https://bugs.exim.org/show_bug.cgi?id=2784
--- Comment #10 from Jeremy Harris <jgh146exb@???> ---
I don't recall either why the "wb" was needed, and that comment is
insufficient.
Mea culpa. But making it "rb" always breaks the ANSWER-11 subtest of
testsuite case 0373 (on Fedora 33).
Your results are interesting. The FIN packet is *expected* to be emitted
due to the closedown(SHUT_WR) preceding that fdopen(), but only if the
do_shutdown flag is set. Which is when the fdopen() should be using "rb"
anyway. You're seeing it in both cases... maybe it's not due to the
shutdown() but to the close() supposedly being done after the cat_file(),
and the latter is behaving differently in some fashion and giving a timing
difference. Do both pcaps have FINs for both directions, or is one getting
a RST? In fact, I need to run up a FreeBSD and play - what version?
--
You are receiving this mail because:
You are on the CC list for the bug.
