public inbox for cygwin@cygwin.com
 help / color / mirror / Atom feed
* Re: Cygwin 64 problem
       [not found] <CAGX=k3HczQGx=51xRwyKrwpKTMqaOVqxj+ud4UH=i1yCPAFGWQ@mail.gmail.com>
@ 2016-11-28 12:03 ` Ken Brown
  2016-11-28 14:34   ` Duncan Roe
  2016-11-30  3:01   ` Ken Brown
  0 siblings, 2 replies; 5+ messages in thread
From: Ken Brown @ 2016-11-28 12:03 UTC (permalink / raw)
  To: Girish Joglekar; +Cc: cygwin

[-- Attachment #1: Type: text/plain, Size: 1581 bytes --]

On 11/27/2016 9:03 AM, Girish Joglekar wrote:
> Hi Ken,
> Thank you for your reply on the Cygwin mailing list. I have created a
> slightly smaller test example smaller than what I had created in Feb.
> Attached is a tar file. After make insall if you run TEST.x and click
> the Open button, it gives segmentation fault.
> Hope you will be able to help.
> Thank you.
> Girish

Please keep the discussion on the mailing list so that people who might 
be able to help will see it.  I personally am not familiar with X11 
programming, so I'm not one of those people.

I do have a few comments, however:

1. This may be slightly smaller than your previous program, but it is 
still *very* far from a small test case.  You really can't expect people 
to debug a program of this size for you.

2. I suggest that you build with -Wall (as Marco already suggested in 
the earlier discussion) and fix all the compiler warnings.  I saw 
several that need attention (aside from the many "unused variable" 
warnings).  It wouldn't hurt to also use -Wextra.

3. I also suggest that you use gdb to find out where the program is 
crashing and why.  You made a start on that in the discussion last 
March, but you didn't follow through.  (By the way, I recommend building 
without optimization; otherwise you may not be able to get accurate 
information about the crash from gdb.)

4. If, after the steps above, you still think the crash is due to a 
Cygwin bug rather than a programming error, then try to extract from 
your program a test case as small as possible that reproduces the crash.

Ken

[-- Attachment #2: test2.tar --]
[-- Type: application/octet-stream, Size: 686080 bytes --]

[-- Attachment #3: Type: text/plain, Size: 218 bytes --]

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: Cygwin 64 problem
  2016-11-28 12:03 ` Cygwin 64 problem Ken Brown
@ 2016-11-28 14:34   ` Duncan Roe
  2016-11-28 16:19     ` Ken Brown
  2016-11-30  3:01   ` Ken Brown
  1 sibling, 1 reply; 5+ messages in thread
From: Duncan Roe @ 2016-11-28 14:34 UTC (permalink / raw)
  To: cygwin

On Sun, Nov 27, 2016 at 11:02:29AM -0500, Ken Brown wrote:
> On 11/27/2016 9:03 AM, Girish Joglekar wrote:
> > Hi Ken,
> > Thank you for your reply on the Cygwin mailing list. I have created a
> > slightly smaller test example smaller than what I had created in Feb.
> > Attached is a tar file. After make insall if you run TEST.x and click
> > the Open button, it gives segmentation fault.
> > Hope you will be able to help.
> > Thank you.
> > Girish
>
> Please keep the discussion on the mailing list so that people who might be
> able to help will see it.  I personally am not familiar with X11
> programming, so I'm not one of those people.
>
> I do have a few comments, however:
>
> 1. This may be slightly smaller than your previous program, but it is still
> *very* far from a small test case.  You really can't expect people to debug
> a program of this size for you.
>
> 2. I suggest that you build with -Wall (as Marco already suggested in the
> earlier discussion) and fix all the compiler warnings.  I saw several that
> need attention (aside from the many "unused variable" warnings).  It
> wouldn't hurt to also use -Wextra.
>
> 3. I also suggest that you use gdb to find out where the program is crashing
> and why.  You made a start on that in the discussion last March, but you
> didn't follow through.  (By the way, I recommend building without
> optimization; otherwise you may not be able to get accurate information
> about the crash from gdb.)
>
> 4. If, after the steps above, you still think the crash is due to a Cygwin
> bug rather than a programming error, then try to extract from your program a
> test case as small as possible that reproduces the crash.
>
> Ken

I thought I would have a crack at this but immediately got the error:

testc.h:8:19: fatal error: Xm/Xm.h: No such file or directory
 #include <Xm/Xm.h>

I thought /usr/include/Xm/ should come with motif, which I have installed.

