[01:40] <CIA-4> nbd * r4028 /branches/buildroot-ng/openwrt/package/ (Makefile rules.mk): add package/download target
[01:49] <CIA-4> nbd * r4029 /branches/buildroot-ng/openwrt/toolchain/ (Makefile sed/Makefile sed/sedcheck.sh utils/Makefile): add toolchain/download target
[01:56] <CIA-4> nbd * r4030 /branches/buildroot-ng/openwrt/target/ (8 files in 6 dirs): add target/download
[01:57] <CIA-4> nbd * r4031 /branches/buildroot-ng/openwrt/Makefile: add global make download target
[04:16] <CIA-4> nbd * r4032 / (96 files in 77 dirs): build system cleanup. move shared include files into $(TOPDIR)/include, move lzma, mkfs.* into toolchain/
[04:32] <CIA-4> nbd * r4033 / (65 files in 63 dirs): add INCLUDE_DIR variable for $(TOPDIR)/include
[04:45] <CIA-4> nbd * r4034 /branches/buildroot-ng/openwrt/scripts/gen_deps.pl: add sdk option for gen_deps.pl
[05:01] <CIA-4> nbd * r4035 /branches/buildroot-ng/openwrt/scripts/gen_deps.pl: don't generate invalid dependency lines in sdk mode
[05:13] <CIA-4> nbd * r4036 /branches/buildroot-ng/openwrt/target/linux/image/ (8 files in 8 dirs): include $(TOPDIR)/rules.mk in the image makefiles
[05:17] <CIA-4> nbd * r4037 /branches/buildroot-ng/openwrt/ (8 files in 6 dirs): fix and enable the sdk
[08:19] <CIA-4> nbd * r4038 /branches/buildroot-ng/openwrt/ (57 files in 48 dirs): massive cleanup of toolchain/
[08:25] <[mbm]> hmm.. that is a rather large change
[08:25] <nbd> yeah
[08:25] <nbd> i converted the bitrot to something cleaner, more buildroot-ng style :)
[08:32] <nbd> [mbm]: btw. about the verbosity control in the build system: couldn't we just use an unused file descriptor for passing the build system trace messages, like echo ... >&3 and then redirecting it to stderr from the main make process?
[08:33] <nbd> or is that something that the shell can't do?
[08:37] <[mbm]> don't think the shell will let you do anything other than 0-2
[08:38] <nbd> hmm... this works:
[08:38] <nbd> ( (echo foo >&3 ) 2>/dev/null >/dev/null ) 3>&1
[08:39] <[mbm]> interesting
[08:39] <nbd> tried it with zsh and bash
[08:39] <nbd> if we could use that, it would make the whole tracing stuff a lot easier
[08:39] <[mbm]> yeah
[08:41] <[mbm]> I'd still like to replace "$MAKE foo" with something more like "$(call make,foo)" just to give us more control
[08:41] <nbd> yeah
[08:42] <[mbm]> and I can't think of any value for $MAKE that would allow that
[08:42] <[mbm]> the cloest would be to define $MAKE as external script
[08:43] <nbd> the thing is: we don't have to change $MAKE for that
[08:43] <nbd> we just patch the places for the buildsystem-internal calls
[08:43] <nbd> and do the rest in include/package.mk
[08:43] <[mbm]> yeah, was thinking about that
[08:45] <nbd> i think when we got this thing implemented and when i've fixed the image builder, we can move trunk to someplace else and replace it with buildroot-ng
[08:45] <nbd> and then slowly start porting all the packages over
[08:45] <[mbm]> yeah I was thinking of just overwriting trunk with buildroot-ng
[08:46] <nbd> we could move it to tags/kamikaze-old
[08:46] <nbd> easier than overwriting
[08:46] <[mbm]> ok
[08:46] Action: [mbm] hasn't played with svn much
[08:46] <nbd> with most of the stuff i made sure that i used svn cp, so that history in most places is preserved
[10:01] <florian__> [mbm], nbd do you well know mips assembly ?
[10:01] <florian__> I am trying to compile a brcm63xx kernel
[10:02] <florian__> and get some errors I am not able to solve
[10:02] <florian__> AS arch/mips/kernel/scall32-o32.o
[10:02] <florian__> arch/mips/kernel/scall32-o32.S: Assembler messages:
[10:02] <florian__> arch/mips/kernel/scall32-o32.S:654: Error: Macro with this name was already defined
[10:02] <florian__> here is the ASM code :
[10:04] <florian__> damn pastebin is lagging
[10:04] <nbd> use a different one
[10:04] <nbd> pastebin.com is slow
[10:05] <florian__> http://pastebin.com/722842
[10:07] <nbd> did you modify that file?
[10:07] <florian__> let me check if is by the patch
[10:08] <florian__> yes it is modified
[10:09] <nbd> got a diff?
[10:09] <florian__> yep
[10:10] <florian__> http://pastebin.ca/68237
[10:10] <florian__> looks like a linux-mips diff
[10:11] <florian__> not broadcom
[10:12] <florian__> and it's not causing the error message
[10:18] <florian__> I will see how it has changed through differente kernel versions
[10:23] <florian__> I should upgrade the 2.6.8.1 patches to 2.6.17
[10:47] <florian__> I really love broadcom people commenting their code "this driver is simple because it is simple"
[11:27] <florian__> I found this in broadcom patches : - __EXPORT_SYMBOL(sym, "_gpl")+ __EXPORT_SYMBOL(sym, "")
[11:27] <florian__> do they have the right to do this ?
[11:32] <common> use phpfi.com
[11:45] <Kaloz> florian__: told you this 2 months ago at least :P
[11:45] <Kaloz> florian__: and no, they have no right to do this
[11:46] <florian__> Kaloz: told me what ?
[11:46] <Kaloz> that symbol change
[11:46] <florian__> ah right
[11:47] <florian__> do their module work if we don't patch the module param as they wish ?
[11:47] <Kaloz> i doubt
[11:48] <florian__> ok
[11:49] <florian__> I found a cleaner brcm63xx source where there was less calls to their external tools
[12:30] <common> Kaloz do you plan adding xscale to the buildroot-ng enviroment?
[12:37] <Kaloz> sure
[12:40] <common> good
[12:49] <wbx> is it possible to get commit-messages via e-mail?
