Dialog-box closing patch (Was: Milestone 11, release candidate 3

On Friday 21 July 2006 12:33, Bronek Kozicki wrote:
> Vladimir Prus wrote:
> > I like this patch. It's now committed.
>
> I'm sorry to say this, but it was still buggy.
Attached you will find
> relevant patch

Bronek,
I've applied this patch. I attach the file I have now in
CVS, can you check 
that it's right?

Thanks for your patience in working on the right solution!


- Volodya
-- 
Vladimir Prus
http://vladimir_pru
s.blogspot.com
Boost.Build V2: http://boost.org/boost-
build2
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build

Vladimir Prus wrote:
> Bronek,
> I've applied this patch. I attach the file I have now
in CVS, can you check 
> that it's right?

that's OK, it's identical with the code I use and it works
fine.

> Thanks for your patience in working on the right
solution!

thank YOU for your patience accepting my endless patches !


B.
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build

Bronek Kozicki wrote:
> Vladimir Prus wrote:
>> Thanks for your patience in working on the right
solution!
> 
> thank YOU for your patience accepting my endless
patches !


Hi Volodya

I still do not get it right  I will
send you another mail with another 
patch today, now I just want to let you know that the code
is still (a little) 
buggy - it closes dialogs displayed by procexp.exe when run
instead of 
taskmgr.exe, which is child of winlogon.exe (which in turn
is child of 
smss.exe). Nothing critical, but annoying.


B.
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build

Bronek Kozicki wrote:

> I still do not get it right  I will
send you another mail with another 
> patch today, now I just want to let you know that the
code is still (a little) 
> buggy - it closes dialogs displayed by procexp.exe when
run instead of 
> taskmgr.exe, which is child of winlogon.exe (which in
turn is child of 
> smss.exe). Nothing critical, but annoying.


This should fix the problem, I started testing.


B.


/* Recursive check if first process is parent (directly or
indirectly) of
the second one. Both processes are passed as process ids,
not handles.
Special return value 2 means that the second processs is
smss.exe and its
parent process is System (first argument is ignored) */
static int
is_parent_child(DWORD parent, DWORD child)
{
     HANDLE process_snapshot_h = INVALID_HANDLE_VALUE;

     if (!child)
         return 0;
     if (parent == child)
         return 1;

     process_snapshot_h =
CreateToolhelp32Snapshot(TH32CS_SNAPPROCESS,0);
     if (INVALID_HANDLE_VALUE != process_snapshot_h)
     {
         BOOL ok = TRUE;
         PROCESSENTRY32 pinfo;
         pinfo.dwSize = sizeof(PROCESSENTRY32);
         for (
             ok = Process32First(process_snapshot_h,
&pinfo);
             ok == TRUE;
             ok = Process32Next(process_snapshot_h,
&pinfo) )
         {
             if (pinfo.th32ProcessID == child)
             {
                 /*
                 Unfortunately, process ids are not really
unique. There might
                 be spurious "parent and child"
relationship match between
                 two non-related processes if real parent
process of a given
                 process has exited (while child process
kept running as an
                 "orphan") and the process id of
such parent process has been
                 reused by internals of the operating system
when creating
                 another process. Thus additional check is
needed - process
                 creation time. This check may fail (ie.
return 0) for system
                 processes due to insufficient privileges,
and that's OK. */
                 double tchild = 0.0;
                 double tparent = 0.0;
                 HANDLE hchild =
OpenProcess(PROCESS_QUERY_INFORMATION, FALSE, 
pinfo.th32ProcessID);

                 CloseHandle(process_snapshot_h);

                 /* csrss.exe may display message box like
following:
                     xyz.exe - Unable To Locate Component
                     This application has failed to start
because
                     boost_foo-bar.dll was not found.
Re-installing the
                     application may fix the problem
                 This actually happens when starting test
process that depends
                 on a dynamic library which failed to build.
We want to
                 automatically close these message boxes
even though csrss.exe
                 is not our child process. We may depend on
the fact that (in
                 all current versions of Windows) csrss.exe
is directly
                 child of smss.exe process, which in turn is
directly child of
                 System process, which always has process id
== 4 .
                 This check must be performed before
comparison of process
                 creation time */
                 if (stricmp(pinfo.szExeFile,
"csrss.exe") == 0
                     && is_parent_child(parent,
pinfo.th32ParentProcessID) == 2)
                 {
                     return 1;
                 }
                 else if (stricmp(pinfo.szExeFile,
"smss.exe") == 0
                     && pinfo.th32ParentProcessID ==
4)
                 {
                     return 2;
                 }

                 if (hchild != 0)
                 {
                     HANDLE hparent =
OpenProcess(PROCESS_QUERY_INFORMATION, 
FALSE, pinfo.th32ParentProcessID);
                     if (hparent != 0)
                     {
                         tchild = creation_time(hchild);
                         tparent = creation_time(hparent);

                         CloseHandle(hparent);
                     }
                     CloseHandle(hchild);
                 }

