Right now, we don't close the read end of the pipe when git-upload-pack runs git-pack-object, so we hang forever (why don't we get SIGALRM?) instead of dying with SIGPIPE if the latter dies, which seems to be the norm if the client disconnects. Thanks to Johannes Schindelin <Johannes.Schindelin@xxxxxx> for pointing out where this close() needed to go. This patch has been tested on kernel.org for several weeks and appear to resolve the problem of git-upload-pack processes hanging around forever. Signed-off-by: H. Peter Anvin <hpa@xxxxxxxxx> --- upload-pack.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/upload-pack.c b/upload-pack.c index 498bf50..d3a09e7 100644 --- a/upload-pack.c +++ b/upload-pack.c @@ -119,6 +119,7 @@ static void create_pack_file(void) int i; struct rev_info revs; + close(lp_pipe[0]); pack_pipe = fdopen(lp_pipe[1], "w"); if (create_full_pack) -- 1.5.0.3 - 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