IRC Logs for #crux-devel Wednesday, 2007-09-05

*** sepen has quit IRC01:05
*** sepen has joined #crux-devel01:05
*** sepen has quit IRC01:12
*** sepen has joined #crux-devel02:44
*** pedja has quit IRC07:10
*** deus_ex has joined #crux-devel07:56
*** deus_ex is now known as pedja07:56
*** jaeger has joined #crux-devel08:15
*** jjpk has joined #crux-devel08:59
jaegernifty... with that external usb thing and syslinux I can boot the updated iso from an HD09:58
jaegerit'd take a fair bit of modification to actually start the install from there but syslinux at least gives a boot menu09:59
tilmanmmh, juergen put opessh 4.6p1 in core on 2007-03-1110:04
tilmanopenssh.org sez: 4.6p1 released sep 4, 200710:04
tilmanstrangeness :P10:04
jaegerodd10:07
tilmanokay, seems 4.7 is out10:07
*** sepen has quit IRC10:10
clbUpdate from core: 5 Sep 15:22 - openssh: updated to 4.7p1.10:36
clbUpdate from xorg: 5 Sep 15:34 - xorg-xf86-video-nsc: updated to 2.1.3.10:56
*** pedja has quit IRC11:05
*** jjpk has quit IRC12:49
*** jjpk has joined #crux-devel12:51
jaegerhrmm... I really need to sit down and figure out why the kernel gets remade every time "make iso" is called13:20
clbUpdate from xorg: 5 Sep 18:24 - xorg-xf86-video-mga: updated to 1.4.7.13:26
jaegerI guess I'll peruse the make -d log manually :P13:31
*** treach has joined #crux-devel13:38
tilmanjaeger: check whether the mtime of kernel/linux-$(KERNEL_VERSION).config changes14:10
tilmanthe other targets look less likely to be a candidate14:11
jaegerI've got the log to look at, just haven't done it yet14:11
jaegerhttp://jaeger.morpheus.net/proxy/iso-build.log14:11
tilmanfilthy large14:11
treachsite #cruxed? :P14:12
treach~30k/s here. :)14:12
jaegerI'll gzip it if you like14:13
jaegerdone14:14
tilman     Prerequisite `kernel/linux-2.6.22.1' is newer than target `kernel/linux-2.6.22.1/arch/i386/boot/bzImage'.14:16
tilman     Prerequisite `kernel/linux-2.6.22.1.config' is older than target `kernel/linux-2.6.22.1/arch/i386/boot/bzImage'.14:16
jaegersomething about kernel/linux-$KERNEL_VERSION being newer than bzImag14:16
tilman    Must remake target `kernel/linux-2.6.22.1/arch/i386/boot/bzImage'.14:16
jaegeryeah14:16
jaegerjust got there, heh14:16
tilmani think it updated the mtime of the directory when it created the bzImage14:16
jaegerline 85 of the Makefile touches bzImage14:17
jaegerafter the build14:17
jaegerwonder if I should remove that14:25
tilmandon't think it will hlp14:26
tilmanjaeger: try to remove the kernel/linux-$(KERNEL_VERSIO) target14:26
tilmanie the one that extracts the kernel14:26
jaegerthe touch shouldn't really be necessary anyway14:27
tilmanyep14:27
tilmanseems bogus to me, too14:27
jaegerit was in the original makefile if I remember right14:28
jaegertilman: the extraction step is kinda useful, though :)14:29
tilmanyes. find a cool way to work around it14:29
tilmanmaybe in the bzimage target:14:29
tilmanif not exist le-dir: extract14:29
jaegerthis doesn't occur in the official iso build, though14:32
jaegerat least, I don't remember this problem14:32
jaegerand it does have the extract phase14:33
jaegeras well as the touch :P14:33
* jaeger scratches his head14:34
*** deus_ex has joined #crux-devel15:05
jaegerok, somewhere after the kernel target, kernel/linux-2.6.22.1 gets its mtime updated15:22
jaegernot sure why15:22
tilmanmy guess is because a file in that dir got updated/written15:23
jaegeryeah, maybe when modules are installed in the squashfs section or something15:24
tilmanno15:25
tilmanit's a file system feature15:25
tilmanor vfs15:25
tilmanmkdir /tmp/foo15:25
tilmanls -ld /tmp/foo15:25
jaegerright, but I'm trying to figure out which part triggers it15:25
tilmanecho ... > /tmp/foo/bar15:25
tilmanjaeger: building the kernel does15:26
jaegerbut this problem doesn't exist in the official build process, only in my updated one15:26
jaegerif you don't change anything in the kernel/ tree between builds, the kernel target should not be remade15:27
jaegermaybe I'm not understanding your point15:30
tilmanthe fact that it only happens in your tree might nullify my theory15:31
tilmanbut i think what happens is this:15:31
tilman1.) kernel/bzImage has an mtime of x, but kernel/ has an mtime of x+115:32
tilman2.) directory is newer than bzImage, so we rebuild bzImage15:32
tilman3.) the new bzImage has an mtime of y, so kernel/ gets this new mtime as well15:32
tilmanmmh15:33
tilmanthere goes the theory15:33
tilmanwhen the mtimes are equal, the target shouldn't trigger15:33
tilmanbleh15:33
jaegeryeah =/15:33
jaegerif I manually touch kernel/linux-$(KERNEL_VERSION).config and kernel/linux-$(KERNEL_VERSION)/arch/i386/boot/bzImage it does what I expect15:36
jaeger(not rebuild kernel)15:36
*** deus_ex has quit IRC15:36
*** deus_ex has joined #crux-devel15:45
tilmanjaeger: maybe diffing working-makefile vs. bad-makefile can give you a clue?15:45
tilmanor is insanely huge? :P15:45
jaegerthe changes are pretty big15:46
jaegerI'm doing some checks during the build now15:46
*** maro has joined #crux-devel16:00
jaegertilman: the squashfs target is where it's happening16:01
jaegermy guess is line 235:16:02
jaeger@make -C kernel/linux-$(KERNEL_VERSION) INSTALL_MOD_PATH=$(SQUASHFS_DIR) modules_install &> modules_install.log16:02
jaegerhttp://jaeger.morpheus.net/proxy/dirtime.txt16:02
jaegerhttp://jaeger.morpheus.net/proxy/dirtime2.txt16:05
jaegerlooks like that's the problem16:05
jaegerI still don't get why this doesn't happen with the official build16:10
*** deus_ex_ has joined #crux-devel16:32
jaegerworked around for now16:37
*** deus_ex has quit IRC16:39
*** jjpk has quit IRC16:42
*** jjpk has joined #crux-devel16:44
*** deus_ex_ is now known as pedja16:45
*** jjpk has quit IRC16:46
*** jaeger has quit IRC17:26
*** treach has quit IRC17:33
*** pedja has quit IRC17:47
*** deus_ex has joined #crux-devel18:33
*** Romster has quit IRC20:27
*** Romster has joined #crux-devel20:28
*** jaeger has joined #crux-devel22:25
*** deus_ex has quit IRC23:10
*** deus_ex_ has joined #crux-devel23:11

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!