[12:56] <common> svn hooks
[12:57] <wbx> common: i mean messages to a mailaccount, when a openwrt developer commit sth.
[12:57] <wbx> common: i know how to configure this on my own servers.
[12:59] <common> it would be possible, but i doubt somebody is willing to do you this favour
[13:08] <florian__> no troll
[13:20] <alxhh> a commit-list would be nice
[13:22] <florian__> right
[13:26] <dbcch> aren't commit messages broadcast to #openwrt-devel?
[13:26] <florian__> usually they are when CIA-4 is up and running
[13:26] <dbcch> i guess you coudl use those as the basis for an emailer
[13:27] <dbcch> i.e. mirc script
[13:30] <florian__> mirc hu hu
[13:31] <dbcch> ;p
[13:31] <dbcch> I mean, x-chat ;p
[14:06] <common> its pretty easy to use svn hooks server side to mail them to a list
[14:07] <nbd> yeah
[14:14] <db90h> like u said, as long as someone is around to do him that favor ;p
[14:20] <alxhh> db90h: stop trolling, wbx is a friend, nor an enemy
[14:20] Action: nbd found that a feed reader is much more convenient for keeping track of svn changes
[14:21] <alxhh> and, separating the projekts by childish behavior is dumb, stupid, contraproduktive
[14:22] <alxhh> viewing freewrt as a place, where things can be tried in a different way to see different approaches and how they work is the way
[14:23] <alxhh> s/nor/not/
[14:24] <db90h> oh, that's what florian meant by 'no troll'.. my bad
[14:25] <db90h> i wondered what the hell he was talking about, i didn't understand it was a sensitive issue
[14:25] <alxhh> and, a list for commits help those, who are not online 24/7
[14:25] <Kaloz> alxhh: well, you know there is a reason why i love our language
[14:26] <alxhh> Kaloz: which language?
[14:26] <Kaloz> hungarian
[14:26] <db90h> Klingon ;p
[14:26] <Kaloz> alxhh: while english has only "friend" really, hungarian has "ismeros" and "barat"
[14:27] <db90h> is one like a good acquantaince or something?
[14:27] <Kaloz> db90h: something like that, but it doesn't mean you have to like the other one, just you know him
[14:28] <Kaloz> db90h: while the second one is about a close link
[14:29] <db90h> ah, ic
[14:30] <Kaloz> so we really distinguish the two
[15:02] <CIA-4> nbd * r4039 /branches/buildroot-ng/openwrt/Config.in: reorder sdk menuconfig entry, so that it does not force the toolchain selection into the main menu
[15:07] <Kaloz> hmz
[15:09] <nbd> hmz?
[15:11] <Kaloz> didn't think you will be that fast, i was adding back the xscale stuff to brng
[15:11] <Kaloz> :P
[15:11] <nbd> :)
[15:11] <nbd> btw. do you like the toolchain/ rewrite?
[15:13] <Kaloz> yep
[16:02] <CIA-4> nbd * r4040 /branches/buildroot-ng/openwrt/ (27 files in 15 dirs): move target/linux/image to target/image
[16:04] <CIA-4> nbd * r4041 /branches/buildroot-ng/openwrt/toolchain/uClibc/ (10 files in 3 dirs): move uClibc/files/config.* to uClibc/config/* and allow board-specific config overrides
[16:43] <nbd> xrg_: ping
[16:50] <xrg_> pong.
[16:50] <xrg_> What's up?
[16:51] <nbd> did sebastian tell you about the state of our current rb532 tree?
[16:51] <xrg_> Not yet.
[16:51] <xrg_> is it better than ever ;) ?
[16:51] <nbd> the performance is somewhat adequate now, but the ethernet ports on the daughterboard lock up under load
[16:52] <nbd> and it looks very much like a pci problem
[16:52] <xrg_> oops..
[16:52] <nbd> and the onboard ports are fine
[16:52] <nbd> it's the same behavior with or without our performance patches
[16:52] <nbd> just with the performance patches it's easier to trigger
[16:52] <nbd> because the lockup requires a certain amount of network load
[16:52] <nbd> processor load is irrelevant
[16:53] <xrg_> sth. like a concurrency problem, then?
[16:54] <xrg_> I wonder why does the VIA have so many bugs. It's not a platform specific chip, anyway..
[16:54] <nbd> i have patched via-rhine to use the special idt pci decoupled access stuff
[16:54] <nbd> i don't think it's a bug in the via chipset this time
[16:54] <nbd> it really looks like an issue with the pci bridge
[16:55] <nbd> and i'm sure that i have integrated tested all the driver changes from mikrotik
[16:58] <xrg_> Does the mikrotik have such a problem? can you reach the same load with it?
[16:58] <nbd> with routeros it works
[16:59] <xrg_> But that would be 2.4 anyway..
[16:59] <xrg_> way too different.
[16:59] <nbd> yeah
[17:00] <nbd> unfortunately i have no idea where to start looking for this problem
[17:01] <xrg_> Me neither ! :(
[17:02] <xrg_> What kind of lockup do you get (for a start)?
[17:03] <nbd> if i transfer for a while, the ethernet port just suddenly stops working
[17:03] <nbd> i think it's tx that dies
[17:03] <nbd> the system continues running
[17:03] <nbd> and ifconfig down/up can bring the port back to life temporarily
[17:04] <nbd> sometimes it dies after a few kb again
[17:04] <nbd> sometimes it can transfer a few megabytes again
[17:04] <nbd> pretty random
[17:04] <xrg_> Like a buffer overflow;
[17:04] <xrg_> or a disabled interrupt?
[17:04] <nbd> i checked the interrupts
[17:04] <nbd> and i increased the ring size as well
[17:05] <nbd> i tried adding a watchdog that reenables interrupts every half a second
[17:05] <nbd> didn't help
[17:14] <xrg_> I'm really sorry I can't help these days. I must finish sth by next week..
[17:14] <nbd> no problem
[17:14] <nbd> i was just telling you in case you had any idea where to look
[17:15] <xrg_> Really, no idea.
[17:16] <nbd> i think i'll start by trying to find a cheap workaround
[17:16] <xrg_> Can you look at the i/f counters when this happens? interrupts?
[17:16] <nbd> until i find the root cause of this
[17:16] <nbd> sometimes rx still works
[17:16] <nbd> but tx is dead
[17:16] <nbd> packet counters don't increase there
[21:56] <florian__> is there any redboot userland tool ?
