**** BEGIN LOGGING AT Thu Jun 28 02:59:57 2007 Jun 28 05:51:59 nbd * r7748 /trunk/ (20 files in 20 dirs): fill in the kernel version info for every target, move the kernel tarball md5sums to a central place Jun 28 06:53:06 nbd * r7749 /trunk/toolchain/kernel-headers/Makefile: disable quilt for kernel-headers Jun 28 07:14:19 nbd * r7750 /trunk/include/quilt.mk: allow packages to override the patch directory using the PATCH_DIR variable Jun 28 07:20:00 nbd * r7751 /trunk/toolchain/Makefile: enable the refresh target in toolchain/ Jun 28 07:20:43 nbd * r7752 /trunk/toolchain/binutils/ (15 files in 4 dirs): remove the custom Build/Prepare override in binutils, use PATCH_DIR, refresh patches using quilt Jun 28 07:30:24 nbd * r7753 /trunk/toolchain/gcc/ (10 files in 3 dirs): replace gcc 4.2 snapshot with 4.2.0 release (sync patches with uclibc buildroot), adjust gcc makefile for PATCH_DIR changes Jun 28 08:45:58 nbd * r7754 /trunk/toolchain/Makefile: add update target to toolchain/ Jun 28 09:08:47 nbd * r7755 /trunk/target/linux/brcm47xx-2.6/patches-2.6.22/170-cpu_wait.patch: disable cpu wait support for bcm4710 (suggested in #1641) Jun 28 10:15:19 nbd * r7756 /trunk/target/linux/brcm47xx-2.6/patches-2.6.22/240-extif_fixes.patch: add extif fixes (#1984) Jun 28 12:56:54 florian * r7757 /trunk/package/iptables/files/firewall.init: Initialise firewall before network (#1988) Jun 28 16:08:28 juhosg * r7758 /trunk/target/linux/ (17 files in 6 dirs): [adm5120] update to 2.6.22-rc6 Jun 28 19:07:35 thepeople * r7759 /packages/net/miniupnpd/patches/101-log_read-fix.patch: fix compiling error Jun 28 19:14:57 thepeople * r7760 /packages/net/miniupnpd/ (Makefile patches/001-makefile-tweaks.patch): update to RC5 Jun 28 19:52:07 nbd * r7761 /trunk/toolchain/gcc/ (Makefile patches/4.2.0/307-long_double_fix.patch): fix gcc 4.2.0 compile for ppc Jun 28 20:21:26 Kaloz, actaully its better here ,, might some question about the adi pronghorn metro and cf ide ? Jun 28 20:42:20 nbd * r7762 /trunk/ (7 files in 4 dirs): update magicbox target to 2.6.22-rc6, select gcc 4.2.0 as default for this target (4.1.2 segfaults at compile time) Jun 28 20:54:39 http://pastebin.ca/594256 error while building 4.2.0 gcc for brcm47xx 2.6 Jun 28 20:57:53 crazy_imp: try newer binutils Jun 28 20:58:46 ok Jun 28 21:00:40 nbd: so binutils 2.16.1 can be marked as broken with gcc 4.2 or gcc should select binutil 2.17 itself (?) Jun 28 21:01:00 first i want to know if it's fixed with 2.17 Jun 28 21:01:19 sure Jun 28 21:03:25 * crazy_imp waits :) Jun 28 21:04:45 thepeople * r7763 /packages/net/ntpclient/files/ntpclient.init: add ntpclient count option and exit after a successful sync closes #1835 Thanks Carlos Sobrinho Jun 28 21:09:42 nbd * r7764 /trunk/package/openssl/patches/180-gcc_breakage.patch: Jun 28 21:09:42 Add a patch for openssl to fix some breakage that happens when compiled with gcc 4.2.0 Jun 28 21:09:42 Note: this does not cover all instances, but it will at least compile. Jun 28 21:09:42 If the broken parts are reached at run time, the program will crash. Jun 28 21:09:42 I wonder whose stupid idea it was to intentionally(!) make programs segfault when they hit something that *might* be unsafe to use. Jun 28 21:09:46 Patch taken from http://www.mail-archive.com/openssl-dev@openssl.org/msg22667.html Jun 28 21:12:09 <[mbm]> that's one amusing thread on the openssl list Jun 28 21:12:35 yeah, reading it right now, wth are they thinking. Jun 28 21:13:37 thepeople * r7765 /packages/net/ (4 files in 4 dirs): reorder hotplug files fixes #1527 Jun 28 21:15:30 <[mbm]> Bartman007: for more bsd related madness - http://marc.info/?l=openbsd-misc&m=118296441702631 Jun 28 21:16:34 yeah, :-\ (running on one of the new Meroms) Jun 28 21:17:18 don't know if it is actually covered though... Jun 28 21:17:18 <[mbm]> I've been buying amd64's the last few rounds Jun 28 21:17:40 <[mbm]> kinda glad I didn't get the last