Follow-up Comment #5, bug #62622 (project findutils):
[comment #4 comment #4:] > Hey all, > > I just wanted to show you the conclusion of my investigation of this bug. > > > > (file #53451) So here in the attached picture, you can see the variables `state->largest_successful_arg_count, state->smallest_failed_arg_count, limit` that I recorded from a gdb session for `update_limit()` when using the reproducer yes | head -n 2000000 > myfile gdb --args xargs/xargs -a myfile -s $limit sh -c 'echo hello world' The green and red dots show the success/fail bounds and the blue dot is the attempt by bc_do_exec(). Every time the blue dot is above the red dot, it goes down because execvp() fails. It goes down by half the difference between the red and green dot. When it ends up below the red dot, it updates the green dot closer; thus the next time (because we start at the top again), the blue dot takes even longer to reach below the red dot. I'm continuing to work on this trying to find a way to fix it. _______________________________________________________ Reply to this item at: <https://savannah.gnu.org/bugs/?62622> _______________________________________________ Message sent via Savannah https://savannah.gnu.org/