[21:57] <CIA-4> nbd * r4042 /branches/buildroot-ng/openwrt/target/linux/ (2 files in 2 dirs): move jffs2 end-of-filesystem detection to the generic 2.6 patches
[21:59] <Kaloz> florian__: no
[21:59] <florian__> ah humm
[21:59] <florian__> I still can't interrupt my inventel livebox from booting :(
[21:59] <Kaloz> except that in the livebox
[21:59] <Kaloz> sure you can't
[21:59] <florian__> I want to test my custom kernel
[21:59] <Kaloz> your redboot has that option disabled
[21:59] <florian__> probably
[21:59] <florian__> I did not find the jtaf
[21:59] <florian__> jtag sorry
[21:59] <Kaloz> not probably, sure :)
[21:59] <Kaloz> there is no jtag :/
[22:00] <florian__> ah, you did not find one too ?
[22:00] <Kaloz> but at least mine have reboot prompt (after downgrading to an odler version)
[22:00] <Kaloz> i sent you that binary iirc
[22:00] <florian__> ah you sure ?
[22:01] <Kaloz> yes.. it's for the livebox, so it's a bit risky to be flashed on the livebox pro, but imho it should work
[22:01] <Kaloz> the only difference is the switch, right?
[22:01] <florian__> I have access to redboot config through mtd
[22:01] <florian__> I guess so
[22:02] <florian__> and the bluetooth by the way
[22:02] <Kaloz> well, redboot config won't do any good for you, as the binary has the shell disabed
[22:02] <florian__> yep
[22:02] <florian__> so I definitively have to change the redboot :(
[22:03] <Kaloz> should i send you the redboot binary? maybe you nuke the livebox, but without that you simply won't be able to get it work
[22:03] <florian__> well send it to me, we'll see what we'll do with it ;)
[22:04] <Kaloz> dcc?
[22:04] <florian__> yep
[22:05] <florian__> how do they appear under irssi ?
[22:05] <Kaloz> in status window
[22:05] <Kaloz> /dcc get should grab it
[22:06] <florian__> it's stalling
[22:07] <Kaloz> you didn't accepted it, yet :p
[22:07] <Kaloz> whatever
[22:07] <Kaloz> http://kaloz.rulez.org/files/boot.500rc11sp
[22:08] <florian__> thanks
[22:08] <Kaloz> sidenote
[22:08] <Kaloz> use dd, forget cat
[22:08] <Kaloz> and used /dev/mtdblock/foo
[22:08] <Kaloz> otherwise you won't get the righ data
[22:08] <florian__> I will first save the original bootloader
[22:09] <Kaloz> also, there is a builtin tftp client, backup the original redboot before
[22:09] <Kaloz> always verify everything with md5sum
[22:09] <CIA-4> mbm * r4043 /branches/ (buildroot-ng/openwrt/rules.mk whiterussian/openwrt/rules.mk): add STRIP to TARGET_CONFIGURE_OPTS (closes #606)
[22:09] <Kaloz> if you save the original one, and for some reason this won't work right (eg. no ethernet, as that's the only difference hopefully)
[22:09] <Kaloz> then you can send the original back via x/ymodem
[22:10] <Kaloz> but 99% it will simply work just fine :)
[22:10] <florian__> oh you know me, I am often the 1% ;)
[22:13] <Kaloz> only if you are trying to solder :P
[22:14] <florian__> thanks ;)
[22:18] <Kaloz> florian__: can you paste me the redboot prompt on bootup?
[22:18] <Kaloz> florian__: eg how does it identify itself
[22:18] <florian__> now ?
[22:18] <Kaloz> sure
[22:19] <Kaloz> i'm asking because the one i've changed from identified itself as BLUE5G.4_mod1 WITHOUT OPTION iirc
[22:19] <florian__> ok
[22:20] <florian__> I reboot the box
[22:20] <florian__> Platform: Blue_5g (MIPS32 4Kc)
[22:20] <Kaloz> after or before that
[22:20] <florian__> before flashing with the new
[22:21] <Kaloz> i mean before or after that line
[22:21] <Kaloz> there is a version info
[22:21] <Kaloz> or just pastebin it
[22:22] <florian__> http://pastebin.ca/68570
[22:22] <Kaloz> BLUE5G.6
[22:23] <florian__> right
[22:23] <Kaloz> md5sum /dev/mtdblock/0?
[22:23] <florian__> 852caf515c47fe3cdc79ceb957150b94
[22:24] <Kaloz> md5sum /dev/mtd0?
[22:24] <florian__> 852caf515c47fe3cdc79ceb957150b94
[22:24] <florian__> looks like the same
[22:25] <florian__> I have changed the bootloader yet
[22:25] <[mbm]> should be the exact same
[22:25] <Kaloz> mine was different (but an older redboot anyways)
[22:25] <Kaloz> have or haven't? :p
[22:25] <[mbm]> well, I mean mtd0 and mtdblock/0 are the same device
[22:26] <Kaloz> gave different values with the firmware version i had on it ;)
[22:26] <florian__> hu hu
[22:26] <Kaloz> it didn't accessed the same partition the same way
[22:26] <florian__> ok, so now it's time to change the bootloader
[22:26] <Kaloz> back'ed up the old one?
[22:27] <florian__> yep
[22:27] <florian__> dd if=new of=/dev/mtdblock0 ?
[22:27] <Kaloz> okay, dd the new one to mtdblock/0
[22:27] <Kaloz> yes, and do an md5sum on it
[22:27] <florian__> no additionnal options ?
[22:27] <Kaloz> well, size should be the same, so no options needed
[22:27] <florian__> 78552d093ca2738bce697b276a3af031
[22:27] <Kaloz> seems right
[22:28] <florian__> ok, so, let's reboot and pray
[22:28] <Kaloz> cross your fingers
[22:28] <Kaloz> :P
[22:28] Action: Kaloz crossed them
[22:28] <florian__> not really convenient ;)
[22:28] Action: [mbm] crosses the wires
[22:28] <Kaloz> :P
[22:28] <florian__> it boots, but I did not had the time to interrupt the bootloader :)
[22:28] <Kaloz> boots? perfct
[22:28] <Kaloz> :)
[22:29] <florian__> hop I am on it :)
[22:29] <florian__> now let's boot my custom 2.6 kernel :)
[22:29] <Kaloz> remove power, start hitting ctrl-c like mad and plug it back
[22:30] <florian__> redboot can load elf binaries right ?
