On Mon, Mar 13, 2017 at 1:12 PM, Marc Stevens <marc.stevens@xxxxxx> wrote:
> Indeed, I've committed a fix, and a small bug fix for the new code just now.
Unrelated side note: there may be some missing dependencies in the
build infrastructure or something, because when I tried Jeff's script
that did that "test the merge and the two parents", and used the
pack-file of the kernel for testing, I got:
5611971c610143e6d38bbdca463f4c9f79a056a0
/home/torvalds/v2.6/linux/.git/objects/pack/pack-153bb8cd11846cf9a27ef7b1069aa9cb9f5b724f.pack
real 0m2.432s
user 0m2.348s
sys 0m0.084s
5611971c610143e6d38bbdca463f4c9f79a056a0
/home/torvalds/v2.6/linux/.git/objects/pack/pack-153bb8cd11846cf9a27ef7b1069aa9cb9f5b724f.pack
real 0m3.747s
user 0m3.672s
sys 0m0.076s
5611971c610143e6d38bbdca463f4c9f79a056a0 *coll*
/home/torvalds/v2.6/linux/.git/objects/pack/pack-153bb8cd11846cf9a27ef7b1069aa9cb9f5b724f.pack
real 0m5.061s
user 0m4.984s
sys 0m0.077s
never mind the performace, notice the *coll* in that last case.
But doing a "git clean -dqfx; make -j8" and re-testing the same tree,
the issue is gone.
I suspect some dependency on a header file is broken, causing some
object file to not be properly re-built, which in turn then
incorrectly causes the 'ctx2.found_collision' test to test the wrong
bit or something.
Linus