valgrind-users

[Valgrind-users] uninitialized value error with packed structure

[David L <id...@gm...>]

With this code:


#include <stdio.h>

class MyClass {
public:
  short int foo_ : 6;
};

int main(int argc, char *argv[]) {
  MyClass *gce = new MyClass();
  double d=0.0;
  if (d == 0.0) {
    printf("d equals zero.\n");
  }
  return 0;
}

I get this output from valgrind:


==32293== Conditional jump or move depends on uninitialised value(s)
==32293==    at 0x804862F: main (main.cpp:11)
==32293==  Uninitialised value was created by a heap allocation
==32293==    at 0x400674E: operator new(unsigned int) (vg_replace_malloc.c:224)
==32293==    by 0x8048610: main (main.cpp:9)

If I remove the : 6 from foo_, I don't get the error.  Is this a valgrind bug?

Thanks,

              David

Re: [Valgrind-users] uninitialized value error with packed structure

[Julian S. <js...@ac...>]

On Thursday 14 May 2009, David L wrote:
> With this code:
>
>
> #include <stdio.h>
>
> class MyClass {
> public:
>   short int foo_ : 6;
> };
>
> int main(int argc, char *argv[]) {
>   MyClass *gce = new MyClass();
>   double d=0.0;
>   if (d == 0.0) {
>     printf("d equals zero.\n");
>   }
>   return 0;
> }
>
> I get this output from valgrind:
>
>
> ==32293== Conditional jump or move depends on uninitialised value(s)
> ==32293==    at 0x804862F: main (main.cpp:11)
> ==32293==  Uninitialised value was created by a heap allocation
> ==32293==    at 0x400674E: operator new(unsigned int)
> (vg_replace_malloc.c:224) ==32293==    by 0x8048610: main (main.cpp:9)
>
> If I remove the : 6 from foo_, I don't get the error.  Is this a valgrind
> bug?

Hmm, that's ungood.  Can you send details of the precise gcc version,
the compilation flags, and the what kind of processor you're running on?

J

Re: [Valgrind-users] uninitialized value error with packed structure

[Julian S. <js...@ac...>]

On Thursday 14 May 2009, Julian Seward wrote:

> Hmm, that's ungood.  Can you send details of the precise gcc version,
> the compilation flags, and the what kind of processor you're running on?

Oh, and of course, the Valgrind version too.

J

Re: [Valgrind-users] uninitialized value error with packed structure

[David L <id...@gm...>]

On Thu, May 14, 2009 at 12:33 AM, Julian Seward wrote:
>
>> Hmm, that's ungood.  Can you send details of the precise gcc version,
>> the compilation flags, and the what kind of processor you're running on?

I'm running on a fedora 10 x86 box:


valgrind --version
valgrind-3.4.1


g++ --version
g++ (GCC) 4.3.2 20081105 (Red Hat 4.3.2-7)

g++ -g3 -o foo main.cpp


valgrind --track-origins=yes foo
==3121== Memcheck, a memory error detector.
==3121== Copyright (C) 2002-2008, and GNU GPL'd, by Julian Seward et al.
==3121== Using LibVEX rev 1884, a library for dynamic binary translation.
==3121== Copyright (C) 2004-2008, and GNU GPL'd, by OpenWorks LLP.
==3121== Using valgrind-3.4.1, a dynamic binary instrumentation framework.
==3121== Copyright (C) 2000-2008, and GNU GPL'd, by Julian Seward et al.
==3121== For more details, rerun with: -v
==3121==
==3121== Conditional jump or move depends on uninitialised value(s)
==3121==    at 0x804852F: main (main.cpp:11)
==3121==  Uninitialised value was created by a heap allocation
==3121==    at 0x400674E: operator new(unsigned int) (vg_replace_malloc.c:224)
==3121==    by 0x8048510: main (main.cpp:9)
d equals zero.
myclass size is 2
==3121==
==3121== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 16 from 1)
==3121== malloc/free: in use at exit: 2 bytes in 1 blocks.
==3121== malloc/free: 1 allocs, 0 frees, 2 bytes allocated.
==3121== For counts of detected errors, rerun with: -v
==3121== searching for pointers to 1 not-freed blocks.
==3121== checked 89,756 bytes.
==3121==
==3121== LEAK SUMMARY:
==3121==    definitely lost: 2 bytes in 1 blocks.
==3121==      possibly lost: 0 bytes in 0 blocks.
==3121==    still reachable: 0 bytes in 0 blocks.
==3121==         suppressed: 0 bytes in 0 blocks.
==3121== Rerun with --leak-check=full to see details of leaked memory.



cat /proc/cpuinfo
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 15
model		: 3
model name	: Intel(R) Pentium(R) 4 CPU 3.00GHz
stepping	: 4
cpu MHz		: 3000.000
cache size	: 1024 KB
fdiv_bug	: no
hlt_bug		: no
f00f_bug	: no
coma_bug	: no
fpu		: yes
fpu_exception	: yes
cpuid level	: 3
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat
pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe constant_tsc up
pebs bts pni monitor ds_cpl cid xtpr
bogomips	: 5984.98
clflush size	: 64
power management:


Thanks,

                David

Re: [Valgrind-users] uninitialized value error with packed structure

[Tom H. <to...@co...>]

David L wrote:

> ==3121== Conditional jump or move depends on uninitialised value(s)
> ==3121==    at 0x804852F: main (main.cpp:11)
> ==3121==  Uninitialised value was created by a heap allocation
> ==3121==    at 0x400674E: operator new(unsigned int) (vg_replace_malloc.c:224)
> ==3121==    by 0x8048510: main (main.cpp:9)
> d equals zero.
> myclass size is 2

What is printing that last line? The code you gave has nothing in it 
that would print that?