[22:30] <Kaloz> it should be able to
[22:30] <florian__> ok, what about the -b option, what do I specify ?
[22:31] <Kaloz> for elf? nothing
[22:31] <Kaloz> load foo
[22:31] <florian__> ok
[22:31] <Kaloz> it will parse the elf headers
[22:32] <florian__> ok -f or -b ?
[22:32] <Kaloz> nothing? :P
[22:32] <florian__> -v ;)
[22:37] <florian__> humm I get network unreachable errors
[22:37] <Kaloz> maybe configure the network before :P
[22:37] <florian__> it's configured
[22:37] <florian__> I can't ping anything
[22:37] <florian__> I will have a test with the default IP address
[22:37] <Kaloz> ip_address -l localaddrss -h tftpserveraddress
[22:38] <Kaloz> also, it's possible that it got mapped to one of the other ports, check those too
[22:38] <florian__> ah yep
[22:38] <florian__> thatt's it ;)
[22:39] <Kaloz> works?
[22:39] <florian__> not with anything else than the default IP strange :)
[22:40] <Kaloz> who cares if it's working? :)
[22:40] <florian__> hu I care :)
[22:41] <florian__> ok it's downloading )
[22:41] <florian__> then I just type go, right ?
[22:42] <florian__> Kaloz: ?
[22:42] <florian__> I never know whether it's exec or go ?
[22:42] <Kaloz> "exec" preferably
[22:42] <florian__> exec
[22:42] <florian__> Now booting linux kernel: Base address 0x8000fc00 Entry 0x80207018 Cmdline : root=1F01 panic=2
[22:44] <florian__> looks like it does not work
[22:47] <florian__> Kaloz: do I need to change the kernel command line ?
[22:48] <Kaloz> dunn owhat's the default command line
[22:48] <florian__> I got it with the fconfig
[22:48] <florian__> exec -c "boot_loader=RedBoot root=1F01 mem=16M" 0x80010000
[22:50] <Kaloz> exec -c "boot_loader=RedBoot root=1F01 mem=16M console=ttyS0,115200"?
[22:51] <florian__> ah
[22:51] <florian__> humm :)
[22:52] <florian__> it looks like it either does not boot, or refuses to boot :(
[22:53] <florian__> I will merge preliminary support, and see that later
[23:01] <jr-> kaloz. is it possible to temporary change xscale machine type from redboot cli or only by replacing entire redboot?
[23:01] <Kaloz> jr-: what do oyu mean?
[23:02] <Kaloz> jr-: theoritically redboot should set the correct mach id before booting linux
[23:02] <jr-> kaloz. i have device that reports itself as 290 (coyote) to linux, but actually doesn't work with coyote config. it does work with generic ixdp425 config (245).
[23:03] <Kaloz> jr-: but on some boards they set the intel dev board's id or nothing, so you have to workaround that from linux
[23:04] <Kaloz> jr-: and it should be a 3rd platform for real :) yeah, saw stuff like that
[23:04] <Kaloz> jr-: most of the time with intel id and coyote config :)
[23:05] <jr-> i think they just used redboot from coyote but changed something so coyote config no longer works. it boots if i swap coyote and ixdp425 id's in mach-types and compile 2.6.17.1 kernel for ixdp425
[23:06] <Kaloz> ah, so openwrt is running on it?
[23:07] <jr-> not exactly. i compiled kernel using openwrt crosscompile tools and just got it booting over tftp. it's telewell tw-646 aka billion biguard2.
[23:07] <Kaloz> got an url?
[23:08] <jr-> http://www.billion.com/product/biguard/biguard2.htm
[23:08] <Kaloz> cpu/flash/ram?
[23:09] <jr-> 420 with 32M ram and 8M flash. wan + 8 lan ports (vlan capable switch inside with necessary tools to config it). empty minipci pads on board and pads for max3xxx
[23:09] <Kaloz> :)
[23:09] <Kaloz> 420?
[23:09] <Kaloz> it should be 422 or 425 for crypto
[23:09] <jr-> well then it was 422. no crypto.
[23:10] <jr-> biguard10 and 30 have 425. otherwise pretty much same hardware with just slightly differently configured vlans ("dual wan")
[23:11] <Kaloz> btw for real the best would be a fixed reboot for it
[23:11] <jr-> i have another xscale router too. it's even more perverted. hacked redboot that loads some wince like os (clavister) from compact flash that's connected to promise ata66 pci chip.
[23:11] <Kaloz> lol
[23:12] <jr-> cf has only couple megabytes data and it has onboard 4MB flash chip that contains only redboot and mac addresses. everything could be on onboard flash but for some reason they wanted to use promise+cf
[23:12] <common> cool
[23:12] <common> should be possible to attach a ide drive
[23:13] <jr-> that one is d-link dfl-200. and talking about weird devices there's dfl-700 too in some pile. ns geode with 16M apacer flash connected to ide bus. just that connector on apacer flash is like 2.5" hdd but female, not male like hdd's do.
[23:14] <jr-> common. you would need adapter with cf-female and 2.5" hdd male (for cabel) or female (directly to disk). they're usually opposite.
[23:14] <common> crimping ide cables is easy
[23:14] <Kaloz> jr-: heh, these stuff are way more expensive then what they worth
[23:15] <common> Kaloz:
[23:15] <common> i got a diff on the ixosal sources to prevent the driver from keeping in d state
[23:15] <jr-> common. you would need to solder wires on one end to cf-connector.