What have I missed?

Cheers ... Duncan.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: Cygwin 64 problem
  2016-11-28 14:34   ` Duncan Roe
@ 2016-11-28 16:19     ` Ken Brown
  0 siblings, 0 replies; 5+ messages in thread
From: Ken Brown @ 2016-11-28 16:19 UTC (permalink / raw)
  To: cygwin

On 11/27/2016 5:30 PM, Duncan Roe wrote:
> On Sun, Nov 27, 2016 at 11:02:29AM -0500, Ken Brown wrote:
>> On 11/27/2016 9:03 AM, Girish Joglekar wrote:
>>> Hi Ken,
>>> Thank you for your reply on the Cygwin mailing list. I have created a
>>> slightly smaller test example smaller than what I had created in Feb.
>>> Attached is a tar file. After make insall if you run TEST.x and click
>>> the Open button, it gives segmentation fault.
>>> Hope you will be able to help.
>>> Thank you.
>>> Girish
>>
>> Please keep the discussion on the mailing list so that people who might be
>> able to help will see it.  I personally am not familiar with X11
>> programming, so I'm not one of those people.
>>
>> I do have a few comments, however:
>>
>> 1. This may be slightly smaller than your previous program, but it is still
>> *very* far from a small test case.  You really can't expect people to debug
>> a program of this size for you.
>>
>> 2. I suggest that you build with -Wall (as Marco already suggested in the
>> earlier discussion) and fix all the compiler warnings.  I saw several that
>> need attention (aside from the many "unused variable" warnings).  It
>> wouldn't hurt to also use -Wextra.
>>
>> 3. I also suggest that you use gdb to find out where the program is crashing
>> and why.  You made a start on that in the discussion last March, but you
>> didn't follow through.  (By the way, I recommend building without
>> optimization; otherwise you may not be able to get accurate information
>> about the crash from gdb.)
>>
>> 4. If, after the steps above, you still think the crash is due to a Cygwin
>> bug rather than a programming error, then try to extract from your program a
>> test case as small as possible that reproduces the crash.
>>
>> Ken
>
> I thought I would have a crack at this but immediately got the error:
>
> testc.h:8:19: fatal error: Xm/Xm.h: No such file or directory
>  #include <Xm/Xm.h>
>
> I thought /usr/include/Xm/ should come with motif, which I have installed.
>
> What have I missed?

libXm-devel

Ken


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: Cygwin 64 problem
  2016-11-28 12:03 ` Cygwin 64 problem Ken Brown
  2016-11-28 14:34   ` Duncan Roe
@ 2016-11-30  3:01   ` Ken Brown
  2016-12-04 15:59     ` Ken Brown
  1 sibling, 1 reply; 5+ messages in thread
From: Ken Brown @ 2016-11-30  3:01 UTC (permalink / raw)
  To: cygwin

[-- Attachment #1: Type: text/plain, Size: 4279 bytes --]

On 11/27/2016 11:02 AM, Ken Brown wrote:
> On 11/27/2016 9:03 AM, Girish Joglekar wrote:
>> Hi Ken,
>> Thank you for your reply on the Cygwin mailing list. I have created a
>> slightly smaller test example smaller than what I had created in Feb.
>> Attached is a tar file. After make insall if you run TEST.x and click
>> the Open button, it gives segmentation fault.
>> Hope you will be able to help.
>> Thank you.
>> Girish
> 
> Please keep the discussion on the mailing list so that people who might
> be able to help will see it.  I personally am not familiar with X11
> programming, so I'm not one of those people.
> 
> I do have a few comments, however:
> 
> 1. This may be slightly smaller than your previous program, but it is
> still *very* far from a small test case.  You really can't expect people
> to debug a program of this size for you.
> 
> 2. I suggest that you build with -Wall (as Marco already suggested in
> the earlier discussion) and fix all the compiler warnings.  I saw
> several that need attention (aside from the many "unused variable"
> warnings).  It wouldn't hurt to also use -Wextra.
> 
> 3. I also suggest that you use gdb to find out where the program is
> crashing and why.  You made a start on that in the discussion last
> March, but you didn't follow through.  (By the way, I recommend building
> without optimization; otherwise you may not be able to get accurate
> information about the crash from gdb.)
> 
> 4. If, after the steps above, you still think the crash is due to a
> Cygwin bug rather than a programming error, then try to extract from
> your program a test case as small as possible that reproduces the crash.

