[00:30] <CIA-4> nbd * r4044 /branches/buildroot-ng/openwrt/target/linux/generic-2.4/patches/227-jffs2_eofdetect.patch: backport jffs2 eofdetect patch to linux 2.4
[01:05] <jr-> kaloz. ok got openwrt shell on that billion device with buildroot-ng 4041. with initrd for now. i need to backup current flash before i wipe it and play more. have to wait until network interfaces are supported. now time to sleep.
[01:27] <CIA-4> mbm * r4045 /branches/buildroot-ng/openwrt/package/iptables/Makefile: avoid includes that aren't required for current target (make download)
[01:29] <nbd> does make not care about the files being included more than once?
[01:29] <[mbm]> doesn't seem to
[01:30] <nbd> ok
[01:31] Action: [mbm] is still thinking the whole thing through .. it's really hard to think about what make would do with our template system
[01:31] <[mbm]> I *think* that the makefile is still correct
[01:32] <nbd> i also think it's correct
[01:32] <nbd> well, if it works
[01:32] <[mbm]> seems to
[01:34] <nbd> hmm... no
[01:34] <nbd> doesn't work for me
[01:34] <[mbm]> ?
[01:34] <nbd> it doesn't get the module names
[01:43] <nbd> i have a better idea on how to fix this
[01:44] <nbd> just do this:
[01:44] <nbd> -include $(LINUX_DIR)/.config
[01:44] <[mbm]> yeah, I was coming to the same conclusion
[01:45] <CIA-4> nbd * r4046 /branches/buildroot-ng/openwrt/package/iptables/Makefile: revert last change and add a different fix
[01:47] <[mbm]> for the verbosity stuff it looks like hooking $(MAKE) is going to be a major pain
[01:47] <nbd> btw. broadcom has this optimize_lib.sh script that strips out all unnecessary functions from libraries by looking at what the binaries on the filesystem use
[01:47] <nbd> i think we should use it for micro builds without ipkg support
[01:47] <[mbm]> what if we just had echo "foo" >&3 statements
[01:48] <[mbm]> and then at the top level we'd do an optional &>/dev/nul
[01:48] <nbd> that's what i meant in the first place
[01:48] <nbd> i thought that was what you were trying to do
[01:48] <[mbm]> :)
[01:48] <[mbm]> well, I wanted to hook it at the make and be able to set the depth
[01:49] <[mbm]> but that's quite impractical with the way make handles some targets
[01:49] <nbd> we should use the $(call make) thing explicitly on calls that descend into a different directory
[01:50] <[mbm]> that $(call make) thing was creating a real mess
[01:50] <nbd> why?
[02:05] <nbd> oh, rstrip.sh has a bug
[02:05] <nbd> it won't strip binaries that have been stripped with the regular strip
[02:05] <nbd> and we have a few of those binaries :)
[02:07] <CIA-4> nbd * r4047 /branches/buildroot-ng/openwrt/scripts/rstrip.sh: sstrip binaries that have been stripped by the regular strip before
[02:27] <[mbm]> doh.. why didn't I think of this before
[02:28] <[mbm]> %:
[02:28] <[mbm]> echo "$@" > &3
[02:39] <[mbm]> heh, yep, works
[02:45] <nbd> btw. why does $(call make) create a mess?
[02:45] <[mbm]> well, hooking via $(call make) creates a mess
[02:46] <nbd> how?
[02:46] <[mbm]> not all targets were called implicitly
[02:46] <[mbm]> so the above %: trick is much simpler and elegant for displaying targets
[02:47] <nbd> maybe we can combine both
[02:47] <nbd> depending on the tracing level
[02:47] <[mbm]> hmm maybe
[02:47] <nbd> i only wanted to use it if we move from, say package/ to package/<foobar>
[02:47] <nbd> not for individual make targets that don't leave the current directory
[02:47] <[mbm]> right now I'm playing with that % in rules.mk
[02:48] <[mbm]> which works really nicely
[02:48] <nbd> if we combine both, we have really fine control over how much logging output we want
[02:48] <[mbm]> @echo " MAKE -C $${PWD##$$TOPDIR/} $@" >&3
[02:49] <nbd> yeah, i think putting both together will work
[02:49] <nbd> but let's finish yours first
[02:49] <[mbm]> Collecting package info... MAKE -C package/alsa all MAKE -C package/base-files all MAKE -C package/bridge all MAKE -C package/broadcom-wl all
[02:50] <[mbm]> hmm irssi put them all on one line
[02:50] <nbd> maybe we should leave them out for the dump
[02:50] <nbd> or at least for lower levels
[02:53] <[mbm]> http://pastebin.ca/68743
[02:53] <[mbm]> looks nice enough
[02:53] <[mbm]> only thing I haven't figured out is why some targets show up multiple times
[02:53] <[mbm]> like target/utils prepare
[02:55] <nbd> yeah
[02:56] <nbd> looks nice
[02:59] <CIA-4> nbd * r4048 /branches/buildroot-ng/openwrt/include/image.mk: add 0xdeadc0de to the jffs2 filesystem images
[03:00] <CIA-4> nbd * r4049 /branches/buildroot-ng/openwrt/include/package.mk: remove stale reference to package/rules.mk - not that useful in the dependency handling anyway
[03:16] <[mbm]> ah, right ..
