From: Takashi Yano <takashi.yano@nifty.ne.jp>
To: cygwin@cygwin.com
Subject: Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
Date: Tue, 11 Aug 2020 13:59:48 +0900 [thread overview]
Message-ID: <20200811135948.959c10de83b8ccfa35f4281a@nifty.ne.jp> (raw)
In-Reply-To: <3d594543-c992-a22c-4631-f786cf34f76b@towo.net>
[-- Attachment #1: Type: text/plain, Size: 3034 bytes --]
Hi Thomas,
On Thu, 6 Aug 2020 15:31:24 +0200
Thomas Wolff wrote:
> Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
> > Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
> >> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
> >> <towo@towo.net> wrote:
> >>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
> >>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
> >>>>> Hi,
> >>>>>
> >>>>> I generally kick off multiple (10) mintty sessions, and place them
> >>>>> around the screen.
> >>>>>
> >>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
> >>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>>> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
> >>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>>> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
> >>>>>
> >>>>> Within the last 6 months or so, about 2-3 of them would hang and
> >>>>> either mintty/tcsh would not start. I put a "sleep 1" in between each
> >>>>> invocation and that seemed to take care of it.
> >>>>>
> >>>>> With the latest cygwin update, about 8 of them just hang even with
> >>>>> the sleep 1. I put in a "sleep 2", and now everything is coming up
> >>>>> again.
> >>>>>
> >>>>> Not sure if this is a mintty or tcsh issue, but just wondering if
> >>>>> others are seeing this before I start trying to debug this.
> >>>> I can reproduce such behaviour with /bin/bash (easy cross-check), and
> >>>> in fact the shell is running in that case (easy test via `echo >
> >>>> .log`), so I have a vague and unpleasant suspicion it might in fact be
> >>>> related to mintty although I have no idea how that would happen. To be
> >>>> analysed.
> >>> When in this state, on the pty from which mintty receives child process
> >>> output, select() does not report a ready for reading condition;
> >>> could it
> >>> be related to the recent poll/select patch? I could not reproduce it in
> >>> cygwin 3.0.7.
> >> for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ; do
> >> mintty -i /Cygwin-Terminal.ico --position @3 --position 120,$i
> >> --size 80x24 /bin/bash & done
> >>
> >> This does not fail for me. Nor does it fail from cmd or powershell
> >> using a script.
> >>
> >> I tried 3.1.4 and 3.1.6 on Windows 10. I tried up to 100 instances
> >> of of mintty...
> > Thanks for testing. It is in fact hard to reproduce, maybe also
> > depending on system load (speculating).
> > I could reproduce one case of one of three terminals being
> > unresponsive also with xterm.
> And it also happens if I drop select() from mintty (and use just
> non-blocking read()).
> > Does anybody familiar with pty/select or recent changes have any idea?
I looked into this problem. After much struggle, I think
I have found a workaround for this issue.
I am not sure why this solves the issue at all, however,
this works for me.
Could you please test a patch attached?
--
Takashi Yano <takashi.yano@nifty.ne.jp>
[-- Attachment #2: 0001-Cygwin-pty-Add-a-workaround-for-issue-of-starting-a-.patch --]
[-- Type: application/octet-stream, Size: 2139 bytes --]
From cc451648bbc7f395fa00cb2b97351d1513827f82 Mon Sep 17 00:00:00 2001
From: Takashi Yano <takashi.yano@nifty.ne.jp>
Date: Tue, 11 Aug 2020 13:16:52 +0900
Subject: [PATCH] Cygwin: pty: Add a workaround for issue of starting a lot of
mintty.
- If a lot of mintty are started in a short time from a mintty, some
of them hang with empty screen, crash immediately or hang on exiting
mintty. The following report seems to be related to this issue.
https://cygwin.com/pipermail/cygwin/2020-August/245751.html
The cause is not clear at all, but this patch seems to solve the
issue.
---
winsup/cygwin/spawn.cc | 15 +++++++++++----
1 file changed, 11 insertions(+), 4 deletions(-)
diff --git a/winsup/cygwin/spawn.cc b/winsup/cygwin/spawn.cc
index 840ec4a86..e70ceb86d 100644
--- a/winsup/cygwin/spawn.cc
+++ b/winsup/cygwin/spawn.cc
@@ -177,7 +177,7 @@ find_exec (const char *name, path_conv& buf, const char *search,
/* Utility for child_info_spawn::worker. */
static HANDLE
-handle (int fd, bool writing)
+handle (int fd, bool writing, bool iscygwin)
{
HANDLE h;
cygheap_fdget cfd (fd);
@@ -188,6 +188,11 @@ handle (int fd, bool writing)
h = INVALID_HANDLE_VALUE;
else if (!writing)
h = cfd->get_handle ();
+ else if (cfd->get_major () == DEV_PTYS_MAJOR && iscygwin)
+ {
+ fhandler_pty_slave *ptys = (fhandler_pty_slave *)(fhandler_base *) cfd;
+ h = ptys->get_output_handle_cyg ();
+ }
else
h = cfd->get_output_handle ();
@@ -625,9 +630,11 @@ child_info_spawn::worker (const char *prog_arg, const char *const *argv,
/* Set up needed handles for stdio */
si.dwFlags = STARTF_USESTDHANDLES;
- si.hStdInput = handle ((in__stdin < 0 ? 0 : in__stdin), false);
- si.hStdOutput = handle ((in__stdout < 0 ? 1 : in__stdout), true);
- si.hStdError = handle (2, true);
+ si.hStdInput = handle ((in__stdin < 0 ? 0 : in__stdin), false,
+ iscygwin ());
+ si.hStdOutput = handle ((in__stdout < 0 ? 1 : in__stdout), true,
+ iscygwin ());
+ si.hStdError = handle (2, true, iscygwin ());
si.cb = sizeof (si);
--
2.27.0
next prev parent reply other threads:[~2020-08-11 5:00 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-08-03 22:13 Brad Wetmore
2020-08-04 10:02 ` Thomas Wolff
2020-08-05 22:56 ` Thomas Wolff
2020-08-05 23:23 ` Kevin Schnitzius
2020-08-06 11:46 ` Thomas Wolff
2020-08-06 13:31 ` Thomas Wolff
2020-08-06 14:26 ` Corinna Vinschen
2020-08-06 14:33 ` Thomas Wolff
2020-08-07 15:34 ` Takashi Yano
2020-08-11 4:59 ` Takashi Yano [this message]
2020-08-11 9:37 ` Corinna Vinschen
2020-08-11 10:02 ` Takashi Yano
2020-08-11 10:30 ` Corinna Vinschen
2020-08-11 17:37 ` Thomas Wolff
2020-08-12 20:27 ` Thomas Wolff
2020-08-06 18:07 ` Brad Wetmore
2020-08-06 18:48 ` Thomas Wolff
2020-08-06 19:25 ` Brad Wetmore
2020-08-13 3:08 ` Brad Wetmore
2020-08-25 18:45 ` Brad Wetmore
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200811135948.959c10de83b8ccfa35f4281a@nifty.ne.jp \
--to=takashi.yano@nifty.ne.jp \
--cc=cygwin@cygwin.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).