This is the mail archive of the
mailing list for the Cygwin project.
Re: [Patch] Fixing the PROCESS_DUP_HANDLE security hole.
On Thu, Nov 11, 2004 at 11:36:32PM -0500, Pierre A. Humblet wrote:
>At 11:24 PM 11/11/2004 -0500, Christopher Faylor wrote:
>>On Thu, Nov 11, 2004 at 10:48:57PM -0500, Pierre A. Humblet wrote:
>>>P.S.: I have no news about the recent patch to /bin/kill -f
>>That is because I was sure that I'd used 'kill -f' to kill windows pids
>>in the past and wanted to check your patch. I haven't been near a
>>WinMe system in a while, though. My vmware version isn't working
>Funny, I had the same feeling. But this is what happens now:
> PID PPID PGID WINPID TTY UID STIME COMMAND
> 606855 1 606855 4294360441 con 740 23:06:35 /c/PROGRAM
> 537691 606855 537691 4294504569 0 740 23:06:36 /c/PROGRAM
> 460171 537691 460171 4294214685 0 740 23:24:07 /c/PROGRAM
>~: /bin/kill -f 4294504569
>couldn't open pid 2147483647
>2147483647 = 0x7FFFFFFF, due to strtol saturating.
That's right. I have seen that from time to time.
>I just researched the ChangeLog and found a possible cause:
>2003-09-20 Christopher Faylor <email@example.com>
> * kill.cc (main): Allow negative pids (indicates process groups).
If that is the cause then bypassing that code when -f is specified should
But, nevertheless, go ahead and check in your patch.