I've looked at this a little, and the crash occurs because of a bad value of a pointer to a font list.  I'm attaching an annotated transcript of a gdb session that illustrates this.  I didn't try to figure out where the bad pointer came from.

As a workaround, you can avoid the crash by removing from mosprshx.c all uses of spreadSheetFont, as in the following patch:

--- mosprshx.c~ 2016-11-27 08:46:12.000000000 -0500
+++ mosprshx.c  2016-11-29 14:58:23.114377500 -0500
@@ -29,8 +29,6 @@
    0x00, 0x7c, 0x07, 0x1e, 0x0f, 0x07, 0x9e, 0x03, 0xfc, 0x01, 0xf8, 0x00,
    0x70, 0x00, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00};
 
-static char *spreadSheetFont = "-adobe-courier-medium-r-normal--12-120-75-75-m-70-iso8859-1";
-
 #ifdef __STDC__
 static void mosprshCreateCheckMarkPixmap(Widget pushButton)
 #else
@@ -208,11 +206,6 @@
    if(isParentManaged)
       XtUnmanageChild(parentFM);
 
-   fromValue.addr = spreadSheetFont;
-   fromValue.size = utStringLength(fromValue.addr) + 1;
-   toValue.addr = NULL;
-   XtConvertAndStore(parentFM,XmRString,&fromValue,XmRFontList,&toValue);
-
    ac = 0;
    XtSetArg(al[ac],XmNuserData,(XtPointer)spreadSheetSize); ac++;
    spreadSheet = XmCreateForm(parentFM,"SpreadSheet",al,ac);
@@ -234,9 +227,6 @@
       XtSetArg(al[ac],XmNtraversalOn,FALSE); ac++;
    }
    XtSetArg(al[ac],XmNcolumns,longestFixedColumnSize+1); ac++;
-   if(toValue.addr) {
-      XtSetArg(al[ac],XmNfontList,*(unsigned int *)toValue.addr); ac++;
-   }
    for(i1=0 ; i1<nRows ; i1++) {
       sprintf(widgetName,"fixedColumnRow%d",i1);
       fixedColumnTF[i1] = XmCreateTextField(fixedColumn,widgetName,al,ac);
@@ -264,9 +254,6 @@
    if(cellSize > 0) {
       XtSetArg(al[ac],XmNcolumns,cellSize); ac++;
    }
-   if(toValue.addr) {
-      XtSetArg(al[ac],XmNfontList,*(unsigned int *)toValue.addr); ac++;
-   }
    for(j1=0 ; j1<nColumns ; j1++) {
       sprintf(widgetName,"fixedRowColumn%d",j1);
       fixedRowTF[j1] = XmCreateTextField(fixedRow,widgetName,al,ac);
@@ -294,9 +281,6 @@
             XtSetArg(al[ac],XmNcolumns,cellSize); ac++;
             XtSetArg(al[ac],XmNmaxLength,cellSize); ac++;
          }
-         if(toValue.addr) {
-            XtSetArg(al[ac],XmNfontList,*(unsigned int *)toValue.addr); ac++;
-         }
          sprintf(widgetName,"column%drow%d",j1,i1);
          variableRowAndColumnTF[loc] = XmCreateTextField(variableColumnFM[j1],widgetName,al,ac); loc++;
       }
@@ -886,15 +870,3 @@
       *nColumns = 0;
    }
 }
-
-
-#ifdef __STDC__
-char *moGetSpreadsheetFont()
-#else
-char *moGetSpreadsheetFont()
-#endif
-{
-   return(spreadSheetFont);
-}
-
-

