diff options
| author | Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org> | 2022-02-17 10:49:20 +0000 |
|---|---|---|
| committer | Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org> | 2022-02-17 10:49:53 +0000 |
| commit | 83bb6ae11ac7cf46b7f5b36892945b98156eb3c6 (patch) | |
| tree | a87c42befebaaa373f5ce3a6608e75c70af7bbb8 | |
| parent | 06d280dc7aedc914dbc003dcfca62f8d4b3cb05d (diff) | |
round-robin-bisect.sh: Don't use interesting_commits in linux-next bisects
Hopefully, this will get us out of an infinite bisect loop we are
currently seeing.
Change-Id: Icb3aac078d8f151ec8c74e59c9ce24accbd88fa3
| -rwxr-xr-x | round-robin-bisect.sh | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/round-robin-bisect.sh b/round-robin-bisect.sh index 78979b11..194978f5 100755 --- a/round-robin-bisect.sh +++ b/round-robin-bisect.sh @@ -420,8 +420,16 @@ cp $commits_to_test $commits_in_range print_tested_revs >> $commits_in_range # This loop can generate lots of console noise. +# +# Bisecting linux-next.git regressions is difficult enough due to how +# the tree is constructed, so we prefer to not use interesting_commits +# when $rebase_workaround is true. This makes linux-next bisects as +# natural as they can be. +# Hopefully, this will get us out of an infinite bisect loop we are +# currently seeing. set +x -while [ x"$(get_first_bad </dev/null)" = x"" ] && read -a arr; do +while [ x"$(get_first_bad </dev/null)" = x"" ] && read -a arr \ + && ! $rebase_workaround; do ( set -euf -o pipefail |