[03:16] <[mbm]> the % only works when the traditional target isn't defined
[03:16] <[mbm]> in which case it's working more often than I had expected
[03:18] <nbd> tried %:: ?
[03:28] <[mbm]> hmm no. didn't think of that
[03:28] <[mbm]> "%: FORCE" is interesting
[03:36] <[mbm]> the %:: looks better
[03:37] <nbd> it still won't catch targets that define commands
[03:38] <nbd> but maybe it doesn't need to
[03:38] <nbd> but it will only display them *after* they're already processed
[04:16] <nbd> [mbm]: i've found an easy way to trace explicit makefile calls:
[04:16] <nbd> $(MAKECMDGOALS): echo-$(MAKECMDGOALS)
[04:16] <nbd> echo-$(MAKECMDGOALS):
[04:16] <nbd> @echo $(MAKECMDGOALS)
[08:39] <srcquench> ho
[08:39] <[mbm]> .
[08:42] <florian__> .
[15:18] <florian__> [mbm], nbd, Kaloz could you suggest a name for the openwrt presentation I have talked about ?
[15:20] <florian__> I was thinking about "OpenWrt : state of art and future" but it's definitively bad
[15:26] <florian__> .. ?
[15:31] <thepeople> florian: "OpenWrt: The future state of functionality"
[15:32] <db90h> "OpenWrt: The base of the future" ;p
[15:33] <florian__> lol
[15:33] <florian__> I mean a real title ;)
[15:33] <db90h> "OpenWrt: May the force be with you!"
[15:34] <florian__> "OpenWrt : Luke, I am your father"
[15:34] <db90h> lol
[15:34] <florian__> "OpenWrt at the beach"
[15:34] <florian__> "OpenWrt at the mountain"
[15:34] <florian__> "Teaching with OpenWrt"
[15:34] <crazy_imp> florian__: OpenWrt, fresh air for (wlan) routers
[15:35] <thepeople> "OpenWrt: The future of wireless state of the art technologies"
[15:35] <florian__> not that bad
[15:35] <db90h> "OpenWrt: Soon to be ClosedWrt, the commercial production"
[15:36] <florian__> lol
[15:36] <thepeople> :-o
[15:36] <db90h> these 'future' and 'state of the art' slogans are overused by all the marketeers
[15:36] <db90h> tho for a reason, they are catchy
[15:38] <thepeople> "OpenWrt: Commercial ripoffs coming soon"
[15:39] <thepeople> db90h: I guess people like hearing about the future
[15:39] <thepeople> even if it isn't
[15:47] <florian__> nbd: any idea for the presentation title ?
[15:47] <nbd> no, sorry
[15:51] <florian__> ah such a pity
[15:52] <florian__> crazy_imp's title sound good, not too commercial ;)
[15:53] <crazy_imp> :)
[15:53] <srcquench> florian__: "Openwrt: What the dudes do all day long"
[15:53] <florian__> too general
[15:54] <florian__> it's not supposed to become a troll ;)
[15:55] <thepeople> "OpenWrt: Why wait? Compile yourself!"
[15:55] <florian__> lol
[19:27] <[mbm]> hmm nobody's looked at rules.mk in years, have they?
[19:27] <andersee> of course not
[19:28] Action: [mbm] sees lines that date back to the original uclibc versions
[19:28] <[mbm]> and have absolutely no effect on openwrt
[19:28] <andersee> probably the little that remains that was written by me
[20:37] <Kaloz> florian__: "OpenWrt - You will be assimilated."
