In the case of git init echo exit >.git/hooks/post-checkout chmod +x .git/hooks/post-checkout touch foo git add foo rm foo git checkout -- foo git-checkout resulted in a Segmentation fault, because there is no new branch set for the post-checkout hook. This patch makes use of the null SHA as it is set for the old branch. While at it, I removed the xstrdup() around the sha1_to_hex(...) calls in builtin-checkout.c/post_checkout_hook() because sha1_to_hex() uses four buffers for the hex-dumped SHA and we only need two. (Duplicating one buffer is only needed if we need more than four.) Signed-off-by: Stephan Beyer <s-beyer@xxxxxxx> --- I checked if all run_hook()-like functions in the code are addressed with my patch and found that the one in builtin-checkout.c isn't. Then I stumbled over this rare-case bug. builtin-checkout.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/builtin-checkout.c b/builtin-checkout.c index b5dd9c0..149343e 100644 --- a/builtin-checkout.c +++ b/builtin-checkout.c @@ -47,8 +47,10 @@ static int post_checkout_hook(struct commit *old, struct commit *new, memset(&proc, 0, sizeof(proc)); argv[0] = name; - argv[1] = xstrdup(sha1_to_hex(old ? old->object.sha1 : null_sha1)); - argv[2] = xstrdup(sha1_to_hex(new->object.sha1)); + argv[1] = sha1_to_hex(old ? old->object.sha1 : null_sha1); + argv[2] = sha1_to_hex(new ? new->object.sha1 : null_sha1); + /* "new" can be NULL when checking out from the index before + a commit exists. */ argv[3] = changed ? "1" : "0"; argv[4] = NULL; proc.argv = argv; -- 1.6.1.160.gecdb -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
