When I recenlty updated to Valgrind 2.1.1 and simultanously updated my gl=
ibc=20
to the newest stuff (devel-packages) offically available for Mandrake 9.0=
, a=20
set of errors was found inside glibc. So, here is the set of suppresions =
for=20
those errors. If have only doubdts for strchr suppresions, if it's ok to =
do=20
an suppresion with so shallow stack trace...
rgds
Sebastian Kaliszewski
{
elf_dynamic_do_rel.4
Memcheck:Cond
fun:elf_dynamic_do_rel.4
}
{
elf_dynamic_do_rel.4/2
Memcheck:Cond
fun:elf_dynamic_do_rel.4
}
{
strchr
Memcheck:Cond
fun:strchr
}
{
elf_dynamic_do_rel.4
Memcheck:Cond
fun:elf_dynamic_do_rel.4
fun:_dl_relocate_object
fun:dl_main
fun:_dl_sysdep_start
}
{
elf_dynamic_do_rel.4
Memcheck:Cond
fun:elf_dynamic_do_rel.4
fun:_dl_relocate_object
fun:dl_main
fun:_dl_sysdep_start
}
{
static_initialisation_.../std_string_leak
Addrcheck,Memcheck:Leak
fun:_Znwj
fun:_ZNSt24__default_alloc_templateILb1ELi0EE14_S_chunk_allocEjRi
fun:_ZNSt24__default_alloc_templateILb1ELi0EE9_S_refillEj
fun:_ZNSs4_Rep9_S_createEjRKSaIcE
}
|
|
From: Nicholas N. <nj...@ca...> - 2004-04-26 08:21:33
|
On Sun, 25 Apr 2004, Sebastian Kaliszewski wrote:
> When I recenlty updated to Valgrind 2.1.1 and simultanously updated my glibc
> to the newest stuff (devel-packages) offically available for Mandrake 9.0, a
> set of errors was found inside glibc. So, here is the set of suppresions for
> those errors. If have only doubdts for strchr suppresions, if it's ok to do
> an suppresion with so shallow stack trace...
>
> {
> elf_dynamic_do_rel.4
> Memcheck:Cond
> fun:elf_dynamic_do_rel.4
> }
> {
> elf_dynamic_do_rel.4/2
> Memcheck:Cond
> fun:elf_dynamic_do_rel.4
> }
> {
> strchr
> Memcheck:Cond
> fun:strchr
> }
> {
> elf_dynamic_do_rel.4
> Memcheck:Cond
> fun:elf_dynamic_do_rel.4
> fun:_dl_relocate_object
> fun:dl_main
> fun:_dl_sysdep_start
> }
> {
> elf_dynamic_do_rel.4
> Memcheck:Cond
> fun:elf_dynamic_do_rel.4
> fun:_dl_relocate_object
> fun:dl_main
> fun:_dl_sysdep_start
> }
> {
> static_initialisation_.../std_string_leak
> Addrcheck,Memcheck:Leak
> fun:_Znwj
> fun:_ZNSt24__default_alloc_templateILb1ELi0EE14_S_chunk_allocEjRi
> fun:_ZNSt24__default_alloc_templateILb1ELi0EE9_S_refillEj
> fun:_ZNSs4_Rep9_S_createEjRKSaIcE
> }
The first two are identical, as are 4 & 5. And the first two will
suppress a superset of those suppressed by 4 & 5 -- are just 4 & 5 enough?
You're right about the strchr stack trace, that suppression will suppress
any 'Cond' error that occurs in strchr, which is a bit broad.
I suggest adding the last two to glibc-2.3.supp. How does that sound?
N
|
Dnia pon 26. kwiecie=F1 2004 10:21, Nicholas Nethercote napisa=B3:
> On Sun, 25 Apr 2004, Sebastian Kaliszewski wrote:
> > When I recenlty updated to Valgrind 2.1.1 and simultanously updated m=
y
> > glibc to the newest stuff (devel-packages) offically available for
> > Mandrake 9.0, a set of errors was found inside glibc. So, here is the=
set
> > of suppresions for those errors. If have only doubdts for strchr
> > suppresions, if it's ok to do an suppresion with so shallow stack
> > trace...
> >
> > {
> > elf_dynamic_do_rel.4
> > Memcheck:Cond
> > fun:elf_dynamic_do_rel.4
> > }
> > {
> > elf_dynamic_do_rel.4/2
> > Memcheck:Cond
> > fun:elf_dynamic_do_rel.4
> > }
> > {
> > strchr
> > Memcheck:Cond
> > fun:strchr
> > }
> > {
> > elf_dynamic_do_rel.4
> > Memcheck:Cond
> > fun:elf_dynamic_do_rel.4
> > fun:_dl_relocate_object
> > fun:dl_main
> > fun:_dl_sysdep_start
> > }
> > {
> > elf_dynamic_do_rel.4
> > Memcheck:Cond
> > fun:elf_dynamic_do_rel.4
> > fun:_dl_relocate_object
> > fun:dl_main
> > fun:_dl_sysdep_start
> > }
> > {
> > static_initialisation_.../std_string_leak
> > Addrcheck,Memcheck:Leak
> > fun:_Znwj
> > fun:_ZNSt24__default_alloc_templateILb1ELi0EE14_S_chunk_allocEjRi
> > fun:_ZNSt24__default_alloc_templateILb1ELi0EE9_S_refillEj
> > fun:_ZNSs4_Rep9_S_createEjRKSaIcE
> > }
>
> The first two are identical, as are 4 & 5. And the first two will
> suppress a superset of those suppressed by 4 & 5 -- are just 4 & 5 enou=
gh?
Unfortunately no. But i think using the first one is not that big of a=20
problem, as elf_dynamic_do_rel.4 is rather initialisation only function.
> You're right about the strchr stack trace, that suppression will suppre=
ss
> any 'Cond' error that occurs in strchr, which is a bit broad.
Too bad...
Maybe it'd be nice to add an extension which would allow to suppres such=20
shallow stacktrace errors without affectin a deep trace ones. For example=
:
{
strchr
Memcheck:Cond
fun:strchr
none:
}
>
> I suggest adding the last two to glibc-2.3.supp. How does that sound?
>
Mandrake 9.0 i glibc 2.2.5 based (so glibc-2.2.supp) , and I'd rather add=
=20
first & last OK. Still that stupid strchr remains.
rgds
Sebastian
--=20
"Never undersetimate the power of human stupidity" -- from notebooks of L=
=2EL.
|