[20:59] <jr-> kaloz. buildroot-ng ramdisk version is handy for testing, but there's small problem. it doesn't create /dev/console and /dev/ttyS0 thus causing kernel panic on boot. adding those manually makes it work. should I file ticket? i was just wondering if it's better not to file ng tickets since it's constantly changing and not expected to work at all.
[20:59] <Kaloz> devfs should create those files
[20:59] <jr-> after boot it of course mounts devfs and other devices work that way, but prior devfs mountint it needs at least console, not sure about ttyS0
[21:05] <Kaloz> jr-: udev needs theose files you are talking about
[21:05] <Kaloz> jr-: udev creates everything itself
[21:06] <tziOm> hey guys
[21:06] <tziOm> any rtorrent available for rc5?
[21:06] <h3sp4wn> Look at macsats site
[21:06] <jr-> kaloz. well anyway i get unable to open initial console unless i add /dev/console and /dev/ttyS0 to cpio ramdisk image. after adding them it works fine.
[21:08] <jr-> kaloz. when i look into /dev in openwrt shell in same device it contains a lot of other devs too that come from udev (that's what i meant with devfs)
[21:09] <Kaloz> err, okay, so
[21:09] <Kaloz> udev needs those files, devfs doesn't need them
[21:14] <jr-> hmm. /etc/mtab says it's devfs
[21:17] <[mbm]> the ramdisk stuff should have a /init that mounts devfs and exec's busybox's init
[21:19] <jr-> mbm. yes that's what it does afaik. at least /init says so. however, since devfs isn't yet mounted and there's no /dev/console kernel doesn't execute /init at all and panics. adding /dev/console fixes that and devfs gets mounted.
[21:20] <[mbm]> the kernel gives you a warning but that's it
[21:22] <jr-> hmm. is it lack of /dev/ttyS0 then? (console=ttyS0,115200 parameter passed to kernel)
[21:22] <[mbm]> you do not need anything in /dev :P
[21:24] <jr-> ok. i'll try again without because i added them after i got stuck with panic. perhaps it was something else wrong with by compile that got fixed on recompile when i added those console and ttyS0
[21:41] <jr-> mbm, kaloz. you were right. it works without files under /dev now. apparently something went wrong on first compile. sorry for wasting your time :)
[22:02] <[mbm]> :)
[00:00] --- Fri Jun 23 2006
[01:05] <jr-> kaloz. ok got openwrt shell on that billion device with buildroot-ng 4041. with initrd for now. i need to backup current flash before i wipe it and play more. have to wait until network interfaces are supported. now time to sleep.
[01:27] <CIA-4> mbm * r4045 /branches/buildroot-ng/openwrt/package/iptables/Makefile: avoid includes that aren't required for current target (make download)
[01:29] <nbd> does make not care about the files being included more than once?
[01:29] <[mbm]> doesn't seem to
[01:30] <nbd> ok
[01:31] Action: [mbm] is still thinking the whole thing through .. it's really hard to think about what make would do with our template system
[01:31] <[mbm]> I *think* that the makefile is still correct
[01:32] <nbd> i also think it's correct
[01:32] <nbd> well, if it works
[01:32] <[mbm]> seems to
[01:34] <nbd> hmm... no
[01:34] <nbd> doesn't work for me
[01:34] <[mbm]> ?
[01:34] <nbd> it doesn't get the module names
[01:43] <nbd> i have a better idea on how to fix this
[01:44] <nbd> just do this:
[01:44] <nbd> -include $(LINUX_DIR)/.config
[01:44] <[mbm]> yeah, I was coming to the same conclusion
[01:45] <CIA-4> nbd * r4046 /branches/buildroot-ng/openwrt/package/iptables/Makefile: revert last change and add a different fix
[01:47] <[mbm]> for the verbosity stuff it looks like hooking $(MAKE) is going to be a major pain
[01:47] <nbd> btw. broadcom has this optimize_lib.sh script that strips out all unnecessary functions from libraries by looking at what the binaries on the filesystem use
[01:47] <nbd> i think we should use it for micro builds without ipkg support
[01:47] <[mbm]> what if we just had echo "foo" >&3 statements
[01:48] <[mbm]> and then at the top level we'd do an optional &>/dev/nul
[01:48] <nbd> that's what i meant in the first place
[01:48] <nbd> i thought that was what you were trying to do
[01:48] <[mbm]> :)
[01:48] <[mbm]> well, I wanted to hook it at the make and be able to set the depth
[01:49] <[mbm]> but that's quite impractical with the way make handles some targets
[01:49] <nbd> we should use the $(call make) thing explicitly on calls that descend into a different directory
[01:50] <[mbm]> that $(call make) thing was creating a real mess
[01:50] <nbd> why?
