**** BEGIN LOGGING AT Sat Mar 31 02:59:59 2012
Mar 31 07:00:21 <builder_> build #134 of rb532 is complete: Failure [failed compile_10]  Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/134
Mar 31 07:49:24 <builder_> build #117 of au1000 is complete: Failure [failed compile_4]  Build details are at http://buildbot.openwrt.org:8010/builders/au1000/builds/117
Mar 31 08:34:37 <builder_> build #139 of ramips is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org:8010/builders/ramips/builds/139
Mar 31 09:38:55 <builder_> build #105 of mpc52xx is complete: Failure [failed compile_4]  Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/105
Mar 31 10:41:51 <Lekensteyn> what is the status for USB support on AR7?
Mar 31 10:48:35 <Lekensteyn> let's start earlier, what is the status of PCI support on the AR7...?
Mar 31 10:57:30 <KanjiMonster> Lekensteyn: afaik AR7 does not have PCI, just something looking similar (this vlync stuff)
Mar 31 10:57:50 <KanjiMonster> it should work though
Mar 31 10:58:23 <Lekensteyn> I managed to get my device booted, but I do not see any log entry of PCI https://forum.openwrt.org/viewtopic.php?pid=162716#p162716
Mar 31 10:58:34 <Lekensteyn> pleaseh ave a look if you've time for it
Mar 31 11:01:06 <KanjiMonster> Lekensteyn: if, then you should see an entry for vlynq, not for pci
Mar 31 11:05:26 <Lekensteyn> I see, what about the NIC driver? machine details can be found on http://wiki.openwrt.org/toh/siemens/sx551
Mar 31 11:05:55 <Lekensteyn> brb lunch
Mar 31 12:39:42 <danitool> KanjiMonster: do you know if there is a CP0 register related to USB in bcm6358 boards?
Mar 31 13:29:59 <KanjiMonster> danitool: no
Mar 31 13:30:10 <KanjiMonster> danitool: or rather, I know there isn't
Mar 31 14:03:45 <[florian]> danitool: cp0 really is about the processor
Mar 31 14:04:12 <[florian]> danitool: I would rather look at the vendor sources (if any) and see if they are not enabling some regulator of any kind through a gpio
Mar 31 14:04:36 <danitool> that is what I'm doing... but I find nothing
Mar 31 14:04:52 <[florian]> if it's not a gpio, it can be some magic writing in some registers
Mar 31 14:05:29 <[florian]> maybe the original bootlog can also sched some light on this
Mar 31 14:05:32 <Lekensteyn> [florian]: I managed to boot the machine :) Now, have you recommendations on building a small initrd? (tools?)
Mar 31 14:06:01 <[florian]> Lekensteyn: ok, cool, so what did you need to change in the end?
Mar 31 14:06:10 <[florian]> Lekensteyn: I would recommend using the initramfs target
Mar 31 14:06:16 <[florian]> so everything runs in ram
Mar 31 14:06:36 <Lekensteyn> disabling parsing of a0,a1,a2,a3 (fw_argX) in prom_init
Mar 31 14:06:43 <Lekensteyn> described at https://forum.openwrt.org/viewtopic.php?pid=162716#p162716
Mar 31 14:06:59 <Lekensteyn> the bootloader possibly does not fill those regs (or garbage)
Mar 31 14:07:31 <[florian]> yes it's possible
Mar 31 14:08:20 <Lekensteyn> I'm currently generating a firmware image manually for the device (from build_dir/linux-ar7/vmlinux), I believe that I need a C runtime and busybox to get started?
Mar 31 14:08:48 <Lekensteyn> +/init file
Mar 31 14:08:52 <[florian]> yes
Mar 31 14:09:09 <[florian]> the base root filesystem comprises busybox + dnsmasq, iptables ...
Mar 31 14:09:14 <[florian]> enough to play with it
Mar 31 14:09:33 <Lekensteyn> oh wait, is that staging_dir or sth?
Mar 31 14:10:12 <[florian]> staging_dir contains libraries, that's build_dir/target-../root-ar7
Mar 31 16:09:12 <Lekensteyn> how do I combine an initrd with vmlinux? (binary, no ELF) I keep getting the error "no initrd found" and on the end it fails because no rootfs can be found
Mar 31 18:00:07 <[florian]> Lekensteyn: you'd rather use the initramfs feature, which embeds your rootfs in the kernel directly
Mar 31 18:00:38 <[florian]> make menuconfig -> Target images -> initramfs
Mar 31 18:03:32 <Lekensteyn> does that work? I had no luck with it :?
Mar 31 18:03:45 <[florian]> of course it works
Mar 31 18:03:46 <Lekensteyn> in which file is the image written then?
Mar 31 18:03:52 <[florian]> in vmlinux directly
Mar 31 18:03:58 <Lekensteyn> in the root folder?
Mar 31 18:04:17 <[florian]> in build_dir/linux-ar7/linux-2.6.x.y/vmlinux
Mar 31 18:04:42 <[florian]> the idea to create a cpio archive of the rootfs which is later linked to the kernel
Mar 31 18:04:49 <[florian]> that makes a bigger kenerl
Mar 31 18:04:54 <[florian]> but it avoids loading an initrd
Mar 31 18:05:05 <[florian]> which usually involves help from the bootloader to setup regiters
Mar 31 18:05:11 <Lekensteyn> ow. now you're mentioning, objcopy has likely stripped it when I used it
Mar 31 18:05:29 <[florian]> no, objcopy only strips what's unecessary
Mar 31 18:05:32 <[florian]> like debugging sections and such
Mar 31 18:05:41 <[florian]> the initramfs image is in a read-only data section
Mar 31 18:06:07 <Lekensteyn> objcopy -O binary -R .reginfo -R .notes -R .note -R     .comment -R .mdebug -R .note.gnu.build-id -S vmlinux /tmp/vmlinux
Mar 31 18:06:15 <Lekensteyn> that?
