This is the mail archive of the cygwin mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: "emacs -nw" hangs in a terminal

On 5/22/2012 7:28 AM, Corinna Vinschen wrote:
On May 21 14:51, Ken Brown wrote:
On 5/21/2012 12:29 PM, Corinna Vinschen wrote:
On May 21 11:31, Ken Brown wrote:
On 5/21/2012 6:02 AM, Ken Brown wrote:
On 5/21/2012 4:50 AM, Filipp Gunbin wrote:
emacs-24.0.96-2 crashes when I am doing the following:

1) emacs -Q -nw
2) M-x shell
3) C-x C-f C-g

I can reproduce this. I'll try again to fix it.

I've discovered something strange by running emacs under gdb. If I start emacs-24 in a terminal (but not under X) and start a shell as you did, then every press of C-g creates a new thread, and these are never destroyed. I'm pretty sure the threads are created by Cygwin, not by emacs.

What does C-g mean in Emacs? What's it supposed to do? Does it call select or poll?

It's supposed to quit whatever operation is in progress. It doesn't call select or poll. In the situation of Filipp's instructions above, C-x C-f has caused emacs to prompt for a file name, and C-g should interrupt that. It also rings the the terminal bell and prints "Quit" in the echo area at the bottom of the screen.

The situation in my instructions is slightly different.  Prior to
the user pressing C-g, emacs is running its idle loop, in which it
repeatedly calls select to see if there's any event it needs to
respond  to.  When C-g is pressed, select returns and emacs reacts
to the keypress.  In this case there's nothing to do but ring the
terminal bell and print "Quit".

Somehow I'm not able to test this. When I start `emacs -Q -nw' in cmd or mintty, emacs takes 100% CPU for some reason. It doesn't matter if I try it under Cygwin 1.7.15 or current CVS.

That's strange. All my tests were done in mintty. Would it help if I sent some strace output and/or a gdb backtrace? I assumed you could get those (or at least the strace output) yourself, and I didn't want to spam the list.


P.S. BTW, I tested today's snapshot, and the problem is still there.

-- Problem reports: FAQ: Documentation: Unsubscribe info:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]