[02:05] <nbd> oh, rstrip.sh has a bug
[02:05] <nbd> it won't strip binaries that have been stripped with the regular strip
[02:05] <nbd> and we have a few of those binaries :)
[02:07] <CIA-4> nbd * r4047 /branches/buildroot-ng/openwrt/scripts/rstrip.sh: sstrip binaries that have been stripped by the regular strip before
[02:27] <[mbm]> doh.. why didn't I think of this before
[02:28] <[mbm]> %:
[02:28] <[mbm]> echo "$@" > &3
[02:39] <[mbm]> heh, yep, works
[02:45] <nbd> btw. why does $(call make) create a mess?
[02:45] <[mbm]> well, hooking via $(call make) creates a mess
[02:46] <nbd> how?
[02:46] <[mbm]> not all targets were called implicitly
[02:46] <[mbm]> so the above %: trick is much simpler and elegant for displaying targets
[02:47] <nbd> maybe we can combine both
[02:47] <nbd> depending on the tracing level
[02:47] <[mbm]> hmm maybe
[02:47] <nbd> i only wanted to use it if we move from, say package/ to package/<foobar>
[02:47] <nbd> not for individual make targets that don't leave the current directory
[02:47] <[mbm]> right now I'm playing with that % in rules.mk
[02:48] <[mbm]> which works really nicely
[02:48] <nbd> if we combine both, we have really fine control over how much logging output we want
[02:48] <[mbm]> @echo " MAKE -C $${PWD##$$TOPDIR/} $@" >&3
[02:49] <nbd> yeah, i think putting both together will work
[02:49] <nbd> but let's finish yours first
[02:49] <[mbm]> Collecting package info... MAKE -C package/alsa all MAKE -C package/base-files all MAKE -C package/bridge all MAKE -C package/broadcom-wl all
[02:50] <[mbm]> hmm irssi put them all on one line
[02:50] <nbd> maybe we should leave them out for the dump
[02:50] <nbd> or at least for lower levels
[02:53] <[mbm]> http://pastebin.ca/68743
[02:53] <[mbm]> looks nice enough
[02:53] <[mbm]> only thing I haven't figured out is why some targets show up multiple times
[02:53] <[mbm]> like target/utils prepare
[02:55] <nbd> yeah
[02:56] <nbd> looks nice
[02:59] <CIA-4> nbd * r4048 /branches/buildroot-ng/openwrt/include/image.mk: add 0xdeadc0de to the jffs2 filesystem images
[03:00] <CIA-4> nbd * r4049 /branches/buildroot-ng/openwrt/include/package.mk: remove stale reference to package/rules.mk - not that useful in the dependency handling anyway
[03:16] <[mbm]> ah, right ..
[03:16] <[mbm]> the % only works when the traditional target isn't defined
[03:16] <[mbm]> in which case it's working more often than I had expected
[03:18] <nbd> tried %:: ?
[03:28] <[mbm]> hmm no. didn't think of that
[03:28] <[mbm]> "%: FORCE" is interesting
[03:36] <[mbm]> the %:: looks better
[03:37] <nbd> it still won't catch targets that define commands
[03:38] <nbd> but maybe it doesn't need to
[03:38] <nbd> but it will only display them *after* they're already processed
[04:16] <nbd> [mbm]: i've found an easy way to trace explicit makefile calls:
[04:16] <nbd> $(MAKECMDGOALS): echo-$(MAKECMDGOALS)
[04:16] <nbd> echo-$(MAKECMDGOALS):
[04:16] <nbd> @echo $(MAKECMDGOALS)
[08:39] <srcquench> ho
[08:39] <[mbm]> .
[08:42] <florian__> .
[15:18] <florian__> [mbm], nbd, Kaloz could you suggest a name for the openwrt presentation I have talked about ?
[15:20] <florian__> I was thinking about "OpenWrt : state of art and future" but it's definitively bad
[15:26] <florian__> .. ?
[15:31] <thepeople> florian: "OpenWrt: The future state of functionality"
[15:32] <db90h> "OpenWrt: The base of the future" ;p
[15:33] <florian__> lol
[15:33] <florian__> I mean a real title ;)
[15:33] <db90h> "OpenWrt: May the force be with you!"
