia/kaxtv
14:49 < rxr> [5] multimedia/mythtv [5] multimedia/avview
14:49 < rxr> [5] audio/ladspa-vst [5] audio/audacity
14:49 < rxr> [5] scientific/xaos [5] network/slidentd
14:49 < rxr> [5] multimedia/ripperx [5] network/centericq
14:49 < rxr> [5] contrib/djvulibre [5] develop/clip
14:49 < rxr> [5] security/ebtables
14:49 < rxr> jsaw: do you want the .cache files beeing injected?
14:50 < mnemoc_> clip?
14:50 < rxr> I just looked
14:50 < rxr> download missing
14:50 < mnemoc_> get it from my mirrors :)
14:52 < mnemoc_> OT: what do you think if Config's choice is able to change default if that key is not on the option list taking the first/last key as new default?
14:54 < jsaw> rxr: no, I was just asking. Only thing I'd really like to know, when the copyright headers will be changed... if it's more than 1week from now, I still have time to start a build.
14:55 < jsaw> anything against linux 2.4.28 -> 2.4.29-rc1???
14:56 < mnemoc_> considering it's a security need i guess we should do it
14:57 < jsaw> (kernel build running...)
14:57 < rxr> well -rc1 should be quite stable so - go for it .. ;-)
14:58 < rxr> INFO: Testing ... errorINFO: Testing ... errorINFO: Testing ... ok 10886.000INFO: Testing ... ok 12520.000INFO: Testing ... error
14:58 < mnemoc_> what about the bleeding 2.6.10 linux-header?
14:58 < rxr> mnemoc_: if you wanna do the patchwork ...
14:59 < mnemoc_> i don't have enough knowledge to #ifdef __KERNEL__ properly
14:59 < rxr> jsaw: well - if we agree about the copyright one day we could inject them ...
14:59 < jsaw> hehe
15:00 < rxr> then let's get some -beta3 out, first before we spend time sanitizing newer kernel headers ...
15:00 < rxr> mnemoc_: thanks for providing mirrors - now that all .de mirrors are down ...
15:01 < mnemoc_> oh
15:01 < mnemoc_> is that coincidence?
15:01 < rxr> I just announced the http://tfh-berlin.de downtime to the list
15:01 < rxr> tfh-berlin.de was planed - but I forgot to announce that
15:02 < rxr> they do some construction work and cut some inter-campus data wires or so
15:02 < rxr> but I do not know the strange details about the stuff happening at my university of applied science ...
15:03 < rxr> the only cunstruction work I can see is a new plant growing glas house ..
15:04 < rxr> no idea how this interferes with data-wires ...
15:04 < rxr> but how knows ...
15:04 < rxr> who even
15:04 < nzg> maybe they whacked some wires when they grounded the greenhous
15:04 < rxr> jsaw: yep - a Emerge-Pkg bash solved the seg-fault ..
15:04 < rxr> no - this already happend weeks ago ...
15:05 < jsaw> ?
15:05 < rxr> they even announced the downtime some days ago ...
15:05 < rxr> the greenhouse even finished - AFAICS only the plans are missing or so ...
15:06 < rxr> jsaw: I could inject the .cache files - but as I wrote some times already, I do not like the noisy nature of the .cache files
15:06 < rxr> I shomehow hesitate to do such monster commits
15:06 < rxr> do you remeber my plans for that files?
15:06 < rxr> and should we start the migration work now ?
15:06 < jsaw> rxr: to be honest, I don't want that noise either.
15:07 < mnemoc_> we need an updated .cache set to migrate
15:07 < jsaw> well, right. That's the other side...
15:08 < rxr> well - the updated set would not have to be commited
15:08 < jsaw> rxr,mnemoc_: would it be enough to keep only "[DEP]" information in the cache files, or will some scripts break?
15:09 < rxr> instead I could directly generate .dep and new-styl .cache files ...
15:09 < rxr> where the new style .cache files would contain less unneeded noisy info ...
15:09 < mnemoc_> do you really have time to do that?
15:09 < rxr> and be rounded so some exponentially limitted value so tine changes would not alter them ..
15:09 < rxr> no
15:10 < mnemoc_> so?
15:10 < rxr> so - you could do it :-)
15:10 < mnemoc_> not today :\
15:11 < jsaw> mnemoc_: rule number one in this channel: never ask such things, otherwise you end up doing it yourself ;)
15:11 < mnemoc_> :)
15:11 < rxr> well - sure - I could inject new .cache files one last time and we do the migration when time permits ...
15:12 < rxr> /usr/include/linux/dvb/dmx.h:164: error: `__u64' does not name a type/usr/include/linux/dvb/dmx.h:164: error: extra semicolon
15:12 < rxr> ieeee
15:12 < rxr> I hate linux-header oddities
15:13 < mnemoc_> santizing 2.6.8 is senseless
15:13 < rxr> nope - it was the newest release when we did it - and all the later kernels only got more broken with each release ...
15:14 -!- ideal [~idealm@211.100.227.107] has quit ["Leaving"]
15:14 < mnemoc_> we have to get used to that
15:14 < mnemoc_> until hurd gets decent ... 10 years?
15:15 < rxr> well - I do much work - but that is really s.th. I do not volunteer to do
15:15 < rxr> I do not say we shoud keep those 10 years - just that is a very low priority for me ...
15:15 < rxr> of course you, ...
15:18 < mnemoc_> i have loooooot to learn before patching kernel files
15:19 < rxr> mnemoc_: I see only negative time windows in january over here ...
15:19 < rxr> if you can wait until february i can do it then ...
15:19 < rxr> alternativly you can find someone paying me as much as i get for the osx driver so I can stop hacking on it right now ...
15:21 < mnemoc_> :)
15:21 < rxr> but hey - on the other side, if this driver gets a success I can hack on whatever I want the remaining whole year ...
15:24 < jsaw> no CIA around anymore?
15:25 < rxr> no
15:25 < rxr> need to migrate mail xmlrpc ...
15:31 < rxr> migrate from mail to ...
15:31 < rxr> sorry was phoning
15:37 < mnemoc_> what do you think is better, add GROUPS_MAX to uclibc's limits.h or include linux/limits.h on gawk's awk.h?
15:43 < rxr> well - it if is normally in limits.h I would consider it a uclibc bug and patch it accoringly ...
15:44 < rxr> I could grep my BSD based Apple system if it is a help ..
15:45 < rxr> /usr/include/limits.h:#define _POSIX_NGROUPS_MAX 0
15:45 < rxr> /usr/include/sys/syslimits.h:#define NGROUPS_MAX 16 /* max supplemental group id's */
15:45 < rxr> /usr/include/sys/param.h:#define NGROUPS NGROUPS_MAX /* max number groups */
15:46 < mnemoc_> param.h includes linux/limits.h
15:46 < rxr> what is the actual define you are looking for
15:46 < rxr> ?
15:46 < mnemoc_> NGROUPS_MAX
15:46 < rxr> my rs6k also has no GROUPS_MAX in limits.h
15:46 < rxr> a - NGROUPS_MAX
15:47 < rxr> yep - that is in both - my OSX and T2/RS6K limits.h ...
15:47 < mnemoc_> libc's ?
15:47 < rxr> hm
15:47 < rxr> or - wait
15:48 < mnemoc_> well, scanline is on a business trip, so it is currently unknown when the bots will be back
15:48 < rxr> well - actually it is in sys/param.h
15:49 < rxr> it is in some way defined when glibc is used - so just patch uclibc to define it "somewhere"
15:49 < mnemoc_> aha
15:49 < rxr> and it is here:
15:49 < rxr> /usr/dietlibc/include/limits.h:#define NGROUPS_MAX 32 /* supplemental group IDs are available */
15:49 < rxr> ;-)
15:50 < mnemoc_> how should i decide the limit?
15:50 < rxr> so far I only saw 32
15:50 < rxr> so just use it to ...
15:51 < mnemoc_> /usr/include/sys/syslimits.h:#define NGROUPS_MAX 16
15:51 < mnemoc_> ok
15:51 < rxr> the Open Group or so wants money for the Posix SPEC - so ...
15:54 < mnemoc_> http://www.opengroup.org/onlinepubs/000095399/toc.htm
15:58 < rxr> oh - good ;-)
16:00 < jsaw> bbl
16:01 < mnemoc_> cu jsaw
16:01 < mnemoc_> rxr: http://www.uclibc.org/cgi-bin/cvsweb/uClibc/include/limits.h?rev=1.5&view=auto <--- their limits.h is amazingly short
16:24 < rxr> then make it a line longer ;-)
16:25 < mnemoc_> NAME_MAX for util-linux is also missing
16:26 < rxr> two lines then ;-)
16:26 < mnemoc_> i'm waiting for uclibc's answer
16:26 < mnemoc_> but i guess your way is the safer
16:27 < mnemoc_> uhm PATH_MAX and LINE_MAX
16:28 < mnemoc_> somethign must be very wrong somewhere
16:45 < rxr> [5] network/xmule [5] e17/efsd
16:45 < rxr> [5] e17/entrance [5] contrib/joystick
16:45 < rxr> [5] multimedia/blender [5] kde/koffice
16:45 < rxr> [5] graphic/pixieplus [5] kde/superkaramba
16:45 < rxr> [5] scientific/qmatplot [5] audio/xmp
16:45 < rxr> [5] multimedia/nvrec [5] audio/vstserver
16:45 < rxr> [5] multimedia/kaxtv [5] multimedia/mythtv
16:45 < rxr> [5] multimedia/avview [5] audio/ladspa-vst
16:45 < rxr> [5] audio/audacity [5] scientific/xaos
16:45 < rxr> [5] multimedia/ripperx [5] contrib/djvulibre
16:45 < rxr> [5] security/ebtables [5] network/ngrep
16:45 < rxr> [5] network/asterisk [5] network/kismet
16:45 < rxr> [5] network/licq
16:45 < rxr> 1370 builds total, 1085 completed fine, 25 with errors.
