On 23.04.2009 1:35 Uhr, Carl-Daniel Hailfinger wrote:
On 23.04.2009 01:28, Stefan Reinauer wrote:
On 23.04.2009 1:09 Uhr, Joseph Smith wrote:
On Wed, 22 Apr 2009 13:50:54 -0700, ron minnich rminnich@gmail.com wrote:
On Tue, Apr 21, 2009 at 4:41 AM, Joseph Smith joe@settoplinux.org
wrote:
This really sounds like an unitialized variable or a bad pointer.
Wouldn't gcc catch these and error when building?
not always. that's what sparse can be useful for.
Why can't I just send the build output to a text file and then look through it for all the warnings/errors?
$ make > buildlog.txt
You can..
btw.. qa.coreboot.org holds logs of each revision too (the last few 200-1000 depending when I clean up)
http://qa.coreboot.org/log_buildbrd.php?revision=4187&device=ip1000&...
But please note that gcc does not catch all the tricky mistakes we make.
Or you run cd util/abuild; abuild -t thomson/ip1000 and look at coreboot-builds/thomson_ip1000/make.log
or
util/abuild/abuild -t thomson/ip1000 $PWD
to build on the top level.
You can also specify a payload directory with -p
mkdir payloads cp /somewhere/filo.elf payloads then add a file payloads/payload.sh which prints the name of the payload to use (and takes the mainboard as a parameter) such as echo "`dirname $0`/build/filo.elf"
then you can build a fully working image with payload by specifying:
util/abuild/abuild -t thomson/ip1000 -p ./payloads $PWD
Specify -c max or -c 2 if you have an SMP system, to speed compilation up.
Stefan