Mar 31 18:06:41 <[florian]> that line does not remove anything which is useful and necessary to the kernel
Mar 31 18:07:18 <[florian]> you *really* need to run make menuconfig -> Target images -> initramfs to get your kernel to link with your rootfs
Mar 31 18:08:07 <[florian]> because that will be enabling the corresponding CONFIG_* symbols to get the kernel to link the rootfs
Mar 31 18:08:11 <Lekensteyn> yay
Mar 31 18:08:50 <Lekensteyn> dejavu
Mar 31 18:10:24 <Lekensteyn> for clarity, I'm now using the kernel source tree (from openwrt) directly, not openwrt' Makefile. Does that make a differnece?
Mar 31 18:11:30 <[florian]> are you using openwrt for building your rootfs somehow?
Mar 31 18:12:37 <[florian]> but yeah, that makes a big difference, because it's openwrt makefile which is configuring the kernel for initramfs
Mar 31 18:13:16 <Lekensteyn> nope, since the image grew too large I've manually built busybox (statically) with almost no options and added it in an empty directory
Mar 31 18:13:57 <[florian]> what about disabling packages you don't need?
Mar 31 18:14:03 <[florian]> are not you loading the image from network?
Mar 31 18:14:31 <Lekensteyn> haven't tried that since it's not an option (the device should be an independent host)
Mar 31 18:14:46 <[florian]> how are you loading the image then?
Mar 31 18:15:12 <[florian]> I mean, you are obviously in a development phase, if you can't load the image easily, either using serial port, or network
Mar 31 18:15:19 <[florian]> then it's definitively going to be painful
Mar 31 18:16:31 <Lekensteyn> oh, I'm building the firmware based on the mkfirm program and then upload it via recovery mode. The kernel boots, but cannot find a rootfs nor initrd
Mar 31 18:17:11 <[florian]> of course, the image you are flashing does only contain the kernel
Mar 31 18:18:01 <Lekensteyn> right, so I just manually appended the initrd and adjusted rd_start and rd_size using dd
Mar 31 18:18:20 <Lekensteyn> ("appended" -> aligned at a 4k boundary after the uncompressed vmlinux)
Mar 31 18:19:47 <Lekensteyn> the image is uncompressed at address 0x94000000 so I thought setting rd_start=0x942a3000 rd_size=168922 would be ok (first number is the offset in the file+base addr, second is the lzma-compressed initrd being appended in bytes)
Mar 31 18:20:00 <Lekensteyn> now I get "Initramfs unpacking failed: junk in compressed archive"
Mar 31 18:20:10 <Lekensteyn> brb
Mar 31 18:20:27 <[florian]> you might want to set the kernel command line to root=/dev/ram0 to force the initrd probing
Mar 31 18:22:44 <[florian]> and check that CONFIG_INITRD and friends are enabled
Mar 31 19:23:20 <philipp64|laptop> [florian]: can you please have a look at 1994?
Mar 31 19:25:26 <KanjiMonster> 1994? did anything interesting happen back then?
Mar 31 20:17:09 <builder_> build #115 of iop32x is complete: Failure [failed compile_10]  Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/115
Mar 31 20:34:43 <philipp64|laptop> KanjiMonster: I think someone was born.
Mar 31 20:37:52 <Lekensteyn> [florian]: is rd_start the base address inside the kernel (0+offset) or in the RAM (0x94000000+offset)?
Mar 31 21:05:37 <builder_> build #151 of s3c24xx is complete: Failure [failed compile_4]  Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/151
Mar 31 22:05:55 <builder_> build #150 of at91 is complete: Failure [failed compile_4]  Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/150
Mar 31 22:08:23 <builder_> build #148 of ubicom32 is complete: Failure [failed compile_1]  Build details are at http://buildbot.openwrt.org:8010/builders/ubicom32/builds/148
Mar 31 23:54:41 <builder_> build #102 of octeon is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/102
Apr 01 00:17:37 <builder_> build #107 of adm5120 is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/107
Apr 01 00:18:00 <builder_> build #144 of cobalt is complete: Failure [failed compile_1]  Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/144
Apr 01 00:19:48 <builder_> build #139 of ps3 is complete: Failure [failed compile_1]  Build details are at http://buildbot.openwrt.org:8010/builders/ps3/builds/139
Apr 01 00:26:34 <builder_> build #129 of uml is complete: Exception [exception interrupted]  Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/129
Apr 01 01:02:43 <builder_> build #130 of x86 is complete: Failure [failed compile_1]  Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/130
Apr 01 01:05:28 <builder_> build #126 of ar7 is complete: Failure [failed compile_2]  Build details are at http://buildbot.openwrt.org:8010/builders/ar7/builds/126
Apr 01 01:09:04 <builder_> build #125 of sibyte is complete: Failure [failed compile_2]  Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/125
Apr 01 01:14:42 <builder_> build #156 of brcm63xx is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org:8010/builders/brcm63xx/builds/156
Apr 01 02:03:56 <builder_> build #124 of avr32 is complete: Failure [failed compile_4]  Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/124
Apr 01 02:15:55 <CIA-54> thepeople * r31153 /trunk/package/kernel/modules/fs.mk: Disable xfs on avr32, it has been failing the build for at least a month
**** ENDING LOGGING AT Sun Apr 01 02:59:58 2012