looking4bI have noticed that the kernel does some printk very early to print out the banner and version of the kernel. It does this before parsing the kernel command line which tells it which ttyS to use. How does the kernel figure out to which UART/Serial/Console port to send the output of printk at that stage? 01:29
looking4bI am trying to get the kernel working on my embedded systems but I do not get any serial output from the kernel. When I debug it then I noticed that is calling printk but I do not see anything on my console (serial port)01:29
looking4bcan anybody help?01:29
looking4bI am talking of the printks in start_kernel01:44
jjohansenlooking4b: those prints use earlyprintk02:12
jjohansenyou can set earlyprintk bootargs02:12
jjohansenI would need to lookup how it determines which console to use by default02:13
looking4bI think my kernel is using uart0 by default instead of using uart1 which is the one my hardware supports02:15
looking4bI did compiled with a hard coded command line that has dev=ttyS1,11520002:16
looking4bbut I guess the kernel is crashing before it can use those settings02:16
looking4bI do not even know if ttyS1 is really going to use uart102:16
looking4bthere are 3 uarts in my system02:17
looking4bI've been reading the source code but in some places they use dynamic function pointers to do stuff and I get lost then because I do not know to what function the pointer is pointing to02:17
looking4bAnother thing that I would like to know is how to make the System.map give me more details02:18
looking4bright now it tells me a symbol and its address but I also want to know where that symbol is located02:18
looking4bin which file.o the symbol was found02:19
looking4bregularly gcc/ld will spit out a .map file with that information by default but I think System.map is being generating by parsing stuf from the .elf02:19
looking4bI guess at that point finding where the symbol came from cannot be done since all the file.o have been combined02:20
jjohansenlooking4b: right, that info isn't available in standard .elf you need the debug information to do that kind of mapping02:46
jjohansenif you know the address you can always map it back using gdb and the debug info02:47
jjohansenI would have to go lookup how again as I haven't done it for a while02:47
looking4byeah, I do not know why the .map file that gcc/ld spits out is not used as system.map02:49
looking4bit has much more info02:49
looking4bI will have to research how to compile the kernel with debug info on02:50
jjohansenare you using custom builds via make or the kernel build scripts02:51
looking4bthe truth is that I am using uCLinux with 2.6 kernel02:51
looking4bbut I thought it should be the same for any distro02:52
looking4bI guess ubuntu patches the kernel with their own stuff but the basics are the same02:53
jjohansenmake  CONFIG_DEBUG_INFO=102:54
jjohansenthat will turn on -g and -gdwarf-2 for the compiler02:55
looking4bjjohansen: Adding a -Map=linux.map to the following line in the root Makefile did the trick03:44
looking4bcmd_vmlinux__ ?= $(LD) -Map=linux.map $(LDFLAGS) $(LDFLAGS_vmlinux) -o $@ \03:44
looking4bnow I see things like 0x40020000                _start03:46
looking4b.head.text     0x40020000       0xde arch/m68knommu/platform/coldfire/head.o03:46
looking4bgood to know in which file my entry point is03:46
looking4bseems like my kernel does not get compressed03:46
looking4bi think misc.c decompress the kernel03:47
jjohansenlooking4b: nice03:51
theRealSaintdpkg-gencontrol: error: package linux-image-2.6.35-rc5-custom+ not in control info07:00
theRealSaintam getting this error while compiling a vanilla kernel on ubuntu07:00
theRealSaintany pointers?07:00
jjohansentheRealSaint: how are you setting up the vanilla kernel for compile?07:07
theRealSaintgit download07:08
theRealSaintam getting this error while compiling a vanilla kernel on ubuntu07:08
theRealSaintfakeroot make-kpkg --initrd --append-to-version=-custom kernel_image kernel_headers07:08
theRealSaintjjohansen, these are the commands07:09
achiangtheRealSaint: can you do a grep LOCALVERSION .config07:09
theRealSaintauto is not set07:10
jjohansentheRealSaint: hrmm I have never used make-kpkg to do vanilla builds07:11
jjohansenI start with git pull of vanilla and either use plain kernel make, make install, make modules_install, update-initramfs, update-grub07:13
achiangthat's what i said earlier in #u-devel. ;)07:13
achiangjjohansen: you forgot make firmware_install. :)07:13
jjohansenor if I want a .deb, copy in the debian, and debian.master files from an ubuntu kernel07:13
jjohansenah yeah, well you know the basic steps07:14
jjohansenand then I twiddle with configs maybe, and do07:14
jjohansenskipabi=true fakeroot debian/rules binary-generic07:15
jjohansenhrmm well actually I usually need to edit the version strings first, then do07:16
jjohansenfakeroot debian/rules clean07:16
jjohansenfakeroot debain/rules prepare-generic07:16
jjohansencheck config, and typing is right, maybe edit debian/change_log to add ~jj to the dpkg version07:17
jjohansenthe do07:17
jjohansenfakeroot debian/rules binary-generic07:17
jjohansentheRealSaint: https://wiki.ubuntu.com/Kernel/Dev/QuickBuildLocal07:21
jjohansencovers the basics of how we build kernels07:21
jjohansenbasically all I do is transplant the debian bits from an ubuntu kernel into a vanilla kernel, and then do standard kernel build07:22
jjohansenerr standard ubuntu kernel build07:22
jk-apt-get install iprint :)09:06
jdstrandapw: hi! fyi only. I have an i7 running lucid that would not resume from suspend well09:20
jdstrandapw: (usb wouldn't come up)09:20
jdstrandapw: kees and I talked, and though his symptoms were totally different than mine, I decided to blacklist tpm09:20
jdstrandapw: and lo and behold... it works fine09:20
jdstrandapw: kees mentioned you had reports of i7 suspend issues, so I thought I'd pass that along. I'll let you know if anything changes on maverick for me09:21
lifelessapw: jdstrand: I have suspend issues on i7 on mav09:31
lifelesslucid base, maverick kernel, been insanely busy :(09:31
jdstrandlifeless: you might try to do 'lsmod|grep tpm', then rmmod those prior to suspend and see if it works better09:32
lifelessok, will give that a shot09:32
jdstrandlifeless: I've only done it a few times, but it has worked every time since then. prior to that, it never worked right09:33
=== persia` is now known as persia
* abogani waves10:19
aboganiapw, Do you have a chance to review -lowlatency kernel (and -realtime also) before of Alpha-3?10:19
vanhoofsconklin: https://patchwork.kernel.org/patch/108727/10:51
diwicabogani, out of interest, what differs the -lowlatency kernel from the -preempt kernel?10:51
diwicI'm trying to understand the differences between the different attempts get lower latency...10:52
aboganidiwic, https://lists.ubuntu.com/archives/kernel-team/2010-February/008708.html11:27
vanhoofsconklin: updated https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/56180213:07
ubot2Launchpad bug 561802 in linux (Ubuntu Lucid) (and 1 other project) "[lucid] [i915] blank screen on Latitude E6410 (affects: 24) (heat: 171)" [Medium,In progress]13:07
=== luftikuss is now known as luftikus2
nessitahi there! can anyone give me a hand with a potential ACPI issue in my toshiba laptop?13:43
hazmatdoes anyone know any details on the http://events.linuxfoundation.org/linuxcon2010/crawford the KSLM monitoring stuff? most i can find are some irc log messages, and an empty launchpad project 17:53

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!