On Wed, Jan 24 2018, Nguyễn Thái Ngọc Duy jotted:
> A follow-up to the recently fixed bugs in the untracked
> invalidation. If opendir() fails it should show a warning, perhaps
> this should die, but if this ever happens the error is probably
> recoverable for the user, and dying would just make things worse.
>
> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@xxxxxxxxx>
> Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx>
> ---
> dir.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/dir.c b/dir.c
> index 44b4dd2ec8..55736d3e2a 100644
> --- a/dir.c
> +++ b/dir.c
> @@ -1787,11 +1787,16 @@ static int open_cached_dir(struct cached_dir *cdir,
> struct strbuf *path,
> int check_only)
> {
> + const char *c_path;
> +
> memset(cdir, 0, sizeof(*cdir));
> cdir->untracked = untracked;
> if (valid_cached_dir(dir, untracked, istate, path, check_only))
> return 0;
> - cdir->fdir = opendir(path->len ? path->buf : ".");
> + c_path = path->len ? path->buf : ".";
> + cdir->fdir = opendir(c_path);
> + if (!cdir->fdir)
> + warning_errno(_("could not open directory '%s'"), c_path);
> if (dir->untracked) {
> invalidate_directory(dir->untracked, untracked);
> dir->untracked->dir_opened++;
I haven't found the root cause yet, but we should not release a 2.17
with this patch.
I tried deploying a 2.16.1 + various patches (including this) internally
today, and on a test set of 236 machines with existing checkouts
(already using untracked cache) 79 would spew "warning: could not open
directory" warnings, warning about a lot of directories that did not
exist at the currently checked-out commit.
The warnings go away on a one-off:
git -c core.untrackedCache=false status
So there's some issue where an existing index with stale UC will be used
by a newer version of git, and will start very verbosely warning about
every directory referenced that can't be found anymore.