16:46 < rxr> who is guilty for this one:
16:46 < rxr> Apply patch /TOOLCHAIN/package/network/kismet/gcc34.patch ...patching file packet.hHunk #1 FAILED at 550.Hunk #2 FAILED at 643.2 out of 2 hunks FAILED -- saving rejects to file packet.h.rej
16:46 < rxr> ?
16:46 < mnemoc_> svn log /package/network/kismet
16:47 < rxr> sure
16:47 < rxr> hm - only /me touched it ...
16:47 < rxr> damn
16:47 < mnemoc_> *G*
16:48 < rxr> I remeber I once deleted the patches - maybe I have just dreamed that and forgot the commit ...
16:48 < mnemoc_> rm instead of svn rm ;)
16:48 < rxr> yeah - like that ...
16:58 < mnemoc_> How Not to Program in C++: 111 Broken Programs and 3 Working Ones, or Why Does 2+2=5986
16:58 < mnemoc_> by Steve Oualline
17:09 < rxr> mnemoc_: url ?
17:09 < mnemoc_> closed, but wait
17:17 < rxr> 1370 builds total, 1102 completed fine, 26 with errors.
17:18 < mnemoc_> :)
17:18 < mnemoc_> your machine is fast
17:19 < rxr> 2.5 pseudo Mhz Athlon
17:19 < rxr> but also already a year old - or no - even older I think ...
17:19 < rxr> Ghz even ...
17:20 < rxr> let's hope I have some bucks for a x86-64 board this year ...
17:20 < mnemoc_> i took 3 days to build 1100 on a piv/2566/512
17:20 < rxr> yeah - P4 ...
17:20 < rxr> I only invested once in my life into a state-of-the-art Intel machine - and that was a Pentium 120 ...
17:21 < rxr> since then I only handed money to AMD, IDT (Winchip 2A) and Motorola/IBM ...
17:22 < mnemoc_> me life was: 8088 -> 8086 -> 80286 -> 486sx -> 486dx4 -> p2/300 -> p3/733 -> p4/2566 (laptop)
17:30 < rxr> Subject: [ANNOUNCE] linux-libc-headers 2.6.10.0
17:30 < rxr> linux kernel ...
17:31 < rxr> http://ep09.pld-linux.org/~mmazur/linux-libc-headers/
17:32 < mnemoc_> :D
17:32 < mnemoc_> will we take pld-linux' work?
17:35 < rxr> no idea
17:35 < rxr> maybe take a look if there is s.th. useful ...
17:37 -!- tschmidt [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
17:39 < tschmidt> re
17:39 < mnemoc_> hi tschmidt
17:40 < tschmidt> "you must boot your system into rescu mode" "it didn't work the last three times" "you must ...."
17:40 < tschmidt> hrmpf
17:40 < rxr> hmgrmpf ...
17:41 < tschmidt> if thats germany's best hoster - how bad must the rest be
17:44 < rxr> any answer why the box does /crash/ so often ?
17:46 < tschmidt> no...
17:47 < tschmidt> thats what the admins wrote in an e-mail:
17:48 -!- tschmidt_ [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
17:48 -!- tschmidt [~tschmidt@p508EB63C.dip.t-dialin.net] has quit [Read error: 104 (Connection reset by peer)]
17:49 < mnemoc_> rxr: can you give a try to Config -> mnemosyne? does it feel too slow? (compared to tipical configs)
17:50 < rxr> oehm
17:50 < rxr> later, ok?
17:50 < mnemoc_> sure
17:50 < mnemoc_> any day ;)
17:50 < rxr> I still USB hack on osx ...
17:52 < tschmidt_> so i have no clue
17:52 < tschmidt_> there is nothing in the logfiles that explains the crashes
17:55 < rxr> so - bad hardware, hm ?
17:56 < tschmidt_> thats what i suppose - it crashes disregarding to kernel-version or entire distribution
17:57 < tschmidt_> and i can't reproduce or provoke the crashes
17:58 -!- tschmidt_ is now known as tschmidt
17:58 < rxr> this damn xcode
17:59 < mnemoc_> rxr: is procinfo needed by rockplug?
18:00 < rxr> oehm
18:00 < rxr> i do not know all requirements of rockplug off-hand
18:00 < rxr> why do you ask=
18:00 < rxr> ? even
18:01 < mnemoc_> to know if i can remove it from mnemosyne-minimal
18:02 < mnemoc_> i'll try on qemu after lunch
18:08 * mnemoc_ moving home
18:09 * tschmidt calling S+P again...
18:20 < rxr> mnemoc_: I think it should work without procutils whatever - IIRC rockplug has very few dependencies ...
18:20 < rxr> tschmidt: :-(
18:21 < tschmidt> now in the line for 15min...
18:21 < tschmidt> still no one picked up
18:26 < tschmidt> wow that guy at least seemed sympathic "i'm gonna kick the admins ass so he really checks the hardware"
18:31 < rxr> .oO
18:31 < tschmidt> i bet h'll just push the rset button
18:32 < tschmidt> want to hold against?
18:32 < rxr> no
18:33 < tschmidt> ;-)
18:39 < rxr> valentin: I think you new code has some problems
18:39 < rxr> preview does crash - with the last commit revert it works with my new USB i/o code ...
18:40 < rxr> ouhm - no query, ...
18:41 < rxr> hm - i/o thruput does not seem to have improved :-(
18:53 < rxr> hm - ok - without the excessive debugging it feals at least slightly faster ...
18:54 -!- tschmidt_ [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
18:54 -!- tschmidt [~tschmidt@p508EB63C.dip.t-dialin.net] has quit [Read error: 104 (Connection reset by peer)]
18:56 * rxr most probably a bit away soon ...
18:58 < rxr> ScannerManager.h
18:58 < rxr> oops
18:58 < rxr> 1370 builds total, 1125 completed fine, 27 with errors.
19:00 < rxr> cu
20:22 -!- sparc-kly [~sparc-kly@64.237.130.154] has joined #t2
22:14 < rxr> re
22:15 < rxr> hi sparc-kly
22:27 < rxr> 1370 builds total, 1204 completed fine, 32 with errors.
22:28 < tschmidt_> rxr, ist back again
22:28 < tschmidt_> s/ist/its
22:28 < rxr> ;-)
22:28 < rxr> and, any additional information ?
22:29 < tschmidt_> not yes - they promised to send an e-mail
22:29 < rxr> U.S. Army Research Lab Opens BRL-CAD Source
22:29 < rxr> I just visited the side via freshmeat yesterday (IIRC) and wondered what it is about ...
22:29 < rxr> not slashdot took it up
22:30 < tschmidt_> if this promise is as reliable as the last ones...
22:30 < tschmidt_> what is brl-cad?
22:30 < rxr> never heared of it before - it is some CAD application
22:31 < rxr> http://brlcad.org/
22:33 < tschmidt_> it can be run on a pdp11 ;-)
22:35 < rxr> more likely could - I hope the current code base is too advanced for such an old beast
22:35 < rxr> otherwise the CAD software would be pretty useless nowadays ...
22:36 -!- mtr_ [~michael@H831b.h.pppool.de] has quit [Read error: 60 (Operation timed out)]
22:38 -!- mtr_ [~michael@Ha526.h.pppool.de] has joined #t2
22:53 < rxr> on osx top needs that much CPU power on this 800Mhz iBook: 16.3%
22:54 < rxr> for 58 running processes (as listed by ps ax)
22:55 -!- tschmidt [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
22:56 -!- tschmidt_ [~tschmidt@p508EB63C.dip.t-dialin.net] has quit [Read error: 104 (Connection reset by peer)]
23:09 -!- mtr_ [~michael@Ha526.h.pppool.de] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
23:16 < rxr> valentin: so far I was not yet able to get into I/O trouble again ...
23:18 < valentin> re
23:18 < valentin> rxr: i'll test it as soon as i booted into OSucX
23:22 < rxr> Read a total of 31 MB in 42 seconds: 0.74MB/s
23:23 < rxr> I added benchmarking to the avtest.cpp ...
23:24 < rxr> that is still at our previously low level under osx ...
23:24 < rxr> under linux we get 2x rates ...
23:24 < rxr> theoretical maximum for this USB 1 port is 12Mbit/8 == 1.5MB/s ...
23:27 < rxr> oehm - f*ck
23:27 < rxr> I have a lot uncommited avision.c core hacking ..
23:27 < rxr> need to deeply review which bits are vital ...
23:28 < rxr> ouhm - my stomarch turns over when I look at that accumulated stalled noise ...
23:28 < rxr> btw I found s.th. very VERY funny today
23:28 < rxr> I reread some parts of the Motorola 68HC12 Reference manual and stumbled over this cool instrcution mnemonic:
23:28 < rxr> EMACS
23:36 -!- tschmidt_ [~tschmidt@p508EAE2A.dip.t-dialin.net] has joined #t2
23:36 < valentin> rxr @private query
23:36 -!- tschmidt__ [~tschmidt@p508EAE2A.dip.t-dialin.net] has joined #t2
23:38 < rxr> to bad the uC does not implement a whole LISP engine
23:39 < rxr> it is just "Extended Multiply and Accumulate (Signed)"
23:54 -!- nzg [~tschmidt@p508EB63C.dip.t-dialin.net] has quit [Read error: 110 (Connection timed out)]
23:58 < rxr> damn - I need some other work place ...
