[rfa/dejagnu] Work around cygwin / expect background problem

[rfa/dejagnu] Work around cygwin / expect background problem

[Andrew Cagney]

[-- Attachment #1: Type: text/plain, Size: 422 bytes --]

Hello,

The attached patch tweeks dejagnu/lib/remote/exp so that it uses the 
shell's background rather than the TCL background operator when trying 
to kill a GDB that's gone wallkabout.

Turns out that expect on cygwin occasionaly and mysteriously die when 
executing the TCL background.  While the cygwin problems now appear 
fixed, the problem is still out there in the field :-(

look ok (fernando I guess?),

Andrew

[-- Attachment #2: diffs --]
[-- Type: text/plain, Size: 2580 bytes --]

2001-11-12  Andrew Cagney  <ac131313@redhat.com>

	* lib/remote.exp (local_exec, standard_close): Use SHELL
	background instead of tcl background to background the kill
	processes.  Work around cygwin/expect problem.

Index: lib/remote.exp
===================================================================
RCS file: /cvs/src/src/dejagnu/lib/remote.exp,v
retrieving revision 1.2
diff -p -r1.2 remote.exp
*** remote.exp	2001/10/30 15:20:05	1.2
--- remote.exp	2001/11/12 18:12:52
*************** proc local_exec { commandline inp outp t
*** 179,185 ****
      # command in order to execute the execution. (English. Gotta love it.)
      if { ! $got_eof } {
  	verbose "killing $pid $pgid";
! 	exec sh -c "exec > /dev/null 2>&1 && (kill -2 $pgid || kill -2 $pid) && sleep 5 && (kill -15 $pgid || kill $pid) && sleep 5 && (kill -9 $pgid || kill -9 $pid)" &;
      }
      # This will hang if the kill doesn't work. Nothin' to do, and it's not ok.
      catch "close -i $spawn_id";
--- 179,188 ----
      # command in order to execute the execution. (English. Gotta love it.)
      if { ! $got_eof } {
  	verbose "killing $pid $pgid";
! 	# This is very, very nasty. SH, instead of EXPECT, is used to
! 	# run this in the background since, on cygwin, a strange file
! 	# I/O error occures.
! 	exec sh -c "exec > /dev/null 2>&1 && (kill -2 $pgid || kill -2 $pid) && sleep 5 && (kill -15 $pgid || kill $pid) && sleep 5 && (kill -9 $pgid || kill -9 $pid) &";
      }
      # This will hang if the kill doesn't work. Nothin' to do, and it's not ok.
      catch "close -i $spawn_id";
*************** proc standard_close { host } {
*** 314,323 ****
  	}
  	if { $pid > 0 } {
  	    verbose "doing kill, pid is $pid";
! 	    # This is very, very nasty. Then again, if after did something
! 	    # reasonable...
  	    set pgid "-[join $pid { -}]";
! 	    exec sh -c "exec > /dev/null 2>&1 && (kill -2 $pgid || kill -2 $pid) && sleep 5 && (kill $pgid || kill $pid) && sleep 5 && (kill -9 $pgid || kill -9 $pid)" &;
  	}
  	verbose "pid is $pid";
  	catch "close -i $shell_id";
--- 317,327 ----
  	}
  	if { $pid > 0 } {
  	    verbose "doing kill, pid is $pid";
! 	    # This is very, very nasty. SH, instead of EXPECT, is used to
! 	    # run this in the background since, on cygwin, a strange file
! 	    # I/O error occures.
  	    set pgid "-[join $pid { -}]";
! 	    exec sh -c "exec > /dev/null 2>&1 && (kill -2 $pgid || kill -2 $pid) && sleep 5 && (kill $pgid || kill $pid) && sleep 5 && (kill -9 $pgid || kill -9 $pid) &";
  	}
  	verbose "pid is $pid";
  	catch "close -i $shell_id";

Re: [rfa/dejagnu] Work around cygwin / expect background problem

[Fernando Nasser]

Andrew Cagney wrote:
> 
> Hello,
> 
> The attached patch tweeks dejagnu/lib/remote/exp so that it uses the
> shell's background rather than the TCL background operator when trying
> to kill a GDB that's gone wallkabout.
> 
> Turns out that expect on cygwin occasionaly and mysteriously die when
> executing the TCL background.  While the cygwin problems now appear
> fixed, the problem is still out there in the field :-(
> 
> look ok (fernando I guess?),
> 
> Andrew
> 

OK with me.  But maybe you should say "on older versions of cygwin", or 
even on versions of cygwin older than <ask Chris>" in this comment:

Fernando


> !       # run this in the background since, on cygwin, a strange file


>   ------------------------------------------------------------------------
> 2001-11-12  Andrew Cagney  <ac131313@redhat.com>
> 
>         * lib/remote.exp (local_exec, standard_close): Use SHELL
>         background instead of tcl background to background the kill
>         processes.  Work around cygwin/expect problem.
> 
> Index: lib/remote.exp
> ===================================================================
> RCS file: /cvs/src/src/dejagnu/lib/remote.exp,v
> retrieving revision 1.2
> diff -p -r1.2 remote.exp
> *** remote.exp  2001/10/30 15:20:05     1.2
> --- remote.exp  2001/11/12 18:12:52
> *************** proc local_exec { commandline inp outp t
> *** 179,185 ****
>       # command in order to execute the execution. (English. Gotta love it.)
>       if { ! $got_eof } {
>         verbose "killing $pid $pgid";
> !       exec sh -c "exec > /dev/null 2>&1 && (kill -2 $pgid || kill -2 $pid) && sleep 5 && (kill -15 $pgid || kill $pid) && sleep 5 && (kill -9 $pgid || kill -9 $pid)" &;
>       }
>       # This will hang if the kill doesn't work. Nothin' to do, and it's not ok.
>       catch "close -i $spawn_id";
> --- 179,188 ----
>       # command in order to execute the execution. (English. Gotta love it.)
>       if { ! $got_eof } {
>         verbose "killing $pid $pgid";
> !       # This is very, very nasty. SH, instead of EXPECT, is used to
> !       # run this in the background since, on cygwin, a strange file
> !       # I/O error occures.
> !       exec sh -c "exec > /dev/null 2>&1 && (kill -2 $pgid || kill -2 $pid) && sleep 5 && (kill -15 $pgid || kill $pid) && sleep 5 && (kill -9 $pgid || kill -9 $pid) &";
>       }
>       # This will hang if the kill doesn't work. Nothin' to do, and it's not ok.
>       catch "close -i $spawn_id";
> *************** proc standard_close { host } {
> *** 314,323 ****
>         }
>         if { $pid > 0 } {
>             verbose "doing kill, pid is $pid";
> !           # This is very, very nasty. Then again, if after did something
> !           # reasonable...
>             set pgid "-[join $pid { -}]";
> !           exec sh -c "exec > /dev/null 2>&1 && (kill -2 $pgid || kill -2 $pid) && sleep 5 && (kill $pgid || kill $pid) && sleep 5 && (kill -9 $pgid || kill -9 $pid)" &;
>         }
>         verbose "pid is $pid";
>         catch "close -i $shell_id";
> --- 317,327 ----
>         }
>         if { $pid > 0 } {
>             verbose "doing kill, pid is $pid";
> !           # This is very, very nasty. SH, instead of EXPECT, is used to
> !           # run this in the background since, on cygwin, a strange file
> !           # I/O error occures.
>             set pgid "-[join $pid { -}]";
> !           exec sh -c "exec > /dev/null 2>&1 && (kill -2 $pgid || kill -2 $pid) && sleep 5 && (kill $pgid || kill $pid) && sleep 5 && (kill -9 $pgid || kill -9 $pid) &";
>         }
>         verbose "pid is $pid";
>         catch "close -i $shell_id";

-- 
Fernando Nasser
Red Hat Canada Ltd.                     E-Mail:  fnasser@redhat.com
2323 Yonge Street, Suite #300
Toronto, Ontario   M4P 2C9

Re: [rfa/dejagnu] Work around cygwin / expect background problem

[Chris Faylor]

On Tue, Nov 13, 2001 at 11:18:48AM -0500, Fernando Nasser wrote:
>Andrew Cagney wrote:
>> 
>> Hello,
>> 
>> The attached patch tweeks dejagnu/lib/remote/exp so that it uses the
>> shell's background rather than the TCL background operator when trying
>> to kill a GDB that's gone wallkabout.
>> 
>> Turns out that expect on cygwin occasionaly and mysteriously die when
>> executing the TCL background.  While the cygwin problems now appear
>> fixed, the problem is still out there in the field :-(
>> 
>> look ok (fernando I guess?),
>
>OK with me.  But maybe you should say "on older versions of cygwin", or 
>even on versions of cygwin older than <ask Chris>" in this comment:

I guess "some older versions of cygwin" would be correct.  I don't know
exactly when this was fixed.

cgf

Re: [rfa/dejagnu] Work around cygwin / expect background problem

[Andrew Cagney]

>>OK with me.  But maybe you should say "on older versions of cygwin", or 
>>even on versions of cygwin older than <ask Chris>" in this comment:
> 
> 
> I guess "some older versions of cygwin" would be correct.  I don't know
> exactly when this was fixed.


Ok, I've checked it in.  (Added a note to mention nickd also who 
originally figured out the problem).

	Andrew