Ken

[-- Attachment #2: gdb.txt --]
[-- Type: text/plain, Size: 3445 bytes --]

Thread 1 "TEST.x" received signal SIGSEGV, Segmentation fault.
0x00000003fc68c4e5 in XmRenderTableCopy (table=0xaeeb0, tags=tags@entry=0x0, 
    tag_count=tag_count@entry=0)
    at /usr/src/debug/motif-2.3.6-1/lib/Xm/XmRenderT.c:1709
1709	  if (_XmRTDisplay(table))

[Crash here because the value of table is bogus.  See below.

+bt
#0  0x00000003fc68c4e5 in XmRenderTableCopy (table=0xaeeb0, 
    tags=tags@entry=0x0, tag_count=tag_count@entry=0)
    at /usr/src/debug/motif-2.3.6-1/lib/Xm/XmRenderT.c:1709
#1  0x00000003fc6831ba in XmFontListCopy (fontlist=<optimized out>)
    at /usr/src/debug/motif-2.3.6-1/lib/Xm/XmFontList.c:637
#2  0x00000003fc657920 in InitializeTextStruct (tf=tf@entry=0x6000af730)
    at /usr/src/debug/motif-2.3.6-1/lib/Xm/TextF.c:7295
#3  0x00000003fc657e67 in Initialize (request=request@entry=0x6000afb40, 
    new_w=new_w@entry=0x6000af730, args=args@entry=0xffffbd60, 
    num_args=num_args@entry=0xffffb320)
    at /usr/src/debug/motif-2.3.6-1/lib/Xm/TextF.c:7817
#4  0x00000003fc469bb1 in CallInitialize (
    class=0x3fc72c440 <xmTextFieldClassRec>, 
    req_widget=req_widget@entry=0x6000afb40, 
    new_widget=new_widget@entry=0x6000af730, args=args@entry=0xffffbd60, 
    num_args=num_args@entry=8) at /usr/src/debug/libXt-1.1.5-1/src/Create.c:226
#5  0x00000003fc46a65c in xtCreate (
    name=name@entry=0xffffb950 "fixedColumnRow0", class=class@entry=0x0, 
    widget_class=widget_class@entry=0x3fc72c440 <xmTextFieldClassRec>, 
    parent=0x1fc5dd590, parent@entry=0x6000af0d0, default_screen=0x600052780, 
    args=args@entry=0xffffbd60, num_args=num_args@entry=8, 
    typed_args=typed_args@entry=0x0, num_typed_args=num_typed_args@entry=0, 
    parent_constraint_class=0x3fc720dc0 <xmFormClassRec>, 
    post_proc=post_proc@entry=0x3fc469bf0 <widgetPostProc>)
    at /usr/src/debug/libXt-1.1.5-1/src/Create.c:416
#6  0x00000003fc46a906 in _XtCreateWidget (
    name=name@entry=0xffffb950 "fixedColumnRow0", 
    widget_class=widget_class@entry=0x3fc72c440 <xmTextFieldClassRec>, 
    parent=parent@entry=0x6000af0d0, args=args@entry=0xffffbd60, 
    num_args=num_args@entry=8, typed_args=typed_args@entry=0x0, 
    num_typed_args=num_typed_args@entry=0)
    at /usr/src/debug/libXt-1.1.5-1/src/Create.c:570
#7  0x00000003fc46abd9 in XtCreateWidget (name=0xffffb950 "fixedColumnRow0", 
    widget_class=0x3fc72c440 <xmTextFieldClassRec>, parent=0x6000af0d0, 
    args=0xffffbd60, num_args=8)
    at /usr/src/debug/libXt-1.1.5-1/src/Create.c:589
#8  0x00000003fc658a81 in XmCreateTextField (parent=<optimized out>, 
    name=<optimized out>, arglist=<optimized out>, argcount=<optimized out>)
    at /usr/src/debug/motif-2.3.6-1/lib/Xm/TextF.c:10645
#9  0x000000010040882d in mosprshCreateSpreadsheet (parentFM=0x6000acc60, 
    spreadSheetTitle=0x100421087 "Measuring Unit Specification", 
    rowNames=0x6000ab4f0, columnNames=0x6000ab4b0, cellSize=0, 
    mosprshCheckButtonCallback=0x0) at mosprshx.c:242
[...]