23:58 < rxr> I can not sit anymore ...
23:58 < rxr> damn - need some pult or so to place my iBook on every few hours to hack standing or so ...
23:58 < rxr> heared that should be more healthy ...
--- Log closed Sun Jan 09 00:00:00 2005
00:02 < mnemoc> me?
00:02 < mnemoc> or we*
00:02 < rxr> no - people that have to use it or more in general ...
00:02 < mnemoc> *g*
00:03 < rxr> and you probably because you have to listen to my nonsense ...
00:04 < rxr> 1370 builds total, 666 completed fine, 3 with errors.
00:04 < mnemoc> your nonsense makes my days shorter
00:09 < rxr> ouhm
00:10 < rxr> then I better stop posting it ...
00:10 < rxr> (leaving time to commit - e.g. xmule fixes ;-)
00:11 < mnemoc> it was a 'good thing', bored days feel looooooong
00:12 < mnemoc> thunderbird crashed again :(
00:12 < rxr> oh
00:12 < rxr> I thought you mean I take away time to do real work
00:13 < mnemoc> :)
00:14 < rxr> so - tux.org reply send
00:14 < rxr> I hope we can mirror some files they
00:14 < rxr> but it seems they have not much space to spare ...
00:14 < mnemoc> :(
00:14 < rxr> http://www.tux.org to make my log parser happy and google rate us a slightly higher
00:15 -!- valentin [~valentin@port-212-202-170-162.dynamic.qsc.de] has quit [Read error: 104 (Connection reset by peer)]
00:15 < rxr> huh????
00:15 < rxr> .oO
00:16 -!- valentin [~valentin@port-212-202-170-162.dynamic.qsc.de] has joined #t2
00:17 < mnemoc> huh what?
00:18 < rxr> that valentin ping-ponged ...
00:19 < rxr> but it was not our server
00:19 < rxr> just his client ...
00:19 < rxr> hm - I do not believe this
00:19 < valentin> what did i do ?
00:19 < rxr> hung your irssi ...
00:19 < valentin> yes - but what is 'huh what' ?
00:20 < rxr> 00:15 -!- valentin [~valentin@port-212-202-170-162.dynamic.qsc.de] has quit
00:20 < rxr> [Read error: 104 (Connection reset by peer)]
00:20 < rxr> 00:15 < rxr> huh????
00:20 < rxr> 00:15 < rxr> .oO
00:20 < valentin> ok
00:20 < rxr> this code change:
00:21 < rxr> - // retrieved = -666;
00:21 < valentin> maybe irssi does not like to run for so long
00:21 < rxr> + retrieved = -666;
00:21 < rxr> causes some illegal instruction somewhere - a place the debugger is not willing to yield ...
00:21 < rxr> rw_completed: ret_io_size: -666, io_size: 96
00:21 < rxr> Illegal instruction
00:22 < valentin> what is the next use of retrieved after that initialisation ?
00:22 < rxr> the pointer to that value is passed to WritePipeAsync
00:22 < valentin> is retrieved an instantce variable of some object ?
00:22 < rxr> ReadPipeAsync even
00:23 < rxr> this is the plain C USB wrapper - nothing with objects ...
00:23 < rxr> no this "objects where one never knows where they are" ;-)
00:23 < rxr> it is thown into the library function like this:
00:24 < rxr> (void *)&retrieved);
00:24 < rxr> and arrives there:
00:24 < rxr> rw_completed: ret_io_size: -666, io_size: 96
00:24 < valentin> void* ???
00:24 < rxr> yeah - custom data passed to the callback ...
00:24 < valentin> ok
00:24 < rxr> fprintf(stderr, "rw_completed: ret_io_size: %d, io_size: %d\n", *(u_int32_t *)ret_io_size, (u_int32_t)io_size);
00:24 < rxr> static void rw_completed(void *ret_io_size, io_return_t result, void *io_size)
00:25 < rxr> I only see one place that could crash - if it is ther (I just try to find out) the compiler seems to be generating invalid code ...)
00:26 < rxr> hm - no
00:26 < rxr> it is even thereafter ...
00:26 < rxr> strange
00:26 * rxr puzzled
00:27 * rxr can not imagine what some system library would do to the custom data pointer that could illegal instruction that way ...
00:27 < valentin> try retrieved = @opcode of some legal instruction
00:27 < rxr> hm - or - wait
00:27 < rxr> it is 0x0002
00:27 < rxr> looks quite random ..
00:27 < rxr> well - better the crash dialog claims it is 0002
00:27 < valentin> and it is realy the one line ?
00:27 < rxr> neither the crash helper wants to show me a backtrace - neither the debugger some memory address
00:28 < rxr> yes
00:28 < rxr> when I remove the assignment it works - as before - with our sporadic errors ...
00:28 < valentin> and it crashes imediatly after the '=' ?
00:28 < rxr> later
00:28 < rxr> and I just verified somewhere after the callback returns
00:29 < valentin> i would guess that parameter causes something in that function to overwrite memory
00:29 < valentin> can you read out the address at startup ?
00:29 < rxr> I posted a line above where it is read out ...
00:30 < rxr> it reads out - and is properly assigned:
00:30 < rxr> rw_completed: ret_io_size: -666, io_size: 96
00:30 < rxr> rw_completed: ret_io_size: 96, io_size: 96
00:30 < rxr> rw_completed: RETURNINGIllegal instruction
00:30 < valentin> the address where the instruction is
00:30 < valentin> the crash pointer
00:30 < rxr> come over and show me how to get osx to yield that info ...
00:30 < rxr> I failed so far ...
00:31 < rxr> the codex debugger is not interested in tracing the callback ...
00:31 < rxr> xcode even ...
00:31 < valentin> isn't it in the crashlog ?
00:31 < rxr> no
00:31 < rxr> it does not want to show the backtrace for the affecte thread
00:31 < rxr> I ranted about this 30 minutes or so ago
00:32 < rxr> you are aware that we dump this into #t2 ?
00:33 < valentin> i get logs for each backtrace
00:33 < valentin> yes i am
00:33 < valentin> will raise our google hits
00:34 < rxr> I get a log
00:34 < valentin> but you are not allowed to paste explicit code :)
00:34 < valentin> Exception: EXC_BAD_ACCESS (0x0001)
00:34 < valentin> Codes: KERN_INVALID_ADDRESS (0x0001) at 0x0117ed4b
00:34 < valentin> that is what i get in my crashlogs
00:34 < rxr> but onyl the backtrace of the thread that are not the one with the illegal instruction with this callback thing
00:34 < rxr> click on the t2 html IRC log and read it there
00:34 < valentin> i get backtraces of all threads
00:34 < rxr> I ranted about 20 minutes while I found that out ...
00:34 < rxr> I doo
00:34 < valentin> which log did you mean ?
00:34 < rxr> too even
00:35 < rxr> but not for this case where this has s.th. to to with the callbck
00:35 < valentin> ok, i belive you
00:35 < rxr> this colorful send bug to apple thing ..
00:35 < valentin> no
00:35 < valentin> i mean the crashlogs
00:35 < valentin> in ~/Library/Logs/CrashReporter/
00:35 < rxr> hread: 0
00:35 < rxr> Exception: EXC_BAD_INSTRUCTION (0x0002)
00:35 < rxr> Code[0]: 0x00000002
00:35 < rxr> Code[1]: 0x03020100
00:35 < rxr> Thread 0 Crashed:
00:35 < rxr> Thread 1:
00:35 < rxr> 0 libSystem.B.dylib 0x900074c8 mach_msg_trap + 0x8
00:35 < rxr> 1 libSystem.B.dylib 0x90007018 mach_msg + 0x38
00:35 < rxr> 2 com.unsanity.ape 0xc0002544 __ape_internal + 0xca4
00:35 < rxr> 3 com.unsanity.ape 0xc0001330 __ape_agent + 0x40
00:35 < rxr> 4 libSystem.B.dylib 0x900246e8 _pthread_body + 0x28
00:35 < rxr> do you see a backtrace for the crashed thread 0 ?
00:36 < valentin> no
00:36 * rxr neither
00:36 < rxr> it is really s.th. with this callback or with the RunLoop around it
00:36 < valentin> did you paste that from the reporter or out of your vi ?
00:37 < rxr> I hacked some more printf's into it (hail to the apple debugger and and/or gdb port) and I get nothing after the callback is called
00:37 < rxr> out of the colorful dialog thing ...
00:37 < rxr> why ?
00:37 < valentin> because this dialog may filter some things out
00:37 < valentin> don't you think ?
00:38 < rxr> no idea - so far I hod no problem with it and thought they just place the text into the scroll view ...