[23:16] <Kaloz> common: nice
[23:16] <common> wthout it i had a load of 1.0 in idle as the driver took it
[23:17] <common> http://phpfi.com/125075
[23:20] <common> I'm looking forward to buildroot-ng for ixp4xx, openixp is pretty dead, and a lot of things would need fixes
[23:24] <jr-> kaloz. ok it boots openwrt kernel too. i used buildroot-ng and added missing lines+files from kamikaze. http://pastebin.ca/68611
[23:25] <Kaloz> yeah, i have to port the gateway7001 stuff over to .17
[23:25] <Kaloz> and submit it upstream
[00:00] --- Thu Jun 22 2006
[01:49] <CIA-4> nbd * r4029 /branches/buildroot-ng/openwrt/toolchain/ (Makefile sed/Makefile sed/sedcheck.sh utils/Makefile): add toolchain/download target
[01:56] <CIA-4> nbd * r4030 /branches/buildroot-ng/openwrt/target/ (8 files in 6 dirs): add target/download
[01:57] <CIA-4> nbd * r4031 /branches/buildroot-ng/openwrt/Makefile: add global make download target
[04:16] <CIA-4> nbd * r4032 / (96 files in 77 dirs): build system cleanup. move shared include files into $(TOPDIR)/include, move lzma, mkfs.* into toolchain/
[04:32] <CIA-4> nbd * r4033 / (65 files in 63 dirs): add INCLUDE_DIR variable for $(TOPDIR)/include
[04:45] <CIA-4> nbd * r4034 /branches/buildroot-ng/openwrt/scripts/gen_deps.pl: add sdk option for gen_deps.pl
[05:01] <CIA-4> nbd * r4035 /branches/buildroot-ng/openwrt/scripts/gen_deps.pl: don't generate invalid dependency lines in sdk mode
[05:13] <CIA-4> nbd * r4036 /branches/buildroot-ng/openwrt/target/linux/image/ (8 files in 8 dirs): include $(TOPDIR)/rules.mk in the image makefiles
[05:17] <CIA-4> nbd * r4037 /branches/buildroot-ng/openwrt/ (8 files in 6 dirs): fix and enable the sdk
[08:19] <CIA-4> nbd * r4038 /branches/buildroot-ng/openwrt/ (57 files in 48 dirs): massive cleanup of toolchain/
[08:25] <[mbm]> hmm.. that is a rather large change
[08:25] <nbd> yeah
[08:25] <nbd> i converted the bitrot to something cleaner, more buildroot-ng style :)
[08:32] <nbd> [mbm]: btw. about the verbosity control in the build system: couldn't we just use an unused file descriptor for passing the build system trace messages, like echo ... >&3 and then redirecting it to stderr from the main make process?
[08:33] <nbd> or is that something that the shell can't do?
[08:37] <[mbm]> don't think the shell will let you do anything other than 0-2
[08:38] <nbd> hmm... this works:
[08:38] <nbd> ( (echo foo >&3 ) 2>/dev/null >/dev/null ) 3>&1
[08:39] <[mbm]> interesting
[08:39] <nbd> tried it with zsh and bash
[08:39] <nbd> if we could use that, it would make the whole tracing stuff a lot easier
[08:39] <[mbm]> yeah
[08:41] <[mbm]> I'd still like to replace "$MAKE foo" with something more like "$(call make,foo)" just to give us more control
[08:41] <nbd> yeah
[08:42] <[mbm]> and I can't think of any value for $MAKE that would allow that
[08:42] <[mbm]> the cloest would be to define $MAKE as external script
[08:43] <nbd> the thing is: we don't have to change $MAKE for that
[08:43] <nbd> we just patch the places for the buildsystem-internal calls
[08:43] <nbd> and do the rest in include/package.mk
[08:43] <[mbm]> yeah, was thinking about that
[08:45] <nbd> i think when we got this thing implemented and when i've fixed the image builder, we can move trunk to someplace else and replace it with buildroot-ng
[08:45] <nbd> and then slowly start porting all the packages over
[08:45] <[mbm]> yeah I was thinking of just overwriting trunk with buildroot-ng
[08:46] <nbd> we could move it to tags/kamikaze-old
[08:46] <nbd> easier than overwriting
[08:46] <[mbm]> ok
[08:46] Action: [mbm] hasn't played with svn much
[08:46] <nbd> with most of the stuff i made sure that i used svn cp, so that history in most places is preserved
[10:01] <florian__> [mbm], nbd do you well know mips assembly ?
[10:01] <florian__> I am trying to compile a brcm63xx kernel
[10:02] <florian__> and get some errors I am not able to solve
[10:02] <florian__> AS arch/mips/kernel/scall32-o32.o
[10:02] <florian__> arch/mips/kernel/scall32-o32.S: Assembler messages:
[10:02] <florian__> arch/mips/kernel/scall32-o32.S:654: Error: Macro with this name was already defined
[10:02] <florian__> here is the ASM code :
[10:04] <florian__> damn pastebin is lagging
[10:04] <nbd> use a different one
[10:04] <nbd> pastebin.com is slow
[10:05] <florian__> http://pastebin.com/722842
[10:07] <nbd> did you modify that file?
[10:07] <florian__> let me check if is by the patch
[10:08] <florian__> yes it is modified
[10:09] <nbd> got a diff?
[10:09] <florian__> yep
[10:10] <florian__> http://pastebin.ca/68237
[10:10] <florian__> looks like a linux-mips diff
[10:11] <florian__> not broadcom
[10:12] <florian__> and it's not causing the error message
[10:18] <florian__> I will see how it has changed through differente kernel versions
[10:23] <florian__> I should upgrade the 2.6.8.1 patches to 2.6.17
[10:47] <florian__> I really love broadcom people commenting their code "this driver is simple because it is simple"
[11:27] <florian__> I found this in broadcom patches : - __EXPORT_SYMBOL(sym, "_gpl")+ __EXPORT_SYMBOL(sym, "")
[11:27] <florian__> do they have the right to do this ?
[11:32] <common> use phpfi.com
[11:45] <Kaloz> florian__: told you this 2 months ago at least :P
[11:45] <Kaloz> florian__: and no, they have no right to do this
[11:46] <florian__> Kaloz: told me what ?
[11:46] <Kaloz> that symbol change
[11:46] <florian__> ah right
[11:47] <florian__> do their module work if we don't patch the module param as they wish ?
[11:47] <Kaloz> i doubt
[11:48] <florian__> ok
[11:49] <florian__> I found a cleaner brcm63xx source where there was less calls to their external tools
[12:30] <common> Kaloz do you plan adding xscale to the buildroot-ng enviroment?
[12:37] <Kaloz> sure
[12:40] <common> good
[12:49] <wbx> is it possible to get commit-messages via e-mail?
[12:56] <common> svn hooks
[12:57] <wbx> common: i mean messages to a mailaccount, when a openwrt developer commit sth.
