[00:09] <nbd> florian__: postinst works just like conffiles
[00:12] <CIA-17> florian * r4343 /packages/net/atftp/ (. Config.in Makefile ipkg/): Port atftp to -ng
[00:13] <florian__> ok, that's what I did with my last commit :)
[00:13] <florian__> I really like buildroot-ng !
[00:13] <florian__> really convenient for package maintaining now
[00:23] <florian__> nbd: I have pending patches for nonmips toolchain, ok to commit when I will have cleanup them ?
[00:38] <CIA-17> florian * r4344 /packages/net/gpsd/ (. Config.in Makefile ipkg/): Port gpsd to -ng, closes #659, thanks to russel !
[00:41] <florian__> good night, see you tomorrow
[01:50] <[g2]> there's a problem in trunk (as of a copule hours ago) with gcom dl isn't gcom0.3 but gcom0-3 (or it's named wrong in the sources)
[01:50] <[g2]> am I supposed to submit a bug on that ?
[03:13] <[g2]> Bartman007 that xscale kernel looks hosed I think I get the same issue you get, the decompressor doesn't even output anything
[04:41] <Bartman007> so it says the message to the effect of loading kernel, at offset 0x0001c400 (don't remember the exact message)
[06:42] <florian__> [g2]: please don't fill any bug regarding comgt/gcom
[07:00] <CIA-17> florian * r4345 /packages/net/autossh/ (. Config.in Makefile ipkg/): Port autossh to -ng
[07:16] <CIA-17> florian * r4346 /packages/utils/bc/ (. Config.in Makefile ipkg/): Port bc to -ng
[10:28] <florian__> nbd: ping ?
[10:29] <nbd> pong
[10:30] <florian__> nbd: how do I make menus with buildroot-ng, and the "symbol" that appears in the menu is not a package ?
[10:30] <florian__> nbd: just like bind for example
[10:31] <florian__> I used define Package/bind and MENU:=1 inside
[10:31] <florian__> but I think it will need Build/Package/bind later right ?
[10:34] <florian__> nbd: maybe there is a package I could use as reference ?
[10:34] <nbd> iirc broadcom-wl has something like that
[10:36] <florian__> ok
[10:37] <florian__> by the way, do you approximately know when you will work on brcm63xx ?
[10:37] <nbd> np
[10:37] <nbd> no
[10:38] <florian__> ok, no problem
[10:39] <florian__> I think I will rewrite the map flash since it's not that complicated
[10:39] <florian__> but I started working on weird implementations
[10:48] <florian__> nbd: I am speaking with johannes berg about brcm63xx adsl driver, would you like to join ?
[10:57] <nbd> sure
[14:38] <florian__> nbd: this weekend with a friend of mine we will do some kind of hackaton, if you think it is worth, and you don't have pending patches, we plan on porting ar7 to 2.6 at least do the preliminary task
[14:43] <CIA-17> florian * r4347 /packages/net/bind/ (. Config.in Makefile ipkg/ patches/cross-compile.patch): Port bind to -ng, and fix the cross_compile patch to detect the right OpenSSL includes/libaries and do not check for /dev/random existence while cross compiling
[14:49] <florian__> nbd: ?
[14:51] <nbd> florian__: sounds great. i can give you my old working tree as a starting point
[14:51] <nbd> florian__: iirc a booting 2.6.15 kernel with initramfs and parts of a new ethernet driver
[14:52] <florian__> nbd: yeah that would be great
[14:57] <Hooligan0> polom
[14:58] <florian__> nbd: I think you might want to talk with Hooligan0 :)
[14:58] <nbd> Hooligan0: hi
[14:59] <Hooligan0> how r you florian__ ?
[15:10] <CIA-17> florian * r4348 /packages/net/bind/Makefile: Add missing PKG_BUILDDEP to openssl
[15:31] <CIA-17> florian * r4349 /packages/net/bitchx/ (. Config.in Makefile ipkg/): Port BitchX to -ng, fix compilation by adding correct CFLAGS
[15:49] <nbd> florian__: http://downloads.openwrt.org/people/nbd/tmp/ar7.patch
[15:50] <nbd> florian__: a snapshot of my ar7 2.6 work
[15:50] <florian__> nbd: thanks a lot !