00:39 < rxr> ouhm
00:39 < rxr> you are right
00:39 < rxr> Thread 0 Crashed:
00:39 < rxr> 0 libobjc.A.dylib 0x908311f4 objc_msgSend + 0x14
00:39 < rxr> 1 <<00000000>> 0x92f65244 0 + 0x92f65244
00:39 < rxr> 2 <<00000000>> 0x92de4a48 0 + 0x92de4a48
00:39 < rxr> 3 <<00000000>> 0x92e77e30 0 + 0x92e77e30
00:39 < rxr> 4 <<00000000>> 0x92e7eb88 0 + 0x92e7eb88
00:39 < rxr> 5 <<00000000>> 0x92eb8e3c 0 + 0x92eb8e3c
00:39 < rxr> 6 <<00000000>> 0x92e58848 0 + 0x92e58848
00:39 < rxr> 7 <<00000000>> 0x92ec9a4c 0 + 0x92ec9a4c
00:39 < rxr> 8 <<00000000>> 0x92e8ec58 0 + 0x92e8ec58
00:39 < rxr> 9 <<00000000>> 0x92e02e28 0 + 0x92e02e28
00:39 < rxr> 10 <<00000000>> 0x92df54ec 0 + 0x92df54ec
00:39 < rxr> 11 <<00000000>> 0x92dfd904 0 + 0x92dfd904
00:39 < rxr> 12 <<00000000>> 0x92eb9f00 0 + 0x92eb9f00
00:39 < rxr> 13 com.apple.TWAINClientCocoa 0x00009870 main + 0x38
00:39 < rxr> 14 com.apple.TWAINClientCocoa 0x00009568 _start + 0x188 (crt.c:267)
00:40 < rxr> 15 com.apple.TWAINClientCocoa 0x000093dc start + 0x30
00:40 < rxr> but - hm
00:41 < rxr> no that is not the log matching my latest crash
00:41 < rxr> some values and the the thread count are different ...
00:41 < valentin> scroll down
00:41 < valentin> there are multiple crashes in one log (find that quite anoying ...)
00:42 < rxr> OH - YES
00:42 < rxr> do they get anything right ?
00:42 < valentin> maybe that is a feature for dau users
00:42 < rxr> ok - there it is the same:
00:42 < rxr> Thread 0 Crashed:
00:42 < rxr> Thread 1:
00:42 < rxr> ...
00:43 < valentin> that sucks ...
00:45 < rxr> Andréé Watter - Si exceptionne
00:45 < rxr> ^- quite nice french rock
00:45 < rxr> at least s.th. good while hacking ...
00:45 < rxr> valentin: the question is: is this an indication of an more underlying bug?
00:46 < rxr> maybe the cause for other real problem?
00:46 < valentin> could be so
00:46 < rxr> or should I just discard this try to use this for my "run-time-verification" code ß
00:46 < rxr> ? even
00:46 < valentin> there are three explanation for an illegal instruction
00:46 < valentin> (0) broken cpu
00:46 < valentin> 1 broken gcc
00:46 < valentin> 2
00:47 < valentin> 2 overwritten memory
00:47 < valentin> 3 jump at wrong address
00:47 < valentin> i guess it is 2 or 3
00:47 < rxr> well - so far I never had problems with this 750FX ...
00:47 < rxr> I also never overclocked it so far ...
00:48 < rxr> altough it might have suffered due to constant T2 builds and the strange Apple fan control ....
00:48 < rxr> and got some sunburn ...
00:48 < rxr> well - it can not be overwritten memory
00:48 < rxr> if the apple patched gcc generates correct code
00:49 < rxr> the value is anyway assigned to zero at the top of the initiating function
00:49 < rxr> it is a function local variable ...
00:49 < valentin> i did not mean the variable but something that happens in between due to the value of the variable
00:49 < valentin> i do not know the code
00:49 < rxr> darwin.c ...
00:50 < rxr> btw do not do svn up
00:50 < rxr> you will get some broken instrucmentation code that will crash as discussed here ...
00:50 < rxr> better news: 1370 builds total, 685 completed fine, 3 with errors.
00:53 < rxr> valentin: throwing a positive value into the value does not crash this internal usb function
00:53 < rxr> hm - maybe the negative assignment generates incorrect code?
00:53 < valentin> what does a negative value mean ?
00:53 < rxr> e.g. writing a 64 value over the int boundaries or so ...
00:54 < valentin> sure this case does not something evil ?
00:54 < rxr> what case?
00:54 < valentin> the negative value case
00:54 < rxr> well it is applicatoin private date ...
00:55 < rxr> this self pointer thing you normally always have with callbacks
00:55 < rxr> osx should not touch it in any way ...
00:55 < valentin> i need to look into the code first, otherwise that discussion is quite pointless
00:56 < valentin> can you point me to some line in darwin.c ?
00:56 < rxr> well yours is slightly different
00:56 < rxr> browse to usb_bulk_read(
00:56 < rxr> and follow received
00:56 < rxr> retrieved even
00:57 < valentin> ok
00:57 < rxr> I'm soon in bed ...
00:58 < rxr> refCon
00:58 < rxr> Arbitrary pointer which is passed as a parameter to the callback routine.
00:59 < rxr> (void *)&retrieved);
00:59 < valentin> i think this may be a bug in ReadPipeAsync that overwrites ram
00:59 < rxr> " (void *)&retrieved);
00:59 < rxr> oops
00:59 < rxr> "Arbitrary pointer" is for me when the library and system stuff leaves it alone ...
01:00 < valentin> so it is evaluated in the callback only - now i understand
01:00 < rxr> btw. what got out of your prime investigation lately ?
01:00 < valentin> i emailed djb
01:00 < rxr> retrieved = (int32_t)-666;
01:00 < valentin> he did not answer but i did not expect that anyway
01:00 < rxr> rw_completed: ret_io_size: -666, io_size: 96
01:00 < rxr> rw_completed: ret_io_size: 96, io_size: 96
01:00 < rxr> rw_completed: RETURNING
01:01 < rxr> Illegal instruction
01:01 < rxr> as before ...
01:02 < valentin> can you point me to the rw_completed implementation ?
01:02 < rxr> it is some lines about the read and write functions in the same file
01:03 < rxr> btw it does not seem to crash for writes
01:03 < rxr> only for reads ...
01:04 < valentin> the signature is
01:04 < valentin> static void rw_completed(void *ret_io_size, io_return_t result, void *io_size)
01:04 < rxr> ehrm
01:04 < rxr> yes and ?
01:04 < rxr> btw I changed the assignemnts to 666
01:04 < rxr> now i get a segmentation fault ...
01:05 < valentin> with a backtrace ?
01:05 < rxr> Thread 0 Crashed:
01:05 < rxr> 0 de.exactcode.AvisionDS 0x0445abb8 sanei_usb_find_devices + 0x70 (sanei_usb.c:506)
01:05 < rxr> too bad we have no valgrind on max osx, hm ?
01:05 < rxr> mac even ..
01:06 < valentin> back to the signature
01:06 < rxr> as far as I can see there is random memory overwritten somewhere
01:06 < valentin> the callback has 3 arguments
01:06 < rxr> otherwise one value would not cause s SIGILL and another some segmentation fault
01:06 < rxr> sepecially not when it is some "this" pointer for private use ...
01:06 < valentin> that is what i say since 30 minutes
01:07 -!- freed [~uni@pD9E77F0E.dip.t-dialin.net] has joined #t2
01:07 < freed> hi @ll
01:07 < valentin> hi freed
01:07 < freed> hi valentin
01:07 < rxr> typedef void (*IOAsyncCallback1)(
01:07 < rxr> void *refcon,
01:07 < rxr> IOReturn result,
01:07 < rxr> void *arg0);
01:07 < rxr> hi freed
01:07 < freed> hi rxr
01:08 < valentin> but ReadPipeAsync only calls the callback with one parameter ?
01:08 < rxr> refcon
01:08 < rxr> The refcon passed into the original I/O request
01:08 < rxr> result
01:08 < rxr> The result of the I/O operation
01:08 < rxr> arg0
01:08 < rxr> Extra argument
01:08 < valentin> ok
01:09 < valentin> now i understand it
01:09 < rxr> IOReturn (*ReadPipeAsync)(
01:09 < rxr> void *self,
01:09 < rxr> UInt8 pipeRef,
01:09 < rxr> void *buf,
01:09 < rxr> UInt32 size,
01:09 < rxr> IOAsyncCallback1 callback,
01:09 < rxr> void *refcon);
01:10 < rxr> freed: I'm off in bed in some seconds
01:10 < valentin> ahm rxr
01:10 < valentin> your code has to crash
01:10 < rxr> in fact I wanted to crawl into it early ...
01:10 < valentin> you overwrite stack memory
01:10 < rxr> why ?
01:10 < rxr> hm ?
01:10 < valentin> if (result == kIOReturnSuccess)
01:10 < valentin> *(u_int32_t *)ret_io_size = (u_int32_t)io_size;
01:10 < valentin> else
01:10 < valentin> *(u_int32_t *)ret_io_size = result;
01:10 < rxr> this was there ...
01:10 < valentin> but that is a bug
01:10 < rxr> AFAIR I did not tuch it ...
01:11 < rxr> oehm
01:11 < rxr> why =
01:11 < valentin> because ret_io_size points to retrieved
01:11 < valentin> which is a local variable ?
01:11 < valentin> or isn't it ?
01:11 < rxr> it is a local, and ?
01:11 < valentin> ok, it isnt
01:11 < rxr> uuhm?
01:11 < valentin> yes it is
01:12 < rxr> it is a local of the callee ...
01:12 < rxr> but this looks all right
01:12 < valentin> int usb_bulk_read(usb_dev_handle *dev, int ep, char *bytes, int size,
01:12 < valentin> int timeout)
01:12 < valentin> {
01:12 < valentin> ....
01:12 < rxr> they cast it to *int32_t and then dereference it ..
01:12 < valentin> u_int32_t retrieved = 0;
01:12 < valentin> but that is a local variable (!!!!)
01:12 < rxr> they pass the pointer to the function
01:12 < rxr> and?
01:12 < valentin> it points to a stack adress ???
01:12 < rxr> and ?
01:13 < valentin> and ReadAsync calls the callback _after_ usb_bulk read returned, doesn't it ?
01:13 < rxr> nope
01:13 < rxr> ReadAsync RunLoops to wait for the completion
01:13 < valentin> ah - ic
01:14 < rxr> so - still not solution ...
01:14 < valentin> what a shitty code - why do they use ReadAsync then ?