[12:57] <wbx> common: i know how to configure this on my own servers.
[12:59] <common> it would be possible, but i doubt somebody is willing to do you this favour
[13:08] <florian__> no troll
[13:20] <alxhh> a commit-list would be nice
[13:22] <florian__> right
[13:26] <dbcch> aren't commit messages broadcast to #openwrt-devel?
[13:26] <florian__> usually they are when CIA-4 is up and running
[13:26] <dbcch> i guess you coudl use those as the basis for an emailer
[13:27] <dbcch> i.e. mirc script
[13:30] <florian__> mirc hu hu
[13:31] <dbcch> ;p
[13:31] <dbcch> I mean, x-chat ;p
[14:06] <common> its pretty easy to use svn hooks server side to mail them to a list
[14:07] <nbd> yeah
[14:14] <db90h> like u said, as long as someone is around to do him that favor ;p
[14:20] <alxhh> db90h: stop trolling, wbx is a friend, nor an enemy
[14:20] Action: nbd found that a feed reader is much more convenient for keeping track of svn changes
[14:21] <alxhh> and, separating the projekts by childish behavior is dumb, stupid, contraproduktive
[14:22] <alxhh> viewing freewrt as a place, where things can be tried in a different way to see different approaches and how they work is the way
[14:23] <alxhh> s/nor/not/
[14:24] <db90h> oh, that's what florian meant by 'no troll'.. my bad
[14:25] <db90h> i wondered what the hell he was talking about, i didn't understand it was a sensitive issue
[14:25] <alxhh> and, a list for commits help those, who are not online 24/7
[14:25] <Kaloz> alxhh: well, you know there is a reason why i love our language
[14:26] <alxhh> Kaloz: which language?
[14:26] <Kaloz> hungarian
[14:26] <db90h> Klingon ;p
[14:26] <Kaloz> alxhh: while english has only "friend" really, hungarian has "ismeros" and "barat"
[14:27] <db90h> is one like a good acquantaince or something?
[14:27] <Kaloz> db90h: something like that, but it doesn't mean you have to like the other one, just you know him
[14:28] <Kaloz> db90h: while the second one is about a close link
[14:29] <db90h> ah, ic
[14:30] <Kaloz> so we really distinguish the two
[15:02] <CIA-4> nbd * r4039 /branches/buildroot-ng/openwrt/Config.in: reorder sdk menuconfig entry, so that it does not force the toolchain selection into the main menu
[15:07] <Kaloz> hmz
[15:09] <nbd> hmz?
[15:11] <Kaloz> didn't think you will be that fast, i was adding back the xscale stuff to brng
[15:11] <Kaloz> :P
[15:11] <nbd> :)
[15:11] <nbd> btw. do you like the toolchain/ rewrite?
[15:13] <Kaloz> yep
[16:02] <CIA-4> nbd * r4040 /branches/buildroot-ng/openwrt/ (27 files in 15 dirs): move target/linux/image to target/image
[16:04] <CIA-4> nbd * r4041 /branches/buildroot-ng/openwrt/toolchain/uClibc/ (10 files in 3 dirs): move uClibc/files/config.* to uClibc/config/* and allow board-specific config overrides
[16:43] <nbd> xrg_: ping
[16:50] <xrg_> pong.
[16:50] <xrg_> What's up?
[16:51] <nbd> did sebastian tell you about the state of our current rb532 tree?
[16:51] <xrg_> Not yet.
[16:51] <xrg_> is it better than ever ;) ?
[16:51] <nbd> the performance is somewhat adequate now, but the ethernet ports on the daughterboard lock up under load
[16:52] <nbd> and it looks very much like a pci problem
[16:52] <xrg_> oops..
[16:52] <nbd> and the onboard ports are fine
[16:52] <nbd> it's the same behavior with or without our performance patches
[16:52] <nbd> just with the performance patches it's easier to trigger
[16:52] <nbd> because the lockup requires a certain amount of network load
[16:52] <nbd> processor load is irrelevant
[16:53] <xrg_> sth. like a concurrency problem, then?
[16:54] <xrg_> I wonder why does the VIA have so many bugs. It's not a platform specific chip, anyway..
[16:54] <nbd> i have patched via-rhine to use the special idt pci decoupled access stuff
[16:54] <nbd> i don't think it's a bug in the via chipset this time
[16:54] <nbd> it really looks like an issue with the pci bridge
[16:55] <nbd> and i'm sure that i have integrated tested all the driver changes from mikrotik
[16:58] <xrg_> Does the mikrotik have such a problem? can you reach the same load with it?
[16:58] <nbd> with routeros it works
[16:59] <xrg_> But that would be 2.4 anyway..
[16:59] <xrg_> way too different.
[16:59] <nbd> yeah
[17:00] <nbd> unfortunately i have no idea where to start looking for this problem
[17:01] <xrg_> Me neither ! :(
[17:02] <xrg_> What kind of lockup do you get (for a start)?
[17:03] <nbd> if i transfer for a while, the ethernet port just suddenly stops working
[17:03] <nbd> i think it's tx that dies
[17:03] <nbd> the system continues running
[17:03] <nbd> and ifconfig down/up can bring the port back to life temporarily
[17:04] <nbd> sometimes it dies after a few kb again
[17:04] <nbd> sometimes it can transfer a few megabytes again
[17:04] <nbd> pretty random
[17:04] <xrg_> Like a buffer overflow;
[17:04] <xrg_> or a disabled interrupt?
[17:04] <nbd> i checked the interrupts
[17:04] <nbd> and i increased the ring size as well
[17:05] <nbd> i tried adding a watchdog that reenables interrupts every half a second
[17:05] <nbd> didn't help
[17:14] <xrg_> I'm really sorry I can't help these days. I must finish sth by next week..
[17:14] <nbd> no problem
[17:14] <nbd> i was just telling you in case you had any idea where to look
[17:15] <xrg_> Really, no idea.
[17:16] <nbd> i think i'll start by trying to find a cheap workaround
[17:16] <xrg_> Can you look at the i/f counters when this happens? interrupts?
[17:16] <nbd> until i find the root cause of this
[17:16] <nbd> sometimes rx still works
[17:16] <nbd> but tx is dead
[17:16] <nbd> packet counters don't increase there
[21:56] <florian__> is there any redboot userland tool ?