[15:52] <florian__> nbd: does acx111 has vlnyq support ?
[15:53] <nbd> no
[15:53] <florian__> ha, so how could we use the driver ? by writing a layer between acx111 and vlnyq ?
[15:54] <nbd> acx100 already has abstraction code for usb and pci
[15:54] <nbd> so adding vlynq to that should not be too hard
[15:54] <nbd> but first we need ethernet
[15:54] <florian__> ok
[15:54] <florian__> cpmac is already working right ?
[15:54] <nbd> no
[15:54] <florian__> no ?
[15:54] <florian__> even for 2.4 ?
[15:55] <nbd> well, for 2.4 we have the weird ti driver
[15:55] <nbd> but it doesn't work right on 2.6
[15:55] <florian__> ah ok, woah there is a lot of work
[15:55] <nbd> that's why i started the rewrite
[15:58] <florian__> ok, you started coding some things in the ethernet cpmac driver
[15:58] <nbd> yes. it still needs lots of work and it can't receive or send anything yet
[15:58] <florian__> do you think one day, all these patchs could go into the official kernel ? I mean broadcom, rb523 ...
[16:00] <nbd> maybe some of them
[16:01] <florian__> this would save us time by not updating patches to the latest kernel revision, and other developpers could work on them as well
[16:01] <nbd> yeah, definitely
[16:04] <florian__> nbd: you did not answer about the non-mips toolchain ?
[16:09] <nbd> sorry, what did you say about it?
[16:10] <florian__> I have pending patches for the nonmips toolchain (lexra cores mainly)
[16:11] <florian__> I just wanted to know whether I could commit them
[16:11] <nbd> sure
[16:11] <florian__> ok, then I will do it later
[16:21] <CIA-17> florian * r4350 /packages/libs/libiconv/ (. Config.in Makefile ipkg/): Port libiconv to -ng
[16:22] <CIA-17> florian * r4351 /packages/net/bitlbee/ (. Config.in Makefile ipkg/): Port bitlbee to -ng
[16:26] <CIA-17> florian * r4352 /packages/net/bwm/ (. Config.in Makefile ipkg/): Port bwm to -ng
[16:48] <CIA-17> florian * r4353 /packages/net/cbtt/ (. Config.in Makefile ipkg/): Port cbtt to -ng
[18:06] <nbd> florian__: ping
[18:33] <florian__> nbd: pong
[18:33] <nbd> florian__: why is NAT marked as 'M' in the kernel config for au1000 and brcm63xx?
[18:35] <florian__> nbd: for au1000 because it only has one ethernet interface
[18:35] <florian__> for brcm63xx it is a mistake
[18:36] <nbd> i think we should enable it for both
[18:36] <florian__> ok
[18:37] <nbd> but i will also fix the build system issue that causes it to break when nat is compiled as module
[18:37] <florian__> ok
[18:37] <florian__> do you want me to enable this or you do it ?
[18:38] <nbd> you can do it, i'll fix the build system issue in the mean time
[18:38] <florian__> ok
[18:40] <CIA-17> nbd * r4354 /branches/buildroot-ng/openwrt/include/netfilter.mk: don't try to package ipt_SNAT and ipt_DNAT when making kernel module packages
[18:40] <florian__> nbd: is there any other symbol I should put built-in ?
[18:41] <nbd> no idea
[18:41] <florian__> maybe basic ip_tables ?
[18:42] <nbd> definitely
[18:42] <nbd> do a diff against the other platform configs
[18:42] <nbd> you'll see the netfilter differences
[18:43] <florian__> yep
[18:44] <nbd> how much flash do the typical au1000 boxes have?
