From: Takashi Yano <takashi.yano@nifty.ne.jp>
To: cygwin@cygwin.com
Cc: Alexey Izbyshev <izbyshev@ispras.ru>
Subject: Re: Deadlock when pressing Ctrl-C at startup of a native console application
Date: Mon, 18 Apr 2022 21:48:23 +0900 [thread overview]
Message-ID: <20220418214823.a01ed743ceb15d3708ee3fd9@nifty.ne.jp> (raw)
In-Reply-To: <20220416181650.cdb31ec9c75189724bf2f498@nifty.ne.jp>
On Sat, 16 Apr 2022 18:16:50 +0900
Takashi Yano wrote:
> On Fri, 15 Apr 2022 21:09:05 +0300
> Alexey Izbyshev wrote:
> > Hi,
> >
> > I've discovered that pressing Ctrl-C in a Cygwin ssh session running the
> > following command can cause a deadlock:
> >
> > $ while python -c ''; do :; done
> >
> > The exact native application doesn't matter here, e.g. "ping -n 1
> > localhost >/dev/null" instead of "python -c ''" also "works".
> >
> > The deadlock is most easily triggered when CPU is fully loaded. It takes
> > no more than 20 tries on my machine to hit it (and when I first saw it,
> > it was on the first try).
> >
> > When the deadlock is triggered, there are conhost.exe and
> > cygwin-console-helper.exe processes, but there is no python.
> >
> > The exact process tree varies depending on Cygwin version. With the
> > vanilla 3.3.4 (the root is bash spawned by sshd):
> >
> > bash---bash-+-conhost.exe
> > `-cygwin-console-helper.exe
> >
> > With 20220301 snapshot
> > (https://cygwin.com/snapshots/x86/cygwin1-20220301.dll.xz), I get one of
> > the two following trees with an extra conhost.exe process:
> >
> > bash---bash-+-conhost.exe
> > |-conhost.exe
> > `-cygwin-console-helper.exe
> >
> > or
> >
> > bash---bash-+-conhost.exe
> > | `-cygwin-console-helper.exe
> > `-conhost.exe
> >
> > I was testing with 32-bit Cygwin on x64 Windows 10 21H2.
>
> Thanks for the report. I could reproduce the issue.
> I also found the current git master does not have this issue,
> while cygwin-3_3-branch does.
>
> Now I am identifying which patch solved the issue.
> One is obviously:
> https://cygwin.com/pipermail/cygwin-patches/2022q2/011870.html
> however, this was applied also for cygwin-3_3-branch. Therefore,
> another patch also should affect this issue.
I identified the cause and applied a workaround patch.
The countermeasure version is available at:
https://tyan0.yr32.net/cygwin/x86/test/cygwin1-20220418.dll.xz
https://tyan0.yr32.net/cygwin/x86_64/test/cygwin1-20220418.dll.xz
Could you please test?
--
Takashi Yano <takashi.yano@nifty.ne.jp>
next prev parent reply other threads:[~2022-04-18 12:49 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-15 18:09 Alexey Izbyshev
2022-04-16 9:16 ` Takashi Yano
2022-04-18 12:48 ` Takashi Yano [this message]
2022-04-18 15:07 ` Alexey Izbyshev
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=20220418214823.a01ed743ceb15d3708ee3fd9@nifty.ne.jp \
--to=takashi.yano@nifty.ne.jp \
--cc=cygwin@cygwin.com \
--cc=izbyshev@ispras.ru \
/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).