[21:57] <CIA-4> nbd * r4042 /branches/buildroot-ng/openwrt/target/linux/ (2 files in 2 dirs): move jffs2 end-of-filesystem detection to the generic 2.6 patches
[21:59] <Kaloz> florian__: no
[21:59] <florian__> ah humm
[21:59] <florian__> I still can't interrupt my inventel livebox from booting :(
[21:59] <Kaloz> except that in the livebox
[21:59] <Kaloz> sure you can't
[21:59] <florian__> I want to test my custom kernel
[21:59] <Kaloz> your redboot has that option disabled
[21:59] <florian__> probably
[21:59] <florian__> I did not find the jtaf
[21:59] <florian__> jtag sorry
[21:59] <Kaloz> not probably, sure :)
[21:59] <Kaloz> there is no jtag :/
[22:00] <florian__> ah, you did not find one too ?
[22:00] <Kaloz> but at least mine have reboot prompt (after downgrading to an odler version)
[22:00] <Kaloz> i sent you that binary iirc
[22:00] <florian__> ah you sure ?
[22:01] <Kaloz> yes.. it's for the livebox, so it's a bit risky to be flashed on the livebox pro, but imho it should work
[22:01] <Kaloz> the only difference is the switch, right?
[22:01] <florian__> I have access to redboot config through mtd
[22:01] <florian__> I guess so
[22:02] <florian__> and the bluetooth by the way
[22:02] <Kaloz> well, redboot config won't do any good for you, as the binary has the shell disabed
[22:02] <florian__> yep
[22:02] <florian__> so I definitively have to change the redboot :(
[22:03] <Kaloz> should i send you the redboot binary? maybe you nuke the livebox, but without that you simply won't be able to get it work
[22:03] <florian__> well send it to me, we'll see what we'll do with it ;)
[22:04] <Kaloz> dcc?
[22:04] <florian__> yep
[22:05] <florian__> how do they appear under irssi ?
[22:05] <Kaloz> in status window
[22:05] <Kaloz> /dcc get should grab it
[22:06] <florian__> it's stalling
[22:07] <Kaloz> you didn't accepted it, yet :p
[22:07] <Kaloz> whatever
[22:07] <Kaloz> http://kaloz.rulez.org/files/boot.500rc11sp
[22:08] <florian__> thanks
[22:08] <Kaloz> sidenote
[22:08] <Kaloz> use dd, forget cat
[22:08] <Kaloz> and used /dev/mtdblock/foo
[22:08] <Kaloz> otherwise you won't get the righ data
[22:08] <florian__> I will first save the original bootloader
[22:09] <Kaloz> also, there is a builtin tftp client, backup the original redboot before
[22:09] <Kaloz> always verify everything with md5sum
[22:09] <CIA-4> mbm * r4043 /branches/ (buildroot-ng/openwrt/rules.mk whiterussian/openwrt/rules.mk): add STRIP to TARGET_CONFIGURE_OPTS (closes #606)
[22:09] <Kaloz> if you save the original one, and for some reason this won't work right (eg. no ethernet, as that's the only difference hopefully)
[22:09] <Kaloz> then you can send the original back via x/ymodem
[22:10] <Kaloz> but 99% it will simply work just fine :)
[22:10] <florian__> oh you know me, I am often the 1% ;)
[22:13] <Kaloz> only if you are trying to solder :P
[22:14] <florian__> thanks ;)
[22:18] <Kaloz> florian__: can you paste me the redboot prompt on bootup?
[22:18] <Kaloz> florian__: eg how does it identify itself
[22:18] <florian__> now ?
[22:18] <Kaloz> sure
[22:19] <Kaloz> i'm asking because the one i've changed from identified itself as BLUE5G.4_mod1 WITHOUT OPTION iirc
[22:19] <florian__> ok
[22:20] <florian__> I reboot the box
[22:20] <florian__> Platform: Blue_5g (MIPS32 4Kc)
[22:20] <Kaloz> after or before that
[22:20] <florian__> before flashing with the new
[22:21] <Kaloz> i mean before or after that line
[22:21] <Kaloz> there is a version info
[22:21] <Kaloz> or just pastebin it
[22:22] <florian__> http://pastebin.ca/68570
[22:22] <Kaloz> BLUE5G.6
[22:23] <florian__> right
[22:23] <Kaloz> md5sum /dev/mtdblock/0?
[22:23] <florian__> 852caf515c47fe3cdc79ceb957150b94
[22:24] <Kaloz> md5sum /dev/mtd0?
[22:24] <florian__> 852caf515c47fe3cdc79ceb957150b94
[22:24] <florian__> looks like the same
[22:25] <florian__> I have changed the bootloader yet
[22:25] <[mbm]> should be the exact same
[22:25] <Kaloz> mine was different (but an older redboot anyways)
[22:25] <Kaloz> have or haven't? :p
[22:25] <[mbm]> well, I mean mtd0 and mtdblock/0 are the same device
[22:26] <Kaloz> gave different values with the firmware version i had on it ;)
[22:26] <florian__> hu hu
[22:26] <Kaloz> it didn't accessed the same partition the same way
[22:26] <florian__> ok, so now it's time to change the bootloader
[22:26] <Kaloz> back'ed up the old one?
[22:27] <florian__> yep
[22:27] <florian__> dd if=new of=/dev/mtdblock0 ?
[22:27] <Kaloz> okay, dd the new one to mtdblock/0
[22:27] <Kaloz> yes, and do an md5sum on it
[22:27] <florian__> no additionnal options ?
[22:27] <Kaloz> well, size should be the same, so no options needed
[22:27] <florian__> 78552d093ca2738bce697b276a3af031
[22:27] <Kaloz> seems right
[22:28] <florian__> ok, so, let's reboot and pray
[22:28] <Kaloz> cross your fingers
[22:28] <Kaloz> :P
[22:28] Action: Kaloz crossed them
[22:28] <florian__> not really convenient ;)
[22:28] Action: [mbm] crosses the wires
[22:28] <Kaloz> :P
[22:28] <florian__> it boots, but I did not had the time to interrupt the bootloader :)
[22:28] <Kaloz> boots? perfct
[22:28] <Kaloz> :)
[22:29] <florian__> hop I am on it :)
[22:29] <florian__> now let's boot my custom 2.6 kernel :)
[22:29] <Kaloz> remove power, start hitting ctrl-c like mad and plug it back
[22:30] <florian__> redboot can load elf binaries right ?