01:15 < rxr> maybe because without the Async the timeout is not controllable ?
01:15 < rxr> freed: how are you?
01:15 < rxr> freed: any AVR Atmel progress on your side ?
01:17 < valentin> rxr: that may be a reason
01:17 < valentin> i am just reading CFRunLoop documentation
01:19 -!- CIA-10 [~CIA@to.je.spocco.com] has quit [Remote closed the connection]
01:22 < rxr> but rereading the code I do not fully understand why to have this RunLoop stuff in it at all
01:24 < rxr> valentin: no the timeout is available with the *To functions
01:25 < rxr> we could try if it makes a difference to strip the Async away and try what happens
01:25 < rxr> but I really need to bed - I do not want to get into such a shifted sleep cycle again ...
01:25 < valentin> there is a function called ReadPipeTO
01:25 < rxr> yes
01:26 < rxr> that is what I meant above ...
01:26 < valentin> with timeout and without Async :)
01:26 < rxr> ack
01:26 < valentin> maybe you can try tomorow using that one
01:26 < rxr> yep - I do ..
01:26 < rxr> so /me off ...
01:26 < valentin> without runloops we do not need in that thread
01:26 < valentin> gn8 rxr
01:27 < rxr> well - but then the gui would hang even more, hm?
01:27 < rxr> so we still might like to keep our io thread ?
01:27 < valentin> yes
01:27 < rxr> what did you mean with "we do not need in that thread" then ?
01:28 < valentin> we do not need to activate the run loop because it is active in the UI thread
01:28 < rxr> yep
01:28 < rxr> hm
01:28 < valentin> i do not even know if RunLoopRun is thread save
01:28 < rxr> well - the docu list "run the current RunLoop"
01:29 < valentin> but is there one run loop for each thread ?
01:29 < rxr> I guess every thread has - it is just useless and most probably empty in our io thing ...
01:29 < valentin> or one run loop per process
01:29 < rxr> the docu speaks of thread ...
01:29 < rxr> and about defautls ..
01:29 < rxr> my conclusion is every thead has a default (most probably empty) one ..
01:30 < valentin> There is exactly one run loop per thread. You neither create nor destroy a thread?s run loop. Core Foundation automatically creates it for you as needed. You obtain the current thread?s run loop with CFRunLoopGetCurrent.
01:30 < rxr> but smart conclusions and osx implementation seems to differ occationaly ... ...
01:30 < valentin> rxr: maybe libusb isn't threadsave ?
01:30 < valentin> they get the run loop only once and use it forever ?
01:30 < valentin> and sane uses bulk_read in different threads ?
01:31 < rxr> 1370 builds total, 705 completed fine, 3 with errors.
01:32 < valentin> ok, there are many calls to CFRunLoopGetCurrent
01:32 < valentin> just remove that runloop / Async stuff tomorow and we will see ...
01:33 < valentin> gn8 rxr
01:36 < rxr> n8 valentin
01:36 < valentin> rxr: still here ?
01:36 < valentin> mom
01:36 < rxr> I hope you go to bed, too ...
01:36 < rxr> yeah - telling freed how to transfer bytes over the network
01:36 < valentin> ReadPipeAsyncTO(device->interface, pipeRef,
01:36 < valentin> bytes + retrieved, size, 0,
01:36 < rxr> .oO
01:36 < rxr> I'm blinded ...
01:36 < rxr> thansk
01:37 < valentin> what happens if that value is negative ?
01:37 < rxr> ugly things ...
01:37 < valentin> take a look tomorow :)
01:37 < valentin> gn8
01:37 < freed> gn8
01:38 < rxr> valentin: what a useless piece of code ...
01:38 < rxr> this retrieved is totally bogus ..
01:38 < rxr> since size not adapted anyway ..
01:38 < rxr> and no loop insight, too ...
01:39 < valentin> hmm
01:39 < valentin> you seem to be rigth
01:39 < valentin> right even
01:39 < rxr> I commented it out ...
01:39 < rxr> now
01:39 < valentin> but this will not fix our IO problems
01:40 < rxr> no
01:40 < rxr> but my runtime verification code ...
01:40 < rxr> thanks for the spot
01:40 < mnemoc> re
01:40 < valentin> only took us 80 minutes *g*
01:40 < valentin> hi menmoc
01:40 < mnemoc> hi valentin
01:40 < rxr> never reuse code not written in house
01:41 < rxr> so - hopefully I can father some data of this run-time check ...
01:41 < rxr> since I only added it to review if the timer ever returns without an associated async event ..
01:43 < rxr> ok - code running again ...
01:43 < rxr> i/o working
01:43 < rxr> preview complete ...
01:43 < rxr> btw: we migth like to review anyway if the runloop code removal brings us speed
01:44 < rxr> maybe the permanent run loop enter for every single USB package causes the big performance impact we see compared to the 2x faster transfer times in linux ...
01:44 < valentin> yep
01:44 < rxr> ouhm
01:44 < rxr> crash
01:45 < rxr> Exception: EXC_BAD_ACCESS (0x0001)
01:45 < rxr> Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x00000028
01:45 < rxr> Thread 0 Crashed:
01:45 < rxr> 0 libobjc.A.dylib 0x908311ec objc_msgSend + 0xc
01:45 < rxr> 1 de.exactcode.AvisionDS 0x0445db2c -[AvisionDS preview:] + 0x78 (AvisionDS.M:382)
01:45 < valentin> maybe that is my code
01:45 < valentin> my scanner never came so far due to IO errors
01:46 < rxr> this was a preview IIRC ...
01:46 < rxr> not a scan
01:46 < valentin> yes i know
01:46 < rxr> hey
01:46 < rxr> valentin:
01:46 < rxr> you do not believe
01:46 < rxr> this
01:46 < valentin> and i never saw a preview using my new multithreaded io error
01:46 < valentin> what ?
01:47 < rxr> o async operation completed. result = -536870165, size = 0
01:47 < rxr> rw_completed: ret_io_size: 3789920, io_size: 0
01:47 < rxr> rw_completed: ret_io_size: -536870165, io_size: 0
01:47 < rxr> rw_completed: RETURNING
01:47 < rxr> CFRunLoop most likely returned without I/O completed!
01:47 < rxr> this is my new checker code
01:47 < rxr> and the code is currently in this always "internal error state" we experienced some times
01:48 < rxr> btw. this is not yet your multithreaded i/o facade ...
01:48 < rxr> I reverted it locally because I thought it caues the segfaults ...
01:48 < rxr> I can svn up again now ...
01:48 < valentin> ok
01:48 < valentin> i am not in OSX currently
01:48 < valentin> so i suggest you remove all this runloop crap
01:49 < valentin> and do an s/Async//g
01:49 < valentin> i guess there are some cocoa timer events getting in the way
01:50 < rxr> Skye Sweetnm - Billy s
01:50 < rxr> is also quite nice
01:50 < valentin> so you can put your mind at rest and go to sleep
01:51 < rxr> in a minute
01:51 < rxr> 1:50 or 2:00 does not matter much ...
01:51 < valentin> i need to drown my cake in chocolade
01:53 < rxr> btw
01:53 < rxr> didn't you need that change:
01:53 < rxr> Skye Sweetnm - Billy s
01:53 < rxr> ehrm
01:53 < rxr> +
01:53 < rxr> + double last_progress;
01:53 < rxr> or did you tweak it locally ?
01:54 < valentin> i just commented it out
01:54 < rxr> valentin: using your threaded wrapper I get dome date
01:55 < rxr> but then this:
01:55 < valentin> because i did not have the declaration
01:55 < valentin> dome date ?
01:55 < valentin> some data ?
01:55 < rxr> well - my i/o core runs:
01:55 < rxr> rw_completed: ret_io_size: -666, io_size: 20304
01:55 < rxr> rw_completed: ret_io_size: 20304, io_size: 20304
01:55 < rxr> rw_completed: RETURNING
01:55 < rxr> [sanei_usb] sanei_usb_read_bulk: wanted 20304 bytes, got 20304 bytes
01:55 < rxr> but your code gets:
01:55 < rxr> 2005-01-08 01:54:34.027 TWAINClientCocoa[909] stream ended because: Error during device I/O
01:55 < rxr> 2005-01-08 01:54:34.132 TWAINClientCocoa[909] scanning ....
01:55 < rxr> and then it crashes ...
01:56 < valentin> when doing what ?
01:56 < rxr> I clicked on scan IIRC
01:56 < valentin> scan does not work
01:56 < rxr> or maybe preview
01:56 < rxr> but I think scan
01:56 < rxr> 0 de.exactcode.AvisionDS 0x044614cc -[AvisionDS transformPreviewDataBytesRGB:] + 0xec (AvisionDS.M:297)
01:56 < rxr> 1 de.exactcode.AvisionDS 0x0446184c -[AvisionDS previewProcessData] + 0x11c (AvisionDS.M:328)
01:56 < rxr> ah - so it was preview ...
01:57 < rxr> yep - preview crahes
01:57 < rxr> was that known ?
01:57 < valentin> i never got any data out of the scanner as i told you 20 minutes before
01:58 < rxr> ok ok
01:58 < valentin> so i did not have a crash
01:58 < valentin> but an empty preview
01:58 < rxr> my brain starts to fall aslepp ...
01:58 < valentin> then go to bed ,)
01:58 < rxr> yeah ..
01:58 < rxr> you just wanna eat your cake alone ...
02:00 < rxr> 1370 builds total, 713 completed fine, 3 with errors.
02:03 * rxr bed ...