Tom

-- 
Tom Hughes (to...@co...)
http://www.compton.nu/

Re: [Valgrind-users] uninitialized value error with packed structure

[David L <id...@gm...>]

On Thu, May 14, 2009 at 6:36 AM, Tom Hughes <to...@co...> wrote:
> David L wrote:
>
>> ==3121== Conditional jump or move depends on uninitialised value(s)
>> ==3121==    at 0x804852F: main (main.cpp:11)
>> ==3121==  Uninitialised value was created by a heap allocation
>> ==3121==    at 0x400674E: operator new(unsigned int)
>> (vg_replace_malloc.c:224)
>> ==3121==    by 0x8048510: main (main.cpp:9)
>> d equals zero.
>> myclass size is 2
>
> What is printing that last line? The code you gave has nothing in it that
> would print that?

oops... I was trying to simplify the example that I posted, so I removed
that printf from the example code, but I forgot to remove it from the
output.  Here's the actual code:

#include <stdio.h>

class MyClass {
public:
  short int foo_ : 6;
};

int main(int argc, char *argv[]) {
  MyClass *gce = new MyClass();
  double d=0.0;
  if (d == 0.0) {
    printf("d equals zero.\n");
    printf("myclass size is %d\n", sizeof(MyClass));
  }
  return 0;
}

Re: [Valgrind-users] uninitialized value error with packed structure

[Tom H. <to...@co...>]

David L wrote:
> On Thu, May 14, 2009 at 6:36 AM, Tom Hughes <to...@co...> wrote:
>> David L wrote:
>>
>>> ==3121== Conditional jump or move depends on uninitialised value(s)
>>> ==3121==    at 0x804852F: main (main.cpp:11)
>>> ==3121==  Uninitialised value was created by a heap allocation
>>> ==3121==    at 0x400674E: operator new(unsigned int)
>>> (vg_replace_malloc.c:224)
>>> ==3121==    by 0x8048510: main (main.cpp:9)
>>> d equals zero.
>>> myclass size is 2
>> What is printing that last line? The code you gave has nothing in it that
>> would print that?
> 
> oops... I was trying to simplify the example that I posted, so I removed
> that printf from the example code, but I forgot to remove it from the
> output.  Here's the actual code:

Thanks. I don't think it actually makes any difference though. I can 
reproduce what you're seeing on F10 x86 with the original program.

Tom

-- 
Tom Hughes (to...@co...)
http://www.compton.nu/

Re: [Valgrind-users] uninitialized value error with packed structure

[Julian S. <js...@ac...>]

Tom,

> Thanks. I don't think it actually makes any difference though. I can
> reproduce what you're seeing on F10 x86 with the original program.

Could you pls send the objdump -d of the insn it's complaining about
+ the few preceding ones, so as to show how the flags get set?

J

Re: [Valgrind-users] uninitialized value error with packed structure

[Tom H. <to...@co...>]

Julian Seward wrote:

>> Thanks. I don't think it actually makes any difference though. I can
>> reproduce what you're seeing on F10 x86 with the original program.
> 
> Could you pls send the objdump -d of the insn it's complaining about
> + the few preceding ones, so as to show how the flags get set?

Here's the dump of main - the jp at 0x80484ff is the one that valgrind 
is complaining about:

080484c4 <main>:
  80484c4:	8d 4c 24 04          	lea    0x4(%esp),%ecx
  80484c8:	83 e4 f0             	and    $0xfffffff0,%esp
  80484cb:	ff 71 fc             	pushl  -0x4(%ecx)
  80484ce:	55                   	push   %ebp
  80484cf:	89 e5                	mov    %esp,%ebp
  80484d1:	51                   	push   %ecx
  80484d2:	83 ec 24             	sub    $0x24,%esp
  80484d5:	c7 04 24 02 00 00 00 	movl   $0x2,(%esp)
  80484dc:	e8 f7 fe ff ff       	call   80483d8 <_Znwj@plt>
  80484e1:	0f b6 10             	movzbl (%eax),%edx
  80484e4:	83 e2 c0             	and    $0xffffffc0,%edx
  80484e7:	88 10                	mov    %dl,(%eax)
  80484e9:	89 45 ec             	mov    %eax,-0x14(%ebp)
  80484ec:	d9 ee                	fldz
  80484ee:	dd 5d f0             	fstpl  -0x10(%ebp)
  80484f1:	dd 45 f0             	fldl   -0x10(%ebp)
  80484f4:	d9 ee                	fldz
  80484f6:	d9 c9                	fxch   %st(1)
  80484f8:	da e9                	fucompp
  80484fa:	df e0                	fnstsw %ax
  80484fc:	9e                   	sahf
  80484fd:	75 0e                	jne    804850d <main+0x49>
  80484ff:	7a 0c                	jp     804850d <main+0x49>
  8048501:	c7 04 24 e4 85 04 08 	movl   $0x80485e4,(%esp)
  8048508:	e8 db fe ff ff       	call   80483e8 <puts@plt>
  804850d:	b8 00 00 00 00       	mov    $0x0,%eax
  8048512:	83 c4 24             	add    $0x24,%esp
  8048515:	59                   	pop    %ecx
  8048516:	5d                   	pop    %ebp
  8048517:	8d 61 fc             	lea    -0x4(%ecx),%esp
  804851a:	c3                   	ret

Tom

-- 
Tom Hughes (to...@co...)
http://www.compton.nu/