When we kill a child we send SIGSTOP first so that we can traverse its
child processes recursively.  On unwinding the recursion we then send
SIGKILL.  Current Cygwin has trouble when both signals are sent in quick
succession by the parent process.  Add a usleep(1) after sending the
first signal to give up our time slice and give Cygwin a chance to
process the first signal before sending the second.

Change-Id: I75ab55cf969a0ea6104a9798c761c0ba1fcd1dbc

If a group of processes needs to be terminated recursively the entry in /proc
doesn't need to be kept open during the recursion as all the interesting
information was already read from it. Especially in case of deeper recursion
this lowers the memory pressure.

Change-Id: I95b969f42c8dd68b57cdaf8fc1b705f358f1dc58

Paul Kunysch authored 12 years ago and Brad King committed 12 years ago

The Single UNIX (R) Specification, Version 2 defines "FD_SETSIZE", but
does not specify the effects if it's (re)defined by an application.

Cygwin's newlib headers document that FD_SETSIZE may be defined
by the user [1] and should be >= NOFILE. [2]  Setting it to 16384
uses a value from Cygwin's poll() implementation. [3]

[1]: http://cygwin.com/cgi-bin/cvsweb.cgi/src/newlib/libc/include/sys/types.h?cvsroot=src
[2]: http://cygwin.com/cgi-bin/cvsweb.cgi/src/winsup/cygwin/include/sys/param.h?cvsroot=src
[3]: http://cygwin.com/cgi-bin/cvsweb.cgi/src/winsup/cygwin/poll.cc?cvsroot=src

Change-Id: Idc43fc28a398979da1e9289e1080a9fc1090c605

Alan Hourihane authored 12 years ago and Brad King committed 12 years ago

Change-Id: I4e44d61c175e1deb4cecdebfdba5fa05cbdd1e6b

Change-Id: I55506c112cfffdb8d56a85bae0eeaae91d4035bf

Change-Id: I881521d84a31b6b1c68930f94acbcea88b5c1dea

Sean McBride authored 12 years ago and Brad King committed 12 years ago

Change-Id: Ic1dcd37dabbf34ce721b81c6a787dafe758105a4

Do not try to execute a child with no command line.  Previously this led
to a silent hang.

kFreeBSD and Hurd have the same userland as Linux.  This change is
necessary to enable kwsysProcessKill() to kill child processes on
kFreeBSD.  The bug was detected by CTestTestTimeout test.

Patch from "Modestas Vainius <modestas@vainius.eu>".
See issue #10432.

Patch from Chuck Atkins <chuck.atkins@kitware.com>

Teach kwsysProcessKill to identify processes on this platform using the "ps"
command just as on Linux.  Patch from Modestas Vainius <modax@debian.org>.
See issue #10432.

Open output files with mode 0666 so that permissions are not more strict
than umask permits.  See issue #10126.

In order to kill process trees we need to list all processes to find
those whose parent we are killing.  We implement process listing on
OpenSolaris by using "ps -ef" and parsing the resulting format:

     UID   PID  PPID   C    STIME TTY         TIME CMD
     %*s    %d    %d   %*[^\n]\n

In order to kill process trees we need to list all processes to find
those whose parent we are killing.  We implement process listing on QNX
using "ps -Af" and parsing the resulting format:

  UID        PID       PPID  C STIME TTY          TIME CMD
  %*d         %d         %d  %*[^\n]\n

On UNIX systems we kill a tree of processes by performing a DFS walk of
the tree.  We send SIGSTOP to each process encountered, recursively
handle its children, and then send SIGKILL.

We once used the above approach in the past, but it was removed by the
commit "Do not send both SIGSTOP and SIGKILL when killing a process".
The commit was meant to work-around an OS X 10.3 bug in which the child
would not always honor SIGKILL after SIGSTOP.  At the time we wrongly
assumed that the process tree remains intact after SIGKILL and before
the child is reaped.  In fact the grandchildren may be re-parented to
ppid=1 even before the child is reaped, which causes the DFS walk to
miss them.

Alex

This converts the KWSys license to a pure 3-clause OSI-approved BSD
License.  We drop the previous license clause requiring modified
versions to be plainly marked.  We also update the KWSys copyright to
cover the full development time range.

Add System_Parse_CommandForUnix to the KWSys System interface as a
utility to parse a unix-style command line.  Move the existing
implementation out of ProcessUNIX.  Add a flags argument reserved for
future use in providing additional behavior.

This helps avoid fixing VMS-specific code for non-VMS compilers where it
isn't needed anyway.

COMP:Fixed warning with gcc 4.3.3: passing argument 1 of kwsysProcessSetVMSFeature discards qualifiers from pointer target type.

This achieves basic process execution on OpenVMS.  We use work-arounds
for different fork()/exec() behavior and a lack of select().

VMS emulates fork/exec using setjmp/longjmp to evaluate the child and
parent return cases from fork.  Therefore both must be invoked from the
same function.

Since select() works only for sockets we use the BeOS-style polling
implementation.  However, non-blocking reads on empty pipes cannot be
distinguished easily from the last read on a closed pipe.  Therefore we
identify end of data by an empty read after the child terminates.

This avoids use of an uninitialized value in the KWSys ProcessUNIX
polling implementation when no timeout is given.

ENH: check in almost building VMS stuff with VMSBuild directory since the bootstrap script will not work on VMS

COMP:Fixed gcc 4.3.2 warning with -O1 and above: ignoring return value of read'), declared with attribute warn_unused_result

According to "man select" on Linux it is possible that select() lies
about data being ready on a pipe in some subtle cases.  We deal with
this by switching to non-blocking i/o and checking for EAGAIN.  See
issue #7180.

ENH: all ctype function have the same issue: char can be signed or unsigned, since isspace only deal with >=0 value (except EOF) one has to first cast it to unsigned char

BUG: Fix hang in Process_Kill on OS X caused by an OS bug in which a pipe read end cannot be closed if the pipe write end is open, the pipe is full, and another process is blocking waiting to write.  Work around the problem by killing the children before closing the pipes.

BUG: Do not send both SIGSTOP and SIGKILL when killing a process.  The SIGSTOP seems to be able to block the SIGKILL occasionally.  Also the SIGKILL is sufficient since the process table entry will still exist until it is reaped with waitpid.