02:03 < valentin> gn8
02:06 < rxr> yeah - n8, too
02:07 < mnemoc> n8 rxr
02:52 -!- rxr_ [~rene@p213.54.196.126.tisdip.tiscali.de] has joined #t2
02:52 -!- Topic for #t2: T2 | 2.1.0-beta2 RELEASED | The next generation of System Development Enviroments (SDE) | http://www.exactcode.de/t2
02:52 -!- Topic set by mnemoc [] [Tue Dec 28 14:55:44 2004]
02:52 [Users #t2]
02:52 [ _Ragnar_] [ freed] [ mnemoc] [ praenti] [ rxr_ ] [ valentin]
02:52 [ daja77 ] [ jsaw ] [ mtr_ ] [ rxr ] [ sparc-kly]
02:52 -!- Irssi: #t2: Total of 11 nicks [0 ops, 0 halfops, 0 voices, 11 normal]
02:52 -!- Channel #t2 created Sun Aug 8 21:15:33 2004
02:52 -!- [freenode-info] help freenode weed out clonebots, please register your IRC nick and auto-identify: http://freenode.net/faq.shtml#nicksetup
02:52 -!- Irssi: Join to #t2 was synced in 10 secs
03:02 < mnemoc> *yawn*
03:08 -!- rxr [~rene@p213.54.194.178.tisdip.tiscali.de] has quit [Read error: 110 (Connection timed out)]
04:20 -!- ideal [~idealm@211.100.227.107] has joined #t2
05:43 -!- Postal [~Postal@ool-43562a7f.dyn.optonline.net] has joined #t2
05:43 < Postal> Hey does anyone know the svn update command?
05:45 -!- Postal [~Postal@ool-43562a7f.dyn.optonline.net] has quit [Client Quit]
05:51 < mnemoc> uhm
06:42 < mnemoc> jsaw: does rock^Whwplug use procinfo?
07:11 -!- mnemoc_ [~amery@200.75.27.47] has joined #t2
07:14 -!- uni_ [~uni@pD95D0F8F.dip.t-dialin.net] has joined #t2
07:27 -!- mnemoc [~amery@200.75.27.37] has quit [Read error: 110 (Connection timed out)]
07:32 -!- freed [~uni@pD9E77F0E.dip.t-dialin.net] has quit [Read error: 110 (Connection timed out)]
08:33 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has joined #t2
08:42 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has quit [Read error: 54 (Connection reset by peer)]
08:42 -!- N0V4K_ [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has joined #t2
08:43 -!- N0V4K_ [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has quit [Client Quit]
08:43 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has joined #t2
09:24 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has quit ["FreeBSD N0V4K.Zero 5.3-RELEASE FreeBSD 5.3-RELEASE Chao"]
09:24 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has joined #t2
09:26 -!- N0V4K [~N0V4K@246.Red-80-33-149.pooles.rima-tde.net] has quit [Remote closed the connection]
09:43 -!- sparc-kly [~sparc@64.237.130.154] has quit [Read error: 104 (Connection reset by peer)]
11:59 -!- You're now known as rxr
11:59 < rxr> moin
11:59 < rxr> [5] network/xmule [5] e17/efsd
11:59 < rxr> [5] e17/entrance [5] contrib/joystick
11:59 < rxr> [5] multimedia/blender [5] kde/koffice
11:59 < rxr> [5] graphic/pixieplus [5] kde/superkaramba
11:59 < rxr> [5] scientific/qmatplot [5] audio/xmp
11:59 < rxr> [5] database/mysqlcc [5] multimedia/nvrec
11:59 < rxr> [5] base/ifhp [5] audio/vstserver
11:59 < rxr> [5] multimedia/kaxtv [5] multimedia/mythtv
11:59 < rxr> [5] multimedia/avview [5] audio/ladspa-vst
11:59 < rxr> [5] audio/audacity [5] scientific/xaos
11:59 < rxr> 1370 builds total, 910 completed fine, 20 with errors.
12:03 < mnemoc_> moin rxr
12:04 < rxr> moin
12:05 < mnemoc_> there is no CIA bot neither on #commits
12:05 < rxr> hm
12:08 < mnemoc_> rxr: are you using XMLRPC?
12:08 < mnemoc_> (me talking about cia)
12:08 < rxr> no
12:09 < rxr> xmlmail or how the nammed it ...
12:09 < rxr> I did not wanted some strange python code or so on the server for security considerations
12:09 < rxr> why ?
12:10 < mnemoc_> "Please upgrade if you're still using a client that doesn't send XML messages" <--- on #commits' topic
12:10 < mnemoc_> and http://cia.navi.cx/stats/project/t2 is outdated
12:11 < rxr> /t2/packages/kaxtv.html
12:11 < rxr> ouhm
12:11 < rxr> ^- top #7 hit on our t2 site
12:12 < mnemoc_> o_O
12:12 < rxr> but most of those people that get to that page thru google exit there, too :-(
12:12 < mnemoc_> :\
12:21 < mnemoc_> how can i ask gcc to show me the #include tree it use on a build?
12:21 < rxr> include tree as tree might not provided by gcc
12:22 < rxr> most often -E (only preprocessing is a help
12:22 < rxr> it still contains marks what got included where ...
12:22 < rxr> recently I found -dD (IIRC) helpfull it will dump the preprocessed source including(!) the macro definitions
12:24 < mnemoc_> sounds as what i want :)
12:28 < mnemoc_> brb
12:30 < rxr> well - some stuff fails due to python library dependencies - as usual ...
12:46 < mnemoc_> -E -dD or just -dD ?
12:48 < mnemoc_> what extension is used for preprocessed files?
12:55 < rxr> I always used -E -dD
12:56 < rxr> when you do not supply -o you get it on stdout ...
12:56 < rxr> when -o is present the specified file is used of course ...
12:56 < mnemoc_> i was lazy and added -e -dD to CFLAGS, now i have text .o files :)
13:14 -!- nzg [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
13:14 < nzg> moin
13:15 < rxr> moin nzg
13:15 [Users #t2]
13:15 [ _Ragnar_] [ ideal] [ mnemoc_] [ nzg ] [ rxr ] [ valentin]
13:15 [ daja77 ] [ jsaw ] [ mtr_ ] [ praenti] [ uni_]
13:15 -!- Irssi: #t2: Total of 11 nicks [0 ops, 0 halfops, 0 voices, 11 normal]
13:15 < nzg> that damn box is down again....
13:15 < rxr> ouch
13:16 < nzg> this time it will be my boss who calls ;-)
13:16 < rxr> ,)
13:31 * nzg is away: einkaufen
13:33 < nzg> sry
13:34 -!- uni_ [~uni@pD95D0F8F.dip.t-dialin.net] has quit [Remote closed the connection]
13:38 < rxr> 1370 builds total, 998 completed fine, 22 with errors.
13:46 < rxr> # ./debug.sh
13:46 < rxr> Segmentation fault (core dumped)
13:46 < rxr> I thought this was solved/work arounded ... ?
13:59 < rxr> kkk1370 builds total, 1019 completed fine, 23 with errors.
14:01 < mnemoc_> i thought it was
14:08 -!- valentin [~valentin@port-212-202-170-162.dynamic.qsc.de] has quit [Read error: 60 (Operation timed out)]
14:08 < mnemoc_> rxr: problems with ec.de server?
14:15 < jsaw> re
14:17 < mnemoc_> re jsaw
14:17 < jsaw> hi mnemoc_
14:18 < jsaw> hmm. again a segfaulting bash... o_O
14:18 < mnemoc_> yep
14:18 < jsaw> rxr: what does valgrind say?
14:18 < mnemoc_> OT: which extension is used for gcc -E output files?
14:19 < jsaw> .d iirc
14:19 < rxr> jsaw: later
14:19 * rxr breakfast ...
14:19 < jsaw> oki
14:23 < mnemoc_> svn: PROPFIND of '/t2/trunk': could not connect to server (http://svn.exactcode.de)
14:25 < jsaw> dito
14:26 < rxr> yep
14:26 < rxr> ack
14:29 < jsaw> btw, I assume, the uselib patch for 2.4 applies to 29-rc... have to check again.
14:29 -!- valentin [~valentin@port-212-202-42-228.dynamic.qsc.de] has joined #t2
14:29 < valentin> test
14:30 < valentin> so we seem to be back online
14:30 < mnemoc_> yes
14:31 < valentin> my gf was cleaning up and stumbled over the cable
14:31 < mnemoc_> hehe
14:32 < jsaw> rxr: will you push the new .cache files into trunk?
14:32 < valentin> cu later
14:32 < jsaw> cu valentin
14:48 < rxr> jsaw: sorry - this was a in-system build of a package on my athlon which had nto yet an up-to-date'ly patched bash installed ...
14:48 < rxr> 1370 builds total, 1045 completed fine, 25 with errors.
14:49 < rxr> [5] network/xmule [5] e17/efsd
14:49 < rxr> [5] e17/entrance [5] contrib/joystick
14:49 < rxr> [5] multimedia/blender [5] kde/koffice
14:49 < rxr> [5] graphic/pixieplus [5] kde/superkaramba
14:49 < rxr> [5] scientific/qmatplot [5] audio/xmp
14:49 < rxr> [5] database/mysqlcc [5] multimedia/nvrec
14:49 < rxr> [5] audio/vstserver [5] multimedia/kaxtv
14:49 < rxr> [5] multimedia/mythtv [5] multimedia/avview
14:49 < rxr> [5] audio/ladspa-vst [5] audio/audacity
14:49 < rxr> [5] scientific/xaos [5] network/slidentd
14:49 < rxr> [5] multimedia/ripperx [5] network/centericq
14:49 < rxr> [5] contrib/djvulibre [5] develop/clip
14:49 < rxr> [5] security/ebtables
14:49 < rxr> jsaw: do you want the .cache files beeing injected?