                 /* return 0 if one of the following is
true:
                 1. we failed to read process creation time
                 2. child was created before alleged parent
*/
                 if (tchild == 0.0 || tparent == 0.0 ||
tchild < tparent)
                     return 0;

                 return is_parent_child(parent,
pinfo.th32ParentProcessID) & 1;
             }
         }

         CloseHandle(process_snapshot_h);
     }

     return 0;
}
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build

On Monday 24 July 2006 01:36, Bronek Kozicki wrote:
> Bronek Kozicki wrote:
> > I still do not get it right  I will
send you another mail with another
> > patch today, now I just want to let you know that
the code is still (a
> > little) buggy - it closes dialogs displayed by
procexp.exe when run
> > instead of taskmgr.exe, which is child of
winlogon.exe (which in turn is
> > child of smss.exe). Nothing critical, but
annoying.
>
> This should fix the problem, I started testing.

Hi Bronek,
sorry for getting picky, but can you send me either diff
(preferred), or a 
modified file? And in either case, as attachment. Pasting
just part of code 
in email is not convenient, as some lines seem to get
wrapped, and it's not 
applicable automatically.

Thanks,
Volodya
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build

Vladimir Prus wrote:
> On Monday 24 July 2006 01:36, Bronek Kozicki wrote:
>> Bronek Kozicki wrote:
>>> I still do not get it right  I will
send you another mail with another
>>> patch today, now I just want to let you know
that the code is still (a
>>> little) buggy - it closes dialogs displayed by
procexp.exe when run
>>> instead of taskmgr.exe, which is child of
winlogon.exe (which in turn is
>>> child of smss.exe). Nothing critical, but
annoying.
>> This should fix the problem, I started testing.
> 
> sorry for getting picky, but can you send me either
diff (preferred), or a 
> modified file? 

Please find attached. Tests results are OK.


B.

--- C:\Documents and Settings\Bronek
Kozicki\Desktop\execnt.c	Mon Jul 24 06:50:15 2006 UTC
+++
C:\DEVEL\BOOST_RTEST\boost\tools\jam\src\execnt.c	Mon
Jul 24 06:53:01 2006 UTC
 -1008,11
+1008,13 
     /* now that the children are all dead, kill the root */
     TerminateProcess(process,-2);
 }
 
 /* Recursive check if first process is parent (directly or
indirectly) of 
-the second one. Both processes are passed as process ids,
not handles */
+the second one. Both processes are passed as process ids,
not handles.
+Special return value 2 means that the second process is
smss.exe and its 
+parent process is System (first argument is ignored) */
 static int 
 is_parent_child(DWORD parent, DWORD child)
 {
     HANDLE process_snapshot_h = INVALID_HANDLE_VALUE;
 
 -1063,14
+1065,18 
                 child of smss.exe process, which in turn is
directly child of
                 System process, which always has process id
== 4 .
                 This check must be performed before
comparison of process 
                 creation time */
                 if (stricmp(pinfo.szExeFile,
"csrss.exe") == 0
-                    || stricmp(pinfo.szExeFile,
"smss.exe") == 0)
+                    && is_parent_child(parent,
pinfo.th32ParentProcessID) == 2)
                 {
-                    if (is_parent_child(4,
pinfo.th32ParentProcessID))
-                        return 1;
+                    return 1;
+                }
+                else if (stricmp(pinfo.szExeFile,
"smss.exe") == 0
+                    && pinfo.th32ParentProcessID ==
4)
+                {
+                    return 2;
                 }
 
                 if (hchild != 0)
                 {
                     HANDLE hparent =
OpenProcess(PROCESS_QUERY_INFORMATION, FALSE,
pinfo.th32ParentProcessID);
 -1088,11
+1094,11 
                 1. we failed to read process creation time
                 2. child was created before alleged parent
*/
                 if (tchild == 0.0 || tparent == 0.0 ||
tchild < tparent)
                     return 0;
 
-                return is_parent_child(parent,
pinfo.th32ParentProcessID);
+                return is_parent_child(parent,
pinfo.th32ParentProcessID) & 1;
             }
         }
 
         CloseHandle(process_snapshot_h);
     }
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build

On Monday 24 July 2006 10:55, Bronek Kozicki wrote:
> Vladimir Prus wrote:
> > On Monday 24 July 2006 01:36, Bronek Kozicki
wrote:
> >> Bronek Kozicki wrote:
> >>> I still do not get it right  I will
send you another mail with
> >>> another patch today, now I just want to
let you know that the code is
> >>> still (a little) buggy - it closes dialogs
displayed by procexp.exe
> >>> when run instead of taskmgr.exe, which is
child of winlogon.exe (which
> >>> in turn is child of smss.exe). Nothing
critical, but annoying.
> >>
> >> This should fix the problem, I started
testing.
> >
> > sorry for getting picky, but can you send me
either diff (preferred), or
> > a modified file?
>
> Please find attached. Tests results are OK.

Applied, thanks.

- Volodya
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build

Vladimir Prus wrote:
>> Please find attached. Tests results are OK.
> 
> Applied, thanks.

Thanks for your patience applying these fixes. I really hope
this one was last.


B.
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build