[15:34] <florian__> "OpenWrt : Luke, I am your father"
[15:34] <db90h> lol
[15:34] <florian__> "OpenWrt at the beach"
[15:34] <florian__> "OpenWrt at the mountain"
[15:34] <florian__> "Teaching with OpenWrt"
[15:34] <crazy_imp> florian__: OpenWrt, fresh air for (wlan) routers
[15:35] <thepeople> "OpenWrt: The future of wireless state of the art technologies"
[15:35] <florian__> not that bad
[15:35] <db90h> "OpenWrt: Soon to be ClosedWrt, the commercial production"
[15:36] <florian__> lol
[15:36] <thepeople> :-o
[15:36] <db90h> these 'future' and 'state of the art' slogans are overused by all the marketeers
[15:36] <db90h> tho for a reason, they are catchy
[15:38] <thepeople> "OpenWrt: Commercial ripoffs coming soon"
[15:39] <thepeople> db90h: I guess people like hearing about the future
[15:39] <thepeople> even if it isn't
[15:47] <florian__> nbd: any idea for the presentation title ?
[15:47] <nbd> no, sorry
[15:51] <florian__> ah such a pity
[15:52] <florian__> crazy_imp's title sound good, not too commercial ;)
[15:53] <crazy_imp> :)
[15:53] <srcquench> florian__: "Openwrt: What the dudes do all day long"
[15:53] <florian__> too general
[15:54] <florian__> it's not supposed to become a troll ;)
[15:55] <thepeople> "OpenWrt: Why wait? Compile yourself!"
[15:55] <florian__> lol
[19:27] <[mbm]> hmm nobody's looked at rules.mk in years, have they?
[19:27] <andersee> of course not
[19:28] Action: [mbm] sees lines that date back to the original uclibc versions
[19:28] <[mbm]> and have absolutely no effect on openwrt
[19:28] <andersee> probably the little that remains that was written by me
[20:37] <Kaloz> florian__: "OpenWrt - You will be assimilated."
[20:59] <jr-> kaloz. buildroot-ng ramdisk version is handy for testing, but there's small problem. it doesn't create /dev/console and /dev/ttyS0 thus causing kernel panic on boot. adding those manually makes it work. should I file ticket? i was just wondering if it's better not to file ng tickets since it's constantly changing and not expected to work at all.
[20:59] <Kaloz> devfs should create those files
[20:59] <jr-> after boot it of course mounts devfs and other devices work that way, but prior devfs mountint it needs at least console, not sure about ttyS0
[21:05] <Kaloz> jr-: udev needs theose files you are talking about
[21:05] <Kaloz> jr-: udev creates everything itself
[21:06] <tziOm> hey guys
[21:06] <tziOm> any rtorrent available for rc5?
[21:06] <h3sp4wn> Look at macsats site
[21:06] <jr-> kaloz. well anyway i get unable to open initial console unless i add /dev/console and /dev/ttyS0 to cpio ramdisk image. after adding them it works fine.
[21:08] <jr-> kaloz. when i look into /dev in openwrt shell in same device it contains a lot of other devs too that come from udev (that's what i meant with devfs)
[21:09] <Kaloz> err, okay, so
[21:09] <Kaloz> udev needs those files, devfs doesn't need them
[21:14] <jr-> hmm. /etc/mtab says it's devfs
[21:17] <[mbm]> the ramdisk stuff should have a /init that mounts devfs and exec's busybox's init
[21:19] <jr-> mbm. yes that's what it does afaik. at least /init says so. however, since devfs isn't yet mounted and there's no /dev/console kernel doesn't execute /init at all and panics. adding /dev/console fixes that and devfs gets mounted.
[21:20] <[mbm]> the kernel gives you a warning but that's it
[21:22] <jr-> hmm. is it lack of /dev/ttyS0 then? (console=ttyS0,115200 parameter passed to kernel)
[21:22] <[mbm]> you do not need anything in /dev :P
[21:24] <jr-> ok. i'll try again without because i added them after i got stuck with panic. perhaps it was something else wrong with by compile that got fixed on recompile when i added those console and ttyS0
[21:41] <jr-> mbm, kaloz. you were right. it works without files under /dev now. apparently something went wrong on first compile. sorry for wasting your time :)
[22:02] <[mbm]> :)
[00:00] --- Fri Jun 23 2006