duo I saw on special Jun 28 21:18:15 [mbm]: I've been buying AMD on the desktop/server side for a long time, but machines that fit my laptop needs tend to reside in the Intel camp Jun 28 21:18:59 <[mbm]> even my laptop is an amd64x2 .. probably doesn't do wonders for the battery life Jun 28 21:20:45 <[mbm]> http://hardware.slashdot.org/comments.pl?sid=242663&cid=19677773 .. wtf? Jun 28 21:22:21 huh? so we'll be able to get xen on arm routers too? Jun 28 21:22:51 [mbm]: sounds like someone found a secret stash went bad a long time ago. Jun 28 21:24:18 <[mbm]> I know there's the harddrives on the market which have an autorun and wedge themselves into providing a "virtual" desktop that you carry with you Jun 28 21:24:27 <[mbm]> but that doesn't quite fit the description given Jun 28 21:25:03 [mbm]: Xen allows live migration of domU's from one host to another, but obviously it would require all hosts to be the same architecture. Jun 28 21:25:26 <[mbm]> or a damn good qemu layer Jun 28 21:25:54 <[mbm]> suppose you could always suspend the vm and save the state to a removable medium Jun 28 21:26:00 <[mbm]> which is a little more practical Jun 28 21:26:16 I don't see ARM making headway into the desktop market anytime soon, so what that guy describes is going to be limited to those that can afford really expensive equipment Jun 28 21:26:53 [mbm]: yeah, that feature is actually fairly handy when you want to reboot the dom0. Jun 28 21:27:06 <[mbm]> yeah, the way I see it we're stuck with legacy x86 for the foreseeable future Jun 28 21:27:51 <[mbm]> and I can't say I'm all that fond of the x86 architecture, although it's amazing how people have figured out how to advance it without completely breaking it Jun 28 21:28:07 <[mbm]> the first amd 64 stuff was an interesting hack Jun 28 21:28:21 [mbm]: I consider many of the advancements to be broken hacks :-P Jun 28 21:28:32 <[mbm]> likewise Jun 28 21:30:07 yeah, x86_64 was the first one that really interested me, it's as close to jumping to a new arch/starting from scratch as we are going to get Jun 28 21:31:39 to some degree, likewise with EFI, unless more companies contribute/open up to LinuxBIOS (or similar) Jun 28 21:31:52 <[mbm]> yeah, was just about to mention efi Jun 28 21:32:00 <[mbm]> I like the concept of replacing bios Jun 28 21:32:20 <[mbm]> but it has some implications about how much control efi can assert over the os Jun 28 21:32:24 <[mbm]> and who controlls efi Jun 28 21:32:55 <[mbm]> but the bootup time of your average pc is just pathetic as most computers spend atleast 10 seconds just in POST Jun 28 21:33:07 yeah, I haven't looked into implementation the Centrino Pro functionality, but from what I understand it pretty much runs as a hypervisor to the OS Jun 28 21:33:10 <[mbm]> so I can't wait for bios to get overhauled Jun 28 21:33:47 [mbm]: and then you've got all the other hardware with their BIOS hooks to display their own stuff. Jun 28 21:37:32 nbd: it's building kernel modules right now, so i think it's working with 2.17 :) Jun 28 21:39:46 <[mbm]> Bartman007: something besides the vendor logo splash screen at post? Jun 28 21:41:34 [mbm]: some videocards display a quick BIOS screen initially, fakeraid cards love displaying something directly after the mobo screens, I think I've seen some other stuff do it too. Jun 28 21:42:17 some nic's do that aswell (until you turn it off) Jun 28 21:42:31 crazy_imp: ah, yes, the netboot hooks Jun 28 21:42:44 press F12 for PXE boot Jun 28 21:43:17 <[mbm]> Bartman007: yep, got a few ide controllers that like to spend 5 seconds just spinning the cursor and probing the drives Jun 28 21:43:19 the e100's were especially long if you didn't kill it. Jun 28 21:44:06 I have an Adaptec SCSI card that takes forever to spin up the drives before moving on to boot Jun 28 21:44:33 i have some realtek cards which offer you press some strange combo to get the menu Jun 28 21:45:04 (and then i can change the way of how they should try to netboot) Jun 28 21:45:49 <[mbm]> I want some efi or linuxbios that as soon as I turn on the machine drops me to a login and runs the startup in the background Jun 28 21:45:50 i also can turn the menu off, but before i do that i would like to know how i can get it back later :D Jun 28 21:46:38 [mbm]: But there are quite a lot of things that have to be in place before even a login will work. Jun 28 21:47:05 <[mbm]> noz: yes, and I'm willing to wait for the video card Jun 28 21:47:12 :) Jun 28 21:47:20 noz: it could fetch the login data and put them later in ;) Jun 28 21:47:43 <[mbm]> but I don't need it probing all 8 drives just for bios when linux is just going to do the same thing again Jun 28 21:47:51 Indeed. Jun 28 21:48:34 <[mbm]> if hibernate/suspend to disk/whatever actually worked worth a damn in linux I'd use it Jun 28 21:48:35 Mostly the BIOS needs to setup core (CPU/mem, possibly video) and boot media Jun 28 21:48:57 and you're right that everything else will get done by linux anyway Jun 28 21:49:22 <[mbm]> as is I tend to just leave my machines running because it's less hassle than waiting for them to boot Jun 28 21:49:59 likewise, plus the vague notion that they need to do housekeeping at 3 in the morning! Jun 28 21:50:31 <[mbm]> atleast one of my laptops hard locks on any attempt to suspend to the point that you need to pull the power cord and yank the batter before you can get it to boot again Jun 28 21:51:45 [mbm]: I got it working reliably on my last laptop, but it took far too long to force the video card to reset (10-30 seconds) quite disappointing. Jun 28 21:52:12 I had a heart-thumping moment in a demo to senior customers several years ago, where the laptop I was demoing on hard crashed. I tried to press the reset button, then ended up popping the battery, all serruptitiously behind my back as I kept talking. Jun 28 21:53:32 <[mbm]> Bartman007: even when suspend works I always have issues like xv not running when it resumes or the soundcard in an undefined state Jun 28 21:53:37 Fundamentally things need to be designed for suspend, and the current expectation of a BIOS initialisation pushes away from that, not towards it Jun 28 21:55:01 <[mbm]> noz: if anything can go wrong it'll go wrong when you attempt to demo it; that's the first rule. second rule is avoid the demo or fake it if at all possible. Jun 28 21:56:49 Oh yes, I've done that too.... I particularly remember a big demo of a large RF comms system (multiple 19" racks) all laid out in 2 stacks on green beize at opposite ends of the room, and the actual comms was a serial cable! Jun 28 21:57:21 (just in case) Jun 28 21:57:45 [mbm]: that soundcard problem is something that appeasr to be fairly prevalent for Ubuntu, I *never* ran into it on Gentoo. Jun 28 21:57:55 appears* Jun 28 21:58:25 <[mbm]> the other common trick is to have several machines, but each at a different step; if you crash one you just move to the next machine and continue on Jun 28 21:58:58 <[mbm]> Bartman007: yeah, ubuntu tries some pretty screwed up things where it likes to unload as many modules as possible Jun 28 21:59:16 [mbm]: Blue Peter style: here's one I prepared earlier. Jun 28 21:59:22 <[mbm]> which causes problems if applitions still hold fds Jun 28 21:59:41 <[mbm]> Bartman007: annoyingly the problem probably wouldn't even exist if they just left the module as-is Jun 28 22:00:05 [mbm]: Jun 28 22:00:09 <[mbm]> noz: yeah it's the style you see on any cooking show Jun 28 22:00:27 Unless hardware starts supporting an explicit "get state", "set state" interface (and that's even possible), suspend is always going to cause problems Jun 28 22:00:39 .. if you let the hardware fully power down Jun 28 22:01:09 <[mbm]> laptop that gives me the most problems is an old dell Jun 28 22:01:18 unfortunately they have to make a solution that works on *everything* (one