+p *table
$1 = (struct __XmRenderTableRec *) 0x56f700001ea1
+p *$
Cannot access memory at address 0x56f700001ea1
+f 2
#2  0x00000003fc657920 in InitializeTextStruct (tf=tf@entry=0x6000af730)
    at /usr/src/debug/motif-2.3.6-1/lib/Xm/TextF.c:7295
7295	     TextF_FontList(tf) = (XmFontList)XmFontListCopy(TextF_FontList(tf));
+p tf->text.font_list
$2 = (XmFontList) 0xaeeb0

[This is the bad pointer that was propagated to XmRenderTableCopy.  I
didn't chase down where it came from.]


[-- Attachment #3: Type: text/plain, Size: 218 bytes --]

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: Cygwin 64 problem
  2016-11-30  3:01   ` Ken Brown
@ 2016-12-04 15:59     ` Ken Brown
  0 siblings, 0 replies; 5+ messages in thread
From: Ken Brown @ 2016-12-04 15:59 UTC (permalink / raw)
  To: cygwin, Girish Joglekar

On 11/29/2016 3:14 PM, Ken Brown wrote:
> I've looked at this a little, and the crash occurs because of a bad value of a pointer to a font list.  I'm attaching an annotated transcript of a gdb session that illustrates this.  I didn't try to figure out where the bad pointer came from.

I've done that now.  There are three places in mosprshx.c where you cast a pointer to a pointer to an object of a different size.  Interesting things happen when you dereference that pointer.  The following patch gets rid of the crash:

--- mosprshx.c~ 2016-12-04 10:02:36.706325200 -0500
+++ mosprshx.c  2016-12-04 10:28:35.160038200 -0500
@@ -235,7 +235,7 @@
    }
    XtSetArg(al[ac],XmNcolumns,longestFixedColumnSize+1); ac++;
    if(toValue.addr) {
-      XtSetArg(al[ac],XmNfontList,*(unsigned int *)toValue.addr); ac++;
+      XtSetArg(al[ac],XmNfontList,*(XtArgVal *)toValue.addr); ac++;
    }
    for(i1=0 ; i1<nRows ; i1++) {
       sprintf(widgetName,"fixedColumnRow%d",i1);
@@ -265,7 +265,7 @@
       XtSetArg(al[ac],XmNcolumns,cellSize); ac++;
    }
    if(toValue.addr) {
-      XtSetArg(al[ac],XmNfontList,*(unsigned int *)toValue.addr); ac++;
+      XtSetArg(al[ac],XmNfontList,*(XtArgVal *)toValue.addr); ac++;
    }
    for(j1=0 ; j1<nColumns ; j1++) {
       sprintf(widgetName,"fixedRowColumn%d",j1);
@@ -295,7 +295,7 @@
             XtSetArg(al[ac],XmNmaxLength,cellSize); ac++;
          }
          if(toValue.addr) {
-            XtSetArg(al[ac],XmNfontList,*(unsigned int *)toValue.addr); ac++;
+            XtSetArg(al[ac],XmNfontList,*(XtArgVal *)toValue.addr); ac++;
          }
          sprintf(widgetName,"column%drow%d",j1,i1);
          variableRowAndColumnTF[loc] = XmCreateTextField(variableColumnFM[j1],widgetName,al,ac); loc++;


Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2016-12-04 15:59 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <CAGX=k3HczQGx=51xRwyKrwpKTMqaOVqxj+ud4UH=i1yCPAFGWQ@mail.gmail.com>
2016-11-28 12:03 ` Cygwin 64 problem Ken Brown
2016-11-28 14:34   ` Duncan Roe
2016-11-28 16:19     ` Ken Brown
2016-11-30  3:01   ` Ken Brown
2016-12-04 15:59     ` Ken Brown

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).