[18:45] <florian__> 32 Mb
[18:45] <florian__> and 64 Mb ram
[18:46] <nbd> ah, ok
[18:47] <nbd> then we can probably afford putting nat back :)
[18:49] <florian__> in fact I hesitate because it's more an access point than a router
[18:49] <nbd> even for an access point it can be useful
[18:50] <nbd> e.g. when you want to run it as a mesh node
[18:51] <florian__> yes
[18:52] <florian__> I enable it
[19:00] <CIA-17> florian * r4355 /branches/buildroot-ng/openwrt/target/linux/brcm63xx-2.6/config: Update netfilter config to match other platforms
[19:11] <nbd> florian__: does au1000 have pcmcia?
[19:15] <CIA-17> nbd * r4356 /branches/buildroot-ng/openwrt/target/Config.in: au1000 has pci support, but no pcmcia
[19:16] <florian__> nbd: no actually it do not I did a dummy copy
[19:16] <florian__> I am fixing the au1000 kernel config
[19:16] <nbd> it's not in the kernel config
[19:16] <nbd> it was only enabled in target/Config.in
[19:17] <nbd> and because it was not in the kernel config, the pcmcia-cs package broke
[19:17] <florian__> no I know but I just tell you that I fix the kernel config
[19:17] <nbd> ah, ok
[19:17] <nbd> thought it was related
[19:17] <florian__> no, but I should have made myself clear
[19:17] <nbd> no problem
[19:20] <florian__> maybe we should do some more quality tests when -ng will be ready for prod
[19:24] <florian__> should be better now
[19:24] <CIA-17> florian * r4357 /branches/buildroot-ng/openwrt/target/linux/au1000-2.6/config: Update au1000 netfilter config to match other devices config
[21:00] <CIA-17> nbd * r4358 /branches/whiterussian/openwrt/target/linux/linux-2.4/Makefile: increase PKG_RELEASE for the linux kernel
[22:37] <florian__> nbd: I am trying to fix the brcm63xx flash map driver
[22:47] <florian__> nbd: here is my code : http://paste.uni.cc/8857
[22:49] <florian__> and the associated error : http://paste.uni.cc/8858
[23:09] <nbd> florian__: weird offset for the jffs2 partition... 57 eraseblocks?
[23:10] <florian__> well I use redboot partitioning
[23:10] <florian__> I have fixed the driver a bit, here is what I get know :
[23:12] <florian__> http://paste.uni.cc/8859
[23:12] <nbd> sizeof(*parsed_parts) / sizeof(struct mtd_partition)
[23:12] <nbd> you can't do that
[23:12] <florian__> but I used amd_flash, instead of cfi_probe
[23:12] <florian__> nbd: ah oki
[23:12] <nbd> sizeof only works for things where you define the size in the c type
[23:13] <florian__> ah by the way, parsed_parts may not be allocated ?
[23:13] <florian__> I think redboot_parse_partitions does it, hence we pass the address
[23:15] <florian__> is not add_mtd_device required ?
[23:25] <florian__> nbd: in fact the problems comes from the cfi_probe
[23:26] <florian__> I don't know why
[23:26] <florian__> is it possible that the flash chip may not be recognized ?
[23:30] <florian__> nbd: can you tell how I could easily use initramfs ?
[23:30] <nbd> just select it in menuconfig
[23:30] <nbd> i mean in openwrt menuconfig
[23:31] <florian__> ok
[23:31] <florian__> for the flash probing, what do you think can cause the kernel to hand while doing cfi_probe ?
[23:32] <nbd> maybe the flash chip doesn't like being probed
[23:33] <florian__> well, actually there seem to be specific code to initialize the flash, but only in the case we have a CFE bootlodaer
[23:35] <florian__> in the case the flash doesn't like to be probed, how do I detect it ?
[23:37] <nbd> uhm... are you sure that you got the mem window of the flash right?
[23:39] <florian__> well I read it from redboot : FLASH: 0xbe400000 - 0xbec00000, 128 blocks of 0x00010000 bytes each
[23:39] <florian__> and the code has : #define WINDOW_ADDR 0xbe400000 /* Start address of the flash */
[23:39] <florian__> #define REDBOOT_FIS_DIRECTORY /* Address of the RedBoot FIS directory */
[23:39] <florian__> #define WINDOW_SIZE 0x400000
[23:41] <florian__> he, WINDOW_SIZE should be 0x800000 then !
