c/9058: structure with flexible array member: offsetof() != sizeof()

c/9058: structure with flexible array member: offsetof() != sizeof()

[manfred]

>Number:         9058
>Category:       c
>Synopsis:       structure with flexible array member: offsetof() != sizeof()
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          wrong-code
>Submitter-Id:   net
>Arrival-Date:   Thu Dec 26 04:36:01 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     Manfred Spraul
>Release:        gcc (GCC) 3.2 20020903 (RedHat Linux 8.0 3.2-7)
>Organization:
>Environment:
Linux i386
>Description:
See the attached test case:
struct x {int a; char b; short c[];};

Now sizeof(struct x) is 8, offsetof(struct x, c) is 6. That seems to contradice the C99 standard: section 6.7.2.1, constraint 16.

The problem occurs both with gcc and g++
>How-To-Repeat:
gcc -o simple simple.c
./simple
>Fix:
Workarounds:
- Do not assume that x_ptr->c == x_ptr+1
- Manually pad structure
>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="simple.c"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="simple.c"
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Re: c/9058: structure with flexible array member: offsetof() != sizeof()

[bangerth]

Synopsis: structure with flexible array member: offsetof() != sizeof()

State-Changed-From-To: feedback->closed
State-Changed-By: bangerth
State-Changed-When: Mon Jan 13 13:13:53 2003
State-Changed-Why:
    Closed on request of submitter. Thanks for your quick reply!
    W.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9058

Re: c/9058: structure with flexible array member: offsetof() != sizeof()

[bangerth]

Synopsis: structure with flexible array member: offsetof() != sizeof()

State-Changed-From-To: open->feedback
State-Changed-By: bangerth
State-Changed-When: Sun Jan 12 16:20:32 2003
State-Changed-Why:
    Manfred,
    what's the status of this report. From the discussion I get
    the impression that the matter has been settled. Is this right?
    Can the report be closed?
    
    Thanks
      W.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9058

Re: c/9058: structure with flexible array member: offsetof() != sizeof()

[Manfred Spraul]

The following reply was made to PR c/9058; it has been noted by GNATS.

From: Manfred Spraul <manfred@colorfullife.com>
To: "Joseph S. Myers" <jsm28@cam.ac.uk>
Cc: gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org
Subject: Re: c/9058: structure with flexible array member: offsetof() != sizeof()
Date: Fri, 27 Dec 2002 00:37:54 +0100

 Joseph S. Myers wrote:
 
 >On Thu, 26 Dec 2002, Manfred Spraul wrote:
 >
 >  
 >
 >>Thanks, I found the thread "C99 conformance bug in gcc-3.1" after filing 
 >>the bug report.
 >>What's the recommended approach to calculate the size for malloc calls? Is
 >>
 >>    struct a {int a; char b; short c[]; };
 >>
 >>    len = offsetof(struct a, c[nr_entries]);
 >>    
 >>
 >
 >You mean offsetof(struct a, c) + nr_entries * sizeof(short).
 >  
 >
 No, I meant what I wrote - is that valid? It would be less error prone, 
 e.g. if the type of the flexible array member is changed.
 linux kernel, i.e gcc portability across platforms.
 
 --
     Manfred
 

Re: c/9058: structure with flexible array member: offsetof() != sizeof()

[Joseph S. Myers]

The following reply was made to PR c/9058; it has been noted by GNATS.

From: "Joseph S. Myers" <jsm28@cam.ac.uk>
To: Manfred Spraul <manfred@colorfullife.com>
Cc: <gcc-gnats@gcc.gnu.org>,  <gcc-bugs@gcc.gnu.org>
Subject: Re: c/9058: structure with flexible array member: offsetof() !=
 sizeof()
Date: Thu, 26 Dec 2002 22:24:11 +0000 (GMT)

 On Thu, 26 Dec 2002, Manfred Spraul wrote:
 
 > Thanks, I found the thread "C99 conformance bug in gcc-3.1" after filing 
 > the bug report.
 > What's the recommended approach to calculate the size for malloc calls? Is
 > 
 >     struct a {int a; char b; short c[]; };
 > 
 >     len = offsetof(struct a, c[nr_entries]);
 
 You mean offsetof(struct a, c) + nr_entries * sizeof(short).
 
 > 
 > valid and portable?
 > 
 >     sizeof(struct a)+nr_entries*sizeof(short);
 > 
 > would waste 2 bytes.
 
 Those would be the bytes of padding (that might or might not nominally be
 needed, depending on the size of the array).  I don't believe their
 absence can cause problems unless you directly assign structures for which
 less memory than the size of the original structure has been allocated (in
 which case you'll lose the flexible array members anyway) but you'll need
 to study the detailed wording of the standard (as proposed to be amended)  
 to work out what is guaranteed portable.
 
 -- 
 Joseph S. Myers
 jsm28@cam.ac.uk
 

Re: c/9058: structure with flexible array member: offsetof() != sizeof()

[Manfred Spraul]

The following reply was made to PR c/9058; it has been noted by GNATS.

From: Manfred Spraul <manfred@colorfullife.com>
To: "Joseph S. Myers" <jsm28@cam.ac.uk>
Cc: gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org
Subject: Re: c/9058: structure with flexible array member: offsetof() != sizeof()
Date: Thu, 26 Dec 2002 22:45:04 +0100

 Joseph S. Myers wrote:
 
 >On 26 Dec 2002 manfred@colorfullife.com wrote:
 >
 >  
 >
 >>Now sizeof(struct x) is 8, offsetof(struct x, c) is 6. That seems to
 >>contradice the C99 standard: section 6.7.2.1, constraint 16.
 >>    
 >>
 >
 >Please read the suggested TC in response to DR 282.  (See N983
 ><http://std.dkuug.dk/JTC1/SC22/WG14/www/docs/n983.htm> and the Santa Cruz
 >minutes <http://std.dkuug.dk/JTC1/SC22/WG14/www/docs/n987.htm> - the DR
 >log hasn't yet been updated.)
 >
 Thanks, I found the thread "C99 conformance bug in gcc-3.1" after filing 
 the bug report.
 What's the recommended approach to calculate the size for malloc calls? Is
 
     struct a {int a; char b; short c[]; };
 
     len = offsetof(struct a, c[nr_entries]);
 
 valid and portable?
 
     sizeof(struct a)+nr_entries*sizeof(short);
 
 would waste 2 bytes.
 
 --
     Manfred
 

Re: c/9058: structure with flexible array member: offsetof() != sizeof()

[Joseph S. Myers]

The following reply was made to PR c/9058; it has been noted by GNATS.

From: "Joseph S. Myers" <jsm28@cam.ac.uk>
To: <manfred@colorfullife.com>
Cc: <gcc-gnats@gcc.gnu.org>,  <gcc-bugs@gcc.gnu.org>
Subject: Re: c/9058: structure with flexible array member: offsetof() !=
 sizeof()
Date: Thu, 26 Dec 2002 20:16:24 +0000 (GMT)

 On 26 Dec 2002 manfred@colorfullife.com wrote:
 
 > Now sizeof(struct x) is 8, offsetof(struct x, c) is 6. That seems to
 > contradice the C99 standard: section 6.7.2.1, constraint 16.
 
 Please read the suggested TC in response to DR 282.  (See N983
 <http://std.dkuug.dk/JTC1/SC22/WG14/www/docs/n983.htm> and the Santa Cruz
 minutes <http://std.dkuug.dk/JTC1/SC22/WG14/www/docs/n987.htm> - the DR
 log hasn't yet been updated.)
 
 -- 
 Joseph S. Myers
 jsm28@cam.ac.uk