diff mbox series

S390: Fix elf/tst-audit25[ab]

Message ID 20220407115948.1664232-1-stli@linux.ibm.com
State Committed
Commit 5325233313c66aea13e86f5dd59618e9dd74b510
Headers
Series S390: Fix elf/tst-audit25[ab] |

Checks

Context Check Description
dj/TryBot-apply_patch success Patch applied to master at the time it was sent
dj/TryBot-32bit success Build for i686

Commit Message

Stefan Liebler April 7, 2022, 11:59 a.m. UTC 
  If glibc is configured with --disable-default-pie and build on
s390 with -O3, the tests elf/tst-audit25a and elf/tst-audit25b are
failing as there are additional la_symbind lines for free and malloc.
It turns out that those belong to the executable. In fact those are
the PLT-stubs. Furthermore la_symbind is also called for calloc and
realloc symbols, but those belong to libc.

Those functions are not called at all, but dlsym'ed in
elf/dl-minimal.c:
__rtld_malloc_init_real (struct link_map *main_map)
{
...
  void *new_calloc = lookup_malloc_symbol (main_map, "calloc", &version);
  void *new_free = lookup_malloc_symbol (main_map, "free", &version);
  void *new_malloc = lookup_malloc_symbol (main_map, "malloc", &version);
  void *new_realloc = lookup_malloc_symbol (main_map, "realloc", &version);
...
}

Therefore, this commit just ignored symbols with LA_SYMB_DLSYM flag.
---
 elf/tst-auditmod25.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Comments

Adhemerval Zanella April 7, 2022, 1:36 p.m. UTC | #1 
On 07/04/2022 08:59, Stefan Liebler via Libc-alpha wrote:
> If glibc is configured with --disable-default-pie and build on
> s390 with -O3, the tests elf/tst-audit25a and elf/tst-audit25b are
> failing as there are additional la_symbind lines for free and malloc.
> It turns out that those belong to the executable. In fact those are
> the PLT-stubs. Furthermore la_symbind is also called for calloc and
> realloc symbols, but those belong to libc.
> 
> Those functions are not called at all, but dlsym'ed in
> elf/dl-minimal.c:
> __rtld_malloc_init_real (struct link_map *main_map)
> {
> ...
>   void *new_calloc = lookup_malloc_symbol (main_map, "calloc", &version);
>   void *new_free = lookup_malloc_symbol (main_map, "free", &version);
>   void *new_malloc = lookup_malloc_symbol (main_map, "malloc", &version);
>   void *new_realloc = lookup_malloc_symbol (main_map, "realloc", &version);
> ...
> }
> 
> Therefore, this commit just ignored symbols with LA_SYMB_DLSYM flag.

LGTM, thank.  Why don't we see in other configuration and/or architecture?
What s390 is doing different here?

Reviewed-by: Adheemrval Zanella  <adhemerval.zanella@linaro.org>

> ---
>  elf/tst-auditmod25.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/elf/tst-auditmod25.c b/elf/tst-auditmod25.c
> index 20640a8daf..0524c5aab1 100644
> --- a/elf/tst-auditmod25.c
> +++ b/elf/tst-auditmod25.c
> @@ -72,7 +72,8 @@ la_symbind32 (Elf32_Sym *sym, unsigned int ndx,
>  	      unsigned int *flags, const char *symname)
>  #endif
>  {
> -  if (*refcook != -1 && *defcook != -1 && symname[0] != '\0')
> +  if (*refcook != -1 && *defcook != -1 && symname[0] != '\0'
> +      && (*flags & LA_SYMB_DLSYM) == 0)
>      fprintf (stderr, "la_symbind: %s %u\n", symname,
>  	     *flags & (LA_SYMB_NOPLTENTER | LA_SYMB_NOPLTEXIT) ? 1 : 0);
>    return sym