On Wed, Aug 28, 2019 at 10:50 AM Thomas Gleixner <t...@linutronix.de> wrote: > > Woody, > > On Wed, 28 Aug 2019, Woody Suwalski wrote: > > > I have tried to "bisect" the config changes, and builds working/not > > working between > > rc3-rc4-rc5, and come out with the same frustrating result, that > > building a "clean" kernel is not producing the same behavoir as > > incremental building while bisecting. > > So what you say is that: > > make clean; make menuconfig (change some option); make > > and > > make menuconfig (change some option); make > > produces different results? > > That needs to be fixed first. If you can't trust your build system then you > cannot trust any result it produces. > > What's you actual build procedure? > The build procedure: a "clean" one - I have a script/make to build a .deb by untar the src, patch, copy .config, run make oldconfig, run make, package.
The bisect and config-change procedures were simpler - I was running "git bisect bad" and make (followed by make modules_install, copy bzImage to boot, rebuild initramfs) and reboot. For config changes - drop new config, hand-merged in steps toward the presumed "good" one, and run make, and install and reboot... So I was not running explicitly make oldconfig every time, however I believe that config has been updated to match other options selected/unselected by make itself (so I have assumed that make oldconfig has been automagically run sometime during the build). But for the bisect procedure I did not run "make clean" at every step, again - in my former bisections it was not needed, and actually saves a lot of compilation time toward the end of bisection... As such I could not directly answer your question - however yes - building "cleanly" from source seems to produce different results than doing it incrementaly... Thanks, Woody
