Re: public-inbox-convert hangs on systems using musl libc

[Eric Wong <e@80x24.org>]

Chris Brannon <chris@the-brannons.com> wrote:
> I can reproduce it reliably on both Alpine Linux and the musl variant of
> Void Linux.  Just clone a mirror of public-inbox.org/meta and try and
> convert it:
> 
> $ git clone --mirror https://public-inbox.org/meta
> $ public-inbox-convert meta metanew
> 
> If I pass --no-index, the process is successful.
> 
> This is a deadlock of some sort, because during the hang, git cat-file
> is blocked on write and other things are blocked on read.

I've requested some packages w/ cfarm admins to test with.

Do you know which pipes are which?  "lsof -p $PID +E" can help
with connectivity checking, as can script/dtas-graph in
https://80x24.org/dtas.git if you have Graph::Easy

What's curious is all the code paths should be independent of
stdio buffering, so I wouldn't think libc differences would
matter...

I also just did the above meta conversion on FreeBSD w/o problems.

> That's as far as I have gotten with debugging it.

Some shots in the dark:

1. force read pipe on our end to be non-blocking

diff --git a/lib/PublicInbox/Git.pm b/lib/PublicInbox/Git.pm
index 882a9a4a..b239ceb9 100644
--- a/lib/PublicInbox/Git.pm
+++ b/lib/PublicInbox/Git.pm
@@ -143,6 +143,7 @@ sub _bidi_pipe {
 		fcntl($out_w, 1031, 4096);
 		fcntl($in_r, 1031, 4096) if $batch eq '--batch-check';
 	}
+	$in_r->blocking(0);
 	$self->{$out} = $out_w;
 	$self->{$in} = $in_r;
 }

2. Tweak $PIPE_BUFSIZ and/or MAX_INFLIGHT to smaller values.  e.g.

diff --git a/lib/PublicInbox/Git.pm b/lib/PublicInbox/Git.pm
index 882a9a4a..ec40edd7 100644
--- a/lib/PublicInbox/Git.pm
+++ b/lib/PublicInbox/Git.pm
@@ -23,13 +23,12 @@ use Carp qw(croak carp);
 use Digest::SHA ();
 use PublicInbox::DS qw(dwaitpid);
 our @EXPORT_OK = qw(git_unquote git_quote);
-our $PIPE_BUFSIZ = 65536; # Linux default
+our $PIPE_BUFSIZ = 4096; # Linux default
 our $in_cleanup;
 our $RDTIMEO = 60_000; # milliseconds
 our $async_warn; # true in read-only daemons
 
-use constant MAX_INFLIGHT => (POSIX::PIPE_BUF * 3) /
-	65; # SHA-256 hex size + "\n" in preparation for git using non-SHA1
+use constant MAX_INFLIGHT => 4;
 
 my %GIT_ESC = (
 	a => "\a",


MAX_INFLIGHT could go down to 1 outside of t/git.t, I think...
But my sleep deprived mind isn't finding anything that jumps out...