14:50 < mnemoc_> clip?
14:50 < rxr> I just looked
14:50 < rxr> download missing
14:50 < mnemoc_> get it from my mirrors :)
14:52 < mnemoc_> OT: what do you think if Config's choice is able to change default if that key is not on the option list taking the first/last key as new default?
14:54 < jsaw> rxr: no, I was just asking. Only thing I'd really like to know, when the copyright headers will be changed... if it's more than 1week from now, I still have time to start a build.
14:55 < jsaw> anything against linux 2.4.28 -> 2.4.29-rc1???
14:56 < mnemoc_> considering it's a security need i guess we should do it
14:57 < jsaw> (kernel build running...)
14:57 < rxr> well -rc1 should be quite stable so - go for it .. ;-)
14:58 < rxr> INFO: Testing
14:58 < mnemoc_> what about the bleeding 2.6.10 linux-header?
14:58 < rxr> mnemoc_: if you wanna do the patchwork ...
14:59 < mnemoc_> i don't have enough knowledge to #ifdef __KERNEL__ properly
14:59 < rxr> jsaw: well - if we agree about the copyright one day we could inject them ...
14:59 < jsaw> hehe
15:00 < rxr> then let's get some -beta3 out, first before we spend time sanitizing newer kernel headers ...
15:00 < rxr> mnemoc_: thanks for providing mirrors - now that all .de mirrors are down ...
15:01 < mnemoc_> oh
15:01 < mnemoc_> is that coincidence?
15:01 < rxr> I just announced the http://tfh-berlin.de downtime to the list
15:01 < rxr> tfh-berlin.de was planed - but I forgot to announce that
15:02 < rxr> they do some construction work and cut some inter-campus data wires or so
15:02 < rxr> but I do not know the strange details about the stuff happening at my university of applied science ...
15:03 < rxr> the only cunstruction work I can see is a new plant growing glas house ..
15:04 < rxr> no idea how this interferes with data-wires ...
15:04 < rxr> but how knows ...
15:04 < rxr> who even
15:04 < nzg> maybe they whacked some wires when they grounded the greenhous
15:04 < rxr> jsaw: yep - a Emerge-Pkg bash solved the seg-fault ..
15:04 < rxr> no - this already happend weeks ago ...
15:05 < jsaw> ?
15:05 < rxr> they even announced the downtime some days ago ...
15:05 < rxr> the greenhouse even finished - AFAICS only the plans are missing or so ...
15:06 < rxr> jsaw: I could inject the .cache files - but as I wrote some times already, I do not like the noisy nature of the .cache files
15:06 < rxr> I shomehow hesitate to do such monster commits
15:06 < rxr> do you remeber my plans for that files?
15:06 < rxr> and should we start the migration work now ?
15:06 < jsaw> rxr: to be honest, I don't want that noise either.
15:07 < mnemoc_> we need an updated .cache set to migrate
15:07 < jsaw> well, right. That's the other side...
15:08 < rxr> well - the updated set would not have to be commited
15:08 < jsaw> rxr,mnemoc_: would it be enough to keep only "[DEP]" information in the cache files, or will some scripts break?
15:09 < rxr> instead I could directly generate .dep and new-styl .cache files ...
15:09 < rxr> where the new style .cache files would contain less unneeded noisy info ...
15:09 < mnemoc_> do you really have time to do that?
15:09 < rxr> and be rounded so some exponentially limitted value so tine changes would not alter them ..
15:09 < rxr> no
15:10 < mnemoc_> so?
15:10 < rxr> so - you could do it :-)
15:10 < mnemoc_> not today :\
15:11 < jsaw> mnemoc_: rule number one in this channel: never ask such things, otherwise you end up doing it yourself ;)
15:11 < mnemoc_> :)
15:11 < rxr> well - sure - I could inject new .cache files one last time and we do the migration when time permits ...
15:12 < rxr> /usr/include/linux/dvb/dmx.h:164: error: `__u64' does not name a type/usr/include/linux/dvb/dmx.h:164: error: extra semicolon
15:12 < rxr> ieeee
15:12 < rxr> I hate linux-header oddities
15:13 < mnemoc_> santizing 2.6.8 is senseless
15:13 < rxr> nope - it was the newest release when we did it - and all the later kernels only got more broken with each release ...
15:14 -!- ideal [~idealm@211.100.227.107] has quit ["Leaving"]
15:14 < mnemoc_> we have to get used to that
15:14 < mnemoc_> until hurd gets decent ... 10 years?
15:15 < rxr> well - I do much work - but that is really s.th. I do not volunteer to do
15:15 < rxr> I do not say we shoud keep those 10 years - just that is a very low priority for me ...
15:15 < rxr> of course you, ...
15:18 < mnemoc_> i have loooooot to learn before patching kernel files
15:19 < rxr> mnemoc_: I see only negative time windows in january over here ...
15:19 < rxr> if you can wait until february i can do it then ...
15:19 < rxr> alternativly you can find someone paying me as much as i get for the osx driver so I can stop hacking on it right now ...
15:21 < mnemoc_> :)
15:21 < rxr> but hey - on the other side, if this driver gets a success I can hack on whatever I want the remaining whole year ...
15:24 < jsaw> no CIA around anymore?
15:25 < rxr> no
15:25 < rxr> need to migrate mail xmlrpc ...
15:31 < rxr> migrate from mail to ...
15:31 < rxr> sorry was phoning
15:37 < mnemoc_> what do you think is better, add GROUPS_MAX to uclibc's limits.h or include linux/limits.h on gawk's awk.h?
15:43 < rxr> well - it if is normally in limits.h I would consider it a uclibc bug and patch it accoringly ...
15:44 < rxr> I could grep my BSD based Apple system if it is a help ..
15:45 < rxr> /usr/include/limits.h:#define _POSIX_NGROUPS_MAX 0
15:45 < rxr> /usr/include/sys/syslimits.h:#define NGROUPS_MAX 16 /* max supplemental group id's */
15:45 < rxr> /usr/include/sys/param.h:#define NGROUPS NGROUPS_MAX /* max number groups */
15:46 < mnemoc_> param.h includes linux/limits.h
15:46 < rxr> what is the actual define you are looking for
15:46 < rxr> ?
15:46 < mnemoc_> NGROUPS_MAX
15:46 < rxr> my rs6k also has no GROUPS_MAX in limits.h
15:46 < rxr> a - NGROUPS_MAX
15:47 < rxr> yep - that is in both - my OSX and T2/RS6K limits.h ...
15:47 < mnemoc_> libc's ?
15:47 < rxr> hm
15:47 < rxr> or - wait
15:48 < mnemoc_>
15:48 < rxr> well - actually it is in sys/param.h
15:49 < rxr> it is in some way defined when glibc is used - so just patch uclibc to define it "somewhere"
15:49 < mnemoc_> aha
15:49 < rxr> and it is here:
15:49 < rxr> /usr/dietlibc/include/limits.h:#define NGROUPS_MAX 32 /* supplemental group IDs are available */
15:49 < rxr> ;-)
15:50 < mnemoc_> how should i decide the limit?
15:50 < rxr> so far I only saw 32
15:50 < rxr> so just use it to ...
15:51 < mnemoc_>
15:51 < mnemoc_> ok
15:51 < rxr> the Open Group or so wants money for the Posix SPEC - so ...
15:54 < mnemoc_> http://www.opengroup.org/onlinepubs/000095399/toc.htm
15:58 < rxr> oh - good ;-)
16:00 < jsaw> bbl
16:01 < mnemoc_> cu jsaw
16:01 < mnemoc_> rxr: http://www.uclibc.org/cgi-bin/cvsweb/uClibc/include/limits.h?rev=1.5&view=auto <--- their limits.h is amazingly short
16:24 < rxr> then make it a line longer ;-)
16:25 < mnemoc_> NAME_MAX for util-linux is also missing
16:26 < rxr> two lines then ;-)
16:26 < mnemoc_> i'm waiting for uclibc's answer
16:26 < mnemoc_> but i guess your way is the safer
16:27 < mnemoc_> uhm PATH_MAX and LINE_MAX
16:28 < mnemoc_> somethign must be very wrong somewhere
16:45 < rxr> [5] network/xmule [5] e17/efsd
16:45 < rxr> [5] e17/entrance [5] contrib/joystick
16:45 < rxr> [5] multimedia/blender [5] kde/koffice
16:45 < rxr> [5] graphic/pixieplus [5] kde/superkaramba
16:45 < rxr> [5] scientific/qmatplot [5] audio/xmp
16:45 < rxr> [5] multimedia/nvrec [5] audio/vstserver
16:45 < rxr> [5] multimedia/kaxtv [5] multimedia/mythtv
16:45 < rxr> [5] multimedia/avview [5] audio/ladspa-vst
16:45 < rxr> [5] audio/audacity [5] scientific/xaos
16:45 < rxr> [5] multimedia/ripperx [5] contrib/djvulibre
16:45 < rxr> [5] security/ebtables [5] network/ngrep
16:45 < rxr> [5] network/asterisk [5] network/kismet
16:45 < rxr> [5] network/licq
16:45 < rxr> 1370 builds total, 1085 completed fine, 25 with errors.
16:46 < rxr> who is guilty for this one:
16:46 < rxr> Apply patch /TOOLCHAIN/package/network/kismet/gcc34.patch ...patching file packet.hHunk #1 FAILED at 550.Hunk #2 FAILED at 643.2 out of 2 hunks FAILED -- saving rejects to file packet.h.rej
16:46 < rxr> ?
