[00:52:17] <FabienD> peda_: I think I have almost solved the above problem with svn mirroring, at least locally (without scanning all svn revs)
[00:52:44] <FabienD> but while doing push I have :
[00:52:48] <FabienD> To git at gitorious dot org:jnode/svn-mirror.git
[00:52:50] <FabienD>  ! [rejected]        master -> master (non-fast forward)
[00:52:51] <FabienD> error: failed to push some refs to 'git at gitorious dot org:jnode/svn-mirror.git'
[00:53:09] <FabienD> do you know what's wrong ?
[00:58:44] <peda_> hmm
[00:58:47] <peda_> no
[01:12:10] <FabienD> I have been told to read that : http://git.wiki.kernel.org/index.php/GitFaq#non-ff
[01:13:11] <FabienD> it appear that the conflict is due to missing svn revs from 1 to <the-svn-rev-have-started-from> in my local repo
[02:24:07] *** FabienD has quit IRC
[18:17:19] *** FabienD has joined #JNode.org
[18:46:10] *** loic has joined #JNode.org
[18:47:56] <loic> hi all
[18:48:31] <loic> peda_: what's the panic int 31
[18:48:33] <loic> ?
[18:53:01] <loic> FabienD: any ideas?
[18:54:07] <loic> i'm sorry to hurry but i won't last long on irc this evening and then, i won't be able to work until monday
[18:57:49] <peda_> 31 or 0x31?
[19:05:26] <peda_> well 31 wouldn't make much sense and 0x31 should be a stack overflow error
[19:09:20] <loic> 0x31
[19:09:35] <loic> peda_: ok for the stackoverflow ;)
[19:10:20] <loic> What is jnode stack size?
[19:10:22] <loic> 64 ko?
[19:10:38] <peda_> I am not sure
[19:10:56] <peda_> but it depends on the context (kernel vs user context)
[19:11:11] <peda_> let me search
[19:12:45] <peda_> 16*1024
[19:14:38] <loic> and the unit?
[19:14:47] <peda_> slots
[19:14:47] <loic> kB?
[19:14:55] <loic> what is a slot?
[19:15:00] <peda_> i.e. 4byte for 32bit, 8byte for 64bit
[19:15:17] <loic> ok
[19:15:21] <loic> just a pointer
[19:17:04] <peda_> actually not.. that's a bit more difficult :)
[19:17:32] <peda_> the language spec says, a byte,char,short,int,float and Ref Item will occupy "one slot"
[19:17:42] <peda_> double and long take two slots
[19:18:12] <peda_> so in order to make that work on 64bit systems without changing the spec you have to make slotsize 8bytes instead of 4
[19:20:39] <FabienD> wow !
[19:21:04] <FabienD> so an array of booleans is very expensive on 64 bits systems
[19:21:18] <peda_> yup
[19:21:26] <peda_> but imho on 32bit systems too :)
[19:21:44] <peda_> though I have no idea if e.g. Sun has some clever optimizations for that
[19:22:06] <peda_> or perhaps they only allow ~32bit address space for apps on 64bit
[19:46:04] <loic> could we use the System.currentTimeMillis() ?
[19:46:23] <loic> or are there any other way to get time in ms
[19:46:49] <loic> when i speak of time, it doesn't matter if its a the date time or a real time clock
[19:49:27] <peda_> currentTimeMillis is ok, nanoTime is ok
[19:49:45] <peda_> and there's another one in Unsafe. that tells you the ticks the CPU did since bootup
[19:55:34] <loic> peda_: what about this http://ups.imagup.com/07/1264839137.png ?
[19:55:54] <loic> i have no idea of what's happening
[19:57:10] <peda_> do you know how to activate kdb?
[19:57:16] <peda_> so that it writes to a file?
[19:57:25] <peda_> I'd need a "full" stacktrace to be sure
[19:58:06] <peda_> ok wait
[19:58:14] <peda_> the call to currentTimeMillis is too early
[19:58:17] <FabienD> it can also write to a telnet console ;)
[19:58:32] <FabienD> i= kvm
[19:58:37] <FabienD> it= kvm
[19:58:51] <FabienD> (and probably qemu too)
[19:59:26] <peda_> it seems the currentTimeMillis needs to be compiled.. jit compilation needs memory allocation which again triggers somehow your currentTimeMillis call
[19:59:35] <peda_> so recursive call-loop -> stackoverflow
[19:59:47] <peda_> FabienD: afair kvm and qemu have merged
[20:00:01] <peda_> they're no different anylonger
[20:01:11] <peda_> loic: if you just need some "timing value" use the Unsafe.getrscd value
[20:01:54] <peda_> sry, Unsafe.getCpuCycles
[20:02:05] <peda_> that does not trigger allocation for sure :)
[20:03:07] <peda_> oh and VmSystem.currentKernelMillis() is ok too
[20:03:17] <peda_> as VmSystem is contained within the bootimage
[20:05:33] <loic> thanks
[22:34:00] *** loic has quit IRC

top