[22:30] <Kaloz> it should be able to
[22:30] <florian__> ok, what about the -b option, what do I specify ?
[22:31] <Kaloz> for elf? nothing
[22:31] <Kaloz> load foo
[22:31] <florian__> ok
[22:31] <Kaloz> it will parse the elf headers
[22:32] <florian__> ok -f or -b ?
[22:32] <Kaloz> nothing? :P
[22:32] <florian__> -v ;)
[22:37] <florian__> humm I get network unreachable errors
[22:37] <Kaloz> maybe configure the network before :P
[22:37] <florian__> it's configured
[22:37] <florian__> I can't ping anything
[22:37] <florian__> I will have a test with the default IP address
[22:37] <Kaloz> ip_address -l localaddrss -h tftpserveraddress
[22:38] <Kaloz> also, it's possible that it got mapped to one of the other ports, check those too
[22:38] <florian__> ah yep
[22:38] <florian__> thatt's it ;)
[22:39] <Kaloz> works?
[22:39] <florian__> not with anything else than the default IP strange :)
[22:40] <Kaloz> who cares if it's working? :)
[22:40] <florian__> hu I care :)
[22:41] <florian__> ok it's downloading )
[22:41] <florian__> then I just type go, right ?
[22:42] <florian__> Kaloz: ?
[22:42] <florian__> I never know whether it's exec or go ?
[22:42] <Kaloz> "exec" preferably
[22:42] <florian__> exec
[22:42] <florian__> Now booting linux kernel: Base address 0x8000fc00 Entry 0x80207018 Cmdline : root=1F01 panic=2
[22:44] <florian__> looks like it does not work
[22:47] <florian__> Kaloz: do I need to change the kernel command line ?
[22:48] <Kaloz> dunn owhat's the default command line
[22:48] <florian__> I got it with the fconfig
[22:48] <florian__> exec -c "boot_loader=RedBoot root=1F01 mem=16M" 0x80010000
[22:50] <Kaloz> exec -c "boot_loader=RedBoot root=1F01 mem=16M console=ttyS0,115200"?
[22:51] <florian__> ah
[22:51] <florian__> humm :)
[22:52] <florian__> it looks like it either does not boot, or refuses to boot :(
[22:53] <florian__> I will merge preliminary support, and see that later
[23:01] <jr-> kaloz. is it possible to temporary change xscale machine type from redboot cli or only by replacing entire redboot?
[23:01] <Kaloz> jr-: what do oyu mean?
[23:02] <Kaloz> jr-: theoritically redboot should set the correct mach id before booting linux
[23:02] <jr-> kaloz. i have device that reports itself as 290 (coyote) to linux, but actually doesn't work with coyote config. it does work with generic ixdp425 config (245).
[23:03] <Kaloz> jr-: but on some boards they set the intel dev board's id or nothing, so you have to workaround that from linux
[23:04] <Kaloz> jr-: and it should be a 3rd platform for real :) yeah, saw stuff like that
[23:04] <Kaloz> jr-: most of the time with intel id and coyote config :)
[23:05] <jr-> i think they just used redboot from coyote but changed something so coyote config no longer works. it boots if i swap coyote and ixdp425 id's in mach-types and compile 2.6.17.1 kernel for ixdp425
[23:06] <Kaloz> ah, so openwrt is running on it?
[23:07] <jr-> not exactly. i compiled kernel using openwrt crosscompile tools and just got it booting over tftp. it's telewell tw-646 aka billion biguard2.
[23:07] <Kaloz> got an url?
[23:08] <jr-> http://www.billion.com/product/biguard/biguard2.htm
[23:08] <Kaloz> cpu/flash/ram?
[23:09] <jr-> 420 with 32M ram and 8M flash. wan + 8 lan ports (vlan capable switch inside with necessary tools to config it). empty minipci pads on board and pads for max3xxx
[23:09] <Kaloz> :)
[23:09] <Kaloz> 420?
[23:09] <Kaloz> it should be 422 or 425 for crypto
[23:09] <jr-> well then it was 422. no crypto.
[23:10] <jr-> biguard10 and 30 have 425. otherwise pretty much same hardware with just slightly differently configured vlans ("dual wan")
[23:11] <Kaloz> btw for real the best would be a fixed reboot for it
[23:11] <jr-> i have another xscale router too. it's even more perverted. hacked redboot that loads some wince like os (clavister) from compact flash that's connected to promise ata66 pci chip.
[23:11] <Kaloz> lol
[23:12] <jr-> cf has only couple megabytes data and it has onboard 4MB flash chip that contains only redboot and mac addresses. everything could be on onboard flash but for some reason they wanted to use promise+cf
[23:12] <common> cool
[23:12] <common> should be possible to attach a ide drive
[23:13] <jr-> that one is d-link dfl-200. and talking about weird devices there's dfl-700 too in some pile. ns geode with 16M apacer flash connected to ide bus. just that connector on apacer flash is like 2.5" hdd but female, not male like hdd's do.
[23:14] <jr-> common. you would need adapter with cf-female and 2.5" hdd male (for cabel) or female (directly to disk). they're usually opposite.
[23:14] <common> crimping ide cables is easy
[23:14] <Kaloz> jr-: heh, these stuff are way more expensive then what they worth
[23:15] <common> Kaloz:
[23:15] <common> i got a diff on the ixosal sources to prevent the driver from keeping in d state
[23:15] <jr-> common. you would need to solder wires on one end to cf-connector.
[23:16] <Kaloz> common: nice
[23:16] <common> wthout it i had a load of 1.0 in idle as the driver took it
[23:17] <common> http://phpfi.com/125075
[23:20] <common> I'm looking forward to buildroot-ng for ixp4xx, openixp is pretty dead, and a lot of things would need fixes
[23:24] <jr-> kaloz. ok it boots openwrt kernel too. i used buildroot-ng and added missing lines+files from kamikaze. http://pastebin.ca/68611
[23:25] <Kaloz> yeah, i have to port the gateway7001 stuff over to .17
[23:25] <Kaloz> and submit it upstream
[00:00] --- Thu Jun 22 2006