[00:00] --- Tue Aug 1 2006
[00:12] <CIA-17> florian * r4343 /packages/net/atftp/ (. Config.in Makefile ipkg/): Port atftp to -ng
[00:13] <florian__> ok, that's what I did with my last commit :)
[00:13] <florian__> I really like buildroot-ng !
[00:13] <florian__> really convenient for package maintaining now
[00:23] <florian__> nbd: I have pending patches for nonmips toolchain, ok to commit when I will have cleanup them ?
[00:38] <CIA-17> florian * r4344 /packages/net/gpsd/ (. Config.in Makefile ipkg/): Port gpsd to -ng, closes #659, thanks to russel !
[00:41] <florian__> good night, see you tomorrow
[01:50] <[g2]> there's a problem in trunk (as of a copule hours ago) with gcom dl isn't gcom0.3 but gcom0-3 (or it's named wrong in the sources)
[01:50] <[g2]> am I supposed to submit a bug on that ?
[03:13] <[g2]> Bartman007 that xscale kernel looks hosed I think I get the same issue you get, the decompressor doesn't even output anything
[04:41] <Bartman007> so it says the message to the effect of loading kernel, at offset 0x0001c400 (don't remember the exact message)
[06:42] <florian__> [g2]: please don't fill any bug regarding comgt/gcom
[07:00] <CIA-17> florian * r4345 /packages/net/autossh/ (. Config.in Makefile ipkg/): Port autossh to -ng
[07:16] <CIA-17> florian * r4346 /packages/utils/bc/ (. Config.in Makefile ipkg/): Port bc to -ng
[10:28] <florian__> nbd: ping ?
[10:29] <nbd> pong
[10:30] <florian__> nbd: how do I make menus with buildroot-ng, and the "symbol" that appears in the menu is not a package ?
[10:30] <florian__> nbd: just like bind for example
[10:31] <florian__> I used define Package/bind and MENU:=1 inside
[10:31] <florian__> but I think it will need Build/Package/bind later right ?
[10:34] <florian__> nbd: maybe there is a package I could use as reference ?
[10:34] <nbd> iirc broadcom-wl has something like that
[10:36] <florian__> ok
[10:37] <florian__> by the way, do you approximately know when you will work on brcm63xx ?
[10:37] <nbd> np
[10:37] <nbd> no
[10:38] <florian__> ok, no problem
[10:39] <florian__> I think I will rewrite the map flash since it's not that complicated
[10:39] <florian__> but I started working on weird implementations
[10:48] <florian__> nbd: I am speaking with johannes berg about brcm63xx adsl driver, would you like to join ?
[10:57] <nbd> sure
[14:38] <florian__> nbd: this weekend with a friend of mine we will do some kind of hackaton, if you think it is worth, and you don't have pending patches, we plan on porting ar7 to 2.6 at least do the preliminary task
[14:43] <CIA-17> florian * r4347 /packages/net/bind/ (. Config.in Makefile ipkg/ patches/cross-compile.patch): Port bind to -ng, and fix the cross_compile patch to detect the right OpenSSL includes/libaries and do not check for /dev/random existence while cross compiling
[14:49] <florian__> nbd: ?
[14:51] <nbd> florian__: sounds great. i can give you my old working tree as a starting point
[14:51] <nbd> florian__: iirc a booting 2.6.15 kernel with initramfs and parts of a new ethernet driver
[14:52] <florian__> nbd: yeah that would be great
[14:57] <Hooligan0> polom
[14:58] <florian__> nbd: I think you might want to talk with Hooligan0 :)
[14:58] <nbd> Hooligan0: hi
[14:59] <Hooligan0> how r you florian__ ?
[15:10] <CIA-17> florian * r4348 /packages/net/bind/Makefile: Add missing PKG_BUILDDEP to openssl
[15:31] <CIA-17> florian * r4349 /packages/net/bitchx/ (. Config.in Makefile ipkg/): Port BitchX to -ng, fix compilation by adding correct CFLAGS
[15:49] <nbd> florian__: http://downloads.openwrt.org/people/nbd/tmp/ar7.patch
[15:50] <nbd> florian__: a snapshot of my ar7 2.6 work
[15:50] <florian__> nbd: thanks a lot !