of the downsides of making "Linux for Humans") and can't tell people to tweak file x with y and z settings. Jun 28 22:01:30 <[mbm]> even doing nothing but sitting on an x11 desktop it can get the temp up to 75C Jun 28 22:01:42 <[mbm]> and it needs both fans on high to lower the temp Jun 28 22:01:54 <[mbm]> rather noisy fans too Jun 28 22:02:23 ouch - wouldn't like that on my lap! Jun 28 22:02:44 <[mbm]> linux doesn't think the laptop supports speedstep Jun 28 22:02:49 <[mbm]> even though I'm pretty sure it does Jun 28 22:03:00 <[mbm]> so I can't downclock it to reduce heat Jun 28 22:03:37 [mbm]: my i8600 (Dell) ran a lot cooler than this macbook pro, especially after I undervolted it. I went from fans going constantly at full load to 10-20 seconds every 3-5 minutes. Jun 28 22:03:42 <[mbm]> and I've dismanted it a few times looking for dust, reapplying thermal paste and reseating the heatsink Jun 28 22:03:52 New Dells (at least some of them) seem to be pretty well supported. (Despite Dell's public face of M$ line-toeing) Jun 28 22:04:50 <[mbm]> temp gets hot it turns the fnas on high Jun 28 22:04:51 aside from the ATI video card, it worked great in linux. The videocard is what caused most of the trouble with suspend/hibernate Jun 28 22:04:58 <[mbm]> 5 sec later it turns them off completely Jun 28 22:05:13 <[mbm]> and another 5-10 sec later the process repeats Jun 28 22:05:30 [mbm]: what model? Jun 28 22:05:37 <[mbm]> inspiron 2650 Jun 28 22:05:47 <[mbm]> has the stupid i8k management interface Jun 28 22:06:08 nbd: finished building the trunk :) Jun 28 22:06:21 :) Jun 28 22:06:23 [mbm]: i8kutils worked fairly well for me in controlling fan noise Jun 28 22:06:52 <[mbm]> Bartman007: i8kutils cna control the fans, problem is that you need both fans on loud to control the heat Jun 28 22:07:34 [mbm]: so you just set the fan to something a bit more sane (going from high to low instead of high to off) Jun 28 22:07:36 <[mbm]> and if you get pissed off about the constant fan noise and turn off the fans it overheats and crashes Jun 28 22:08:20 <[mbm]> Bartman007: Bartman007 it gets pretty much 1C every second until 85C unless both fans are on high Jun 28 22:08:29 <[mbm]> I have no idea what generates that much heat Jun 28 22:08:47 it is a p4. Jun 28 22:08:53 <[mbm]> yep Jun 28 22:08:59 noz: it's booting *happy* :D Jun 28 22:09:26 crazy_imp: offering a router to the gcc gods? Jun 28 22:09:39 Excellent. Guess we'll never know which bit it really was. You've changed quite a bit... Jun 28 22:09:40 <[mbm]> you can throttle the cpu but that seems to do nothing otehr than making the system less responsive Jun 28 22:10:29 noz: indeed, but i'll make a clean build with the old toolchain stuff to see if it's still there Jun 28 22:11:01 OK. If you're prepared to do that sort of regression testing, brilliant - thanks. Jun 28 22:12:39 [mbm]: if you are sure it supports speedstep there are ways to force the it to switch freqs instead of throttling Jun 28 22:13:23 <[mbm]> Bartman007: pretty sure I managed to run the windows speedstep utils back when I first bought it but linux speedstep probe fails at bootup Jun 28 22:13:55 <[mbm]> bios is one of those useless pheonix deals where you can set the clock Jun 28 22:17:23 I really feel like I ran into a similar thing with a P3, (it supported speedstep but cpuinfo, etc didn't think it did) I thought I forced it to work, I just don't remember if it was patching the speedstep driver, throwing it an arguement, or what... Jun 28 22:17:56 <[mbm]> I'll try forcing the driver and seeing what happens Jun 28 22:22:34 hey, farnz you about? Jun 28 23:07:26 mbm * r7766 /trunk/package/base-files/files/lib/network/config.sh: fix broadcast address (#1611) Jun 29 00:07:44 nbd * r7767 /trunk/package/busybox/patches/260-broadcast_plus.patch: check for + in the broadcast address before attempting a dns lookup **** ENDING LOGGING AT Fri Jun 29 02:59:56 2007