16:46 < mnemoc_> svn log /package/network/kismet
16:47 < rxr> sure
16:47 < rxr> hm - only /me touched it ...
16:47 < rxr> damn
16:47 < mnemoc_> *G*
16:48 < rxr> I remeber I once deleted the patches - maybe I have just dreamed that and forgot the commit ...
16:48 < mnemoc_> rm instead of svn rm ;)
16:48 < rxr> yeah - like that ...
16:58 < mnemoc_> How Not to Program in C++: 111 Broken Programs and 3 Working Ones, or Why Does 2+2=5986
16:58 < mnemoc_> by Steve Oualline
17:09 < rxr> mnemoc_: url ?
17:09 < mnemoc_> closed, but wait
17:17 < rxr> 1370 builds total, 1102 completed fine, 26 with errors.
17:18 < mnemoc_> :)
17:18 < mnemoc_> your machine is fast
17:19 < rxr> 2.5 pseudo Mhz Athlon
17:19 < rxr> but also already a year old - or no - even older I think ...
17:19 < rxr> Ghz even ...
17:20 < rxr> let's hope I have some bucks for a x86-64 board this year ...
17:20 < mnemoc_> i took 3 days to build 1100 on a piv/2566/512
17:20 < rxr> yeah - P4 ...
17:20 < rxr> I only invested once in my life into a state-of-the-art Intel machine - and that was a Pentium 120 ...
17:21 < rxr> since then I only handed money to AMD, IDT (Winchip 2A) and Motorola/IBM ...
17:22 < mnemoc_> me life was: 8088 -> 8086 -> 80286 -> 486sx -> 486dx4 -> p2/300 -> p3/733 -> p4/2566 (laptop)
17:30 < rxr> Subject: [ANNOUNCE] linux-libc-headers 2.6.10.0
17:30 < rxr> linux kernel ...
17:31 < rxr> http://ep09.pld-linux.org/~mmazur/linux-libc-headers/
17:32 < mnemoc_> :D
17:32 < mnemoc_> will we take pld-linux' work?
17:35 < rxr> no idea
17:35 < rxr> maybe take a look if there is s.th. useful ...
17:37 -!- tschmidt [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
17:39 < tschmidt> re
17:39 < mnemoc_> hi tschmidt
17:40 < tschmidt> "you must boot your system into rescu mode" "it didn't work the last three times" "you must ...."
17:40 < tschmidt> hrmpf
17:40 < rxr> hmgrmpf ...
17:41 < tschmidt> if thats germany's best hoster - how bad must the rest be
17:44 < rxr> any answer why the box does /crash/ so often ?
17:46 < tschmidt> no...
17:47 < tschmidt> thats what the admins wrote in an e-mail:
wir haben gesehen das bei Ihrer Maschine die Daten von Bios und 3ware Raid Controller nicht mehr auf dem aktuellsten Stand waren. Diese haben wir geupdatet. Sonst konnten wir keinen Fehler entdecken. Ich hoffe wir konnten Ihnen hiermit weiterhelfen.
17:48 -!- tschmidt_ [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
17:48 -!- tschmidt [~tschmidt@p508EB63C.dip.t-dialin.net] has quit [Read error: 104 (Connection reset by peer)]
17:49 < mnemoc_> rxr: can you give a try to Config -> mnemosyne? does it feel too slow? (compared to tipical configs)
17:50 < rxr> oehm
17:50 < rxr> later, ok?
17:50 < mnemoc_> sure
17:50 < mnemoc_> any day ;)
17:50 < rxr> I still USB hack on osx ...
17:52 < tschmidt_> so i have no clue
17:52 < tschmidt_> there is nothing in the logfiles that explains the crashes
17:55 < rxr> so - bad hardware, hm ?
17:56 < tschmidt_> thats what i suppose - it crashes disregarding to kernel-version or entire distribution
17:57 < tschmidt_> and i can't reproduce or provoke the crashes
17:58 -!- tschmidt_ is now known as tschmidt
17:58 < rxr> this damn xcode
17:59 < mnemoc_> rxr: is procinfo needed by rockplug?
18:00 < rxr> oehm
18:00 < rxr> i do not know all requirements of rockplug off-hand
18:00 < rxr> why do you ask=
18:00 < rxr> ? even
18:01 < mnemoc_> to know if i can remove it from mnemosyne-minimal
18:02 < mnemoc_> i'll try on qemu after lunch
18:08 * mnemoc_ moving home
18:09 * tschmidt calling S+P again...
18:20 < rxr> mnemoc_: I think it should work without procutils whatever - IIRC rockplug has very few dependencies ...
18:20 < rxr> tschmidt: :-(
18:21 < tschmidt> now in the line for 15min...
18:21 < tschmidt> still no one picked up
18:26 < tschmidt> wow that guy at least seemed sympathic "i'm gonna kick the admins ass so he really checks the hardware"
18:31 < rxr> .oO
18:31 < tschmidt> i bet h'll just push the rset button
18:32 < tschmidt> want to hold against?
18:32 < rxr> no
18:33 < tschmidt> ;-)
18:39 < rxr> valentin: I think you new code has some problems
18:39 < rxr> preview does crash - with the last commit revert it works with my new USB i/o code ...
18:40 < rxr> ouhm - no query, ...
18:41 < rxr> hm - i/o thruput does not seem to have improved :-(
18:53 < rxr> hm - ok - without the excessive debugging it feals at least slightly faster ...
18:54 -!- tschmidt_ [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
18:54 -!- tschmidt [~tschmidt@p508EB63C.dip.t-dialin.net] has quit [Read error: 104 (Connection reset by peer)]
18:56 * rxr most probably a bit away soon ...
18:58 < rxr> ScannerManager.h
18:58 < rxr> oops
18:58 < rxr> 1370 builds total, 1125 completed fine, 27 with errors.
19:00 < rxr> cu
20:22 -!- sparc-kly [~sparc-kly@64.237.130.154] has joined #t2
22:14 < rxr> re
22:15 < rxr> hi sparc-kly
22:27 < rxr> 1370 builds total, 1204 completed fine, 32 with errors.
22:28 < tschmidt_> rxr, ist back again
22:28 < tschmidt_> s/ist/its
22:28 < rxr> ;-)
22:28 < rxr> and, any additional information ?
22:29 < tschmidt_> not yes - they promised to send an e-mail
22:29 < rxr> U.S. Army Research Lab Opens BRL-CAD Source
22:29 < rxr> I just visited the side via freshmeat yesterday (IIRC) and wondered what it is about ...
22:29 < rxr> not slashdot took it up
22:30 < tschmidt_> if this promise is as reliable as the last ones...
22:30 < tschmidt_> what is brl-cad?
22:30 < rxr> never heared of it before - it is some CAD application
22:31 < rxr> http://brlcad.org/
22:33 < tschmidt_> it can be run on a pdp11 ;-)
22:35 < rxr> more likely could - I hope the current code base is too advanced for such an old beast
22:35 < rxr> otherwise the CAD software would be pretty useless nowadays ...
22:36 -!- mtr_ [~michael@H831b.h.pppool.de] has quit [Read error: 60 (Operation timed out)]
22:38 -!- mtr_ [~michael@Ha526.h.pppool.de] has joined #t2
22:53 < rxr> on osx top needs that much CPU power on this 800Mhz iBook: 16.3%
22:54 < rxr> for 58 running processes (as listed by ps ax)
22:55 -!- tschmidt [~tschmidt@p508EB63C.dip.t-dialin.net] has joined #t2
22:56 -!- tschmidt_ [~tschmidt@p508EB63C.dip.t-dialin.net] has quit [Read error: 104 (Connection reset by peer)]
23:09 -!- mtr_ [~michael@Ha526.h.pppool.de] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
23:16 < rxr> valentin: so far I was not yet able to get into I/O trouble again ...
23:18 < valentin> re
23:18 < valentin> rxr: i'll test it as soon as i booted into OSucX
23:22 < rxr> Read a total of 31 MB in 42 seconds: 0.74MB/s
23:23 < rxr> I added benchmarking to the avtest.cpp ...
23:24 < rxr> that is still at our previously low level under osx ...
23:24 < rxr> under linux we get 2x rates ...
23:24 < rxr> theoretical maximum for this USB 1 port is 12Mbit/8 == 1.5MB/s ...
23:27 < rxr> oehm - f*ck
23:27 < rxr> I have a lot uncommited avision.c core hacking ..
23:27 < rxr> need to deeply review which bits are vital ...
23:28 < rxr> ouhm - my stomarch turns over when I look at that accumulated stalled noise ...
23:28 < rxr> btw I found s.th. very VERY funny today
23:28 < rxr> I reread some parts of the Motorola 68HC12 Reference manual and stumbled over this cool instrcution mnemonic:
23:28 < rxr> EMACS
23:36 -!- tschmidt_ [~tschmidt@p508EAE2A.dip.t-dialin.net] has joined #t2
23:36 < valentin> rxr @private query
23:36 -!- tschmidt__ [~tschmidt@p508EAE2A.dip.t-dialin.net] has joined #t2
23:38 < rxr> to bad the uC does not implement a whole LISP engine
23:39 < rxr> it is just "Extended Multiply and Accumulate (Signed)"
23:54 -!- nzg [~tschmidt@p508EB63C.dip.t-dialin.net] has quit [Read error: 110 (Connection timed out)]
23:58 < rxr> damn - I need some other work place ...
23:58 < rxr> I can not sit anymore ...
23:58 < rxr> damn - need some pult or so to place my iBook on every few hours to hack standing or so ...
23:58 < rxr> heared that should be more healthy ...
--- Log closed Sun Jan 09 00:00:00 2005