[15:52] <florian__> nbd: does acx111 has vlnyq support ?
[15:53] <nbd> no
[15:53] <florian__> ha, so how could we use the driver ? by writing a layer between acx111 and vlnyq ?
[15:54] <nbd> acx100 already has abstraction code for usb and pci
[15:54] <nbd> so adding vlynq to that should not be too hard
[15:54] <nbd> but first we need ethernet
[15:54] <florian__> ok
[15:54] <florian__> cpmac is already working right ?
[15:54] <nbd> no
[15:54] <florian__> no ?
[15:54] <florian__> even for 2.4 ?
[15:55] <nbd> well, for 2.4 we have the weird ti driver
[15:55] <nbd> but it doesn't work right on 2.6
[15:55] <florian__> ah ok, woah there is a lot of work
[15:55] <nbd> that's why i started the rewrite
[15:58] <florian__> ok, you started coding some things in the ethernet cpmac driver
[15:58] <nbd> yes. it still needs lots of work and it can't receive or send anything yet
[15:58] <florian__> do you think one day, all these patchs could go into the official kernel ? I mean broadcom, rb523 ...
[16:00] <nbd> maybe some of them
[16:01] <florian__> this would save us time by not updating patches to the latest kernel revision, and other developpers could work on them as well
[16:01] <nbd> yeah, definitely
[16:04] <florian__> nbd: you did not answer about the non-mips toolchain ?
[16:09] <nbd> sorry, what did you say about it?
[16:10] <florian__> I have pending patches for the nonmips toolchain (lexra cores mainly)
[16:11] <florian__> I just wanted to know whether I could commit them
[16:11] <nbd> sure
[16:11] <florian__> ok, then I will do it later
[16:21] <CIA-17> florian * r4350 /packages/libs/libiconv/ (. Config.in Makefile ipkg/): Port libiconv to -ng
[16:22] <CIA-17> florian * r4351 /packages/net/bitlbee/ (. Config.in Makefile ipkg/): Port bitlbee to -ng
[16:26] <CIA-17> florian * r4352 /packages/net/bwm/ (. Config.in Makefile ipkg/): Port bwm to -ng
[16:48] <CIA-17> florian * r4353 /packages/net/cbtt/ (. Config.in Makefile ipkg/): Port cbtt to -ng
[18:06] <nbd> florian__: ping
[18:33] <florian__> nbd: pong
[18:33] <nbd> florian__: why is NAT marked as 'M' in the kernel config for au1000 and brcm63xx?
[18:35] <florian__> nbd: for au1000 because it only has one ethernet interface
[18:35] <florian__> for brcm63xx it is a mistake
[18:36] <nbd> i think we should enable it for both
[18:36] <florian__> ok
[18:37] <nbd> but i will also fix the build system issue that causes it to break when nat is compiled as module
[18:37] <florian__> ok
[18:37] <florian__> do you want me to enable this or you do it ?
[18:38] <nbd> you can do it, i'll fix the build system issue in the mean time
[18:38] <florian__> ok
[18:40] <CIA-17> nbd * r4354 /branches/buildroot-ng/openwrt/include/netfilter.mk: don't try to package ipt_SNAT and ipt_DNAT when making kernel module packages
[18:40] <florian__> nbd: is there any other symbol I should put built-in ?
[18:41] <nbd> no idea
[18:41] <florian__> maybe basic ip_tables ?
[18:42] <nbd> definitely
[18:42] <nbd> do a diff against the other platform configs
[18:42] <nbd> you'll see the netfilter differences
[18:43] <florian__> yep
[18:44] <nbd> how much flash do the typical au1000 boxes have?
[18:45] <florian__> 32 Mb
[18:45] <florian__> and 64 Mb ram
[18:46] <nbd> ah, ok
[18:47] <nbd> then we can probably afford putting nat back :)
[18:49] <florian__> in fact I hesitate because it's more an access point than a router
[18:49] <nbd> even for an access point it can be useful
[18:50] <nbd> e.g. when you want to run it as a mesh node
[18:51] <florian__> yes
[18:52] <florian__> I enable it
[19:00] <CIA-17> florian * r4355 /branches/buildroot-ng/openwrt/target/linux/brcm63xx-2.6/config: Update netfilter config to match other platforms
[19:11] <nbd> florian__: does au1000 have pcmcia?
[19:15] <CIA-17> nbd * r4356 /branches/buildroot-ng/openwrt/target/Config.in: au1000 has pci support, but no pcmcia
[19:16] <florian__> nbd: no actually it do not I did a dummy copy
[19:16] <florian__> I am fixing the au1000 kernel config
[19:16] <nbd> it's not in the kernel config
[19:16] <nbd> it was only enabled in target/Config.in
[19:17] <nbd> and because it was not in the kernel config, the pcmcia-cs package broke
[19:17] <florian__> no I know but I just tell you that I fix the kernel config
[19:17] <nbd> ah, ok
[19:17] <nbd> thought it was related
[19:17] <florian__> no, but I should have made myself clear
[19:17] <nbd> no problem
[19:20] <florian__> maybe we should do some more quality tests when -ng will be ready for prod
[19:24] <florian__> should be better now
[19:24] <CIA-17> florian * r4357 /branches/buildroot-ng/openwrt/target/linux/au1000-2.6/config: Update au1000 netfilter config to match other devices config
[21:00] <CIA-17> nbd * r4358 /branches/whiterussian/openwrt/target/linux/linux-2.4/Makefile: increase PKG_RELEASE for the linux kernel
[22:37] <florian__> nbd: I am trying to fix the brcm63xx flash map driver
[22:47] <florian__> nbd: here is my code : http://paste.uni.cc/8857
[22:49] <florian__> and the associated error : http://paste.uni.cc/8858
[23:09] <nbd> florian__: weird offset for the jffs2 partition... 57 eraseblocks?
[23:10] <florian__> well I use redboot partitioning
[23:10] <florian__> I have fixed the driver a bit, here is what I get know :
[23:12] <florian__> http://paste.uni.cc/8859
[23:12] <nbd> sizeof(*parsed_parts) / sizeof(struct mtd_partition)
[23:12] <nbd> you can't do that
[23:12] <florian__> but I used amd_flash, instead of cfi_probe
[23:12] <florian__> nbd: ah oki
[23:12] <nbd> sizeof only works for things where you define the size in the c type
[23:13] <florian__> ah by the way, parsed_parts may not be allocated ?
[23:13] <florian__> I think redboot_parse_partitions does it, hence we pass the address
[23:15] <florian__> is not add_mtd_device required ?
[23:25] <florian__> nbd: in fact the problems comes from the cfi_probe
[23:26] <florian__> I don't know why
[23:26] <florian__> is it possible that the flash chip may not be recognized ?
[23:30] <florian__> nbd: can you tell how I could easily use initramfs ?
[23:30] <nbd> just select it in menuconfig
[23:30] <nbd> i mean in openwrt menuconfig
[23:31] <florian__> ok
[23:31] <florian__> for the flash probing, what do you think can cause the kernel to hand while doing cfi_probe ?
[23:32] <nbd> maybe the flash chip doesn't like being probed
[23:33] <florian__> well, actually there seem to be specific code to initialize the flash, but only in the case we have a CFE bootlodaer
[23:35] <florian__> in the case the flash doesn't like to be probed, how do I detect it ?
[23:37] <nbd> uhm... are you sure that you got the mem window of the flash right?
[23:39] <florian__> well I read it from redboot : FLASH: 0xbe400000 - 0xbec00000, 128 blocks of 0x00010000 bytes each
[23:39] <florian__> and the code has : #define WINDOW_ADDR 0xbe400000 /* Start address of the flash */
[23:39] <florian__> #define REDBOOT_FIS_DIRECTORY /* Address of the RedBoot FIS directory */
[23:39] <florian__> #define WINDOW_SIZE 0x400000
[23:41] <florian__> he, WINDOW_SIZE should be 0x800000 then !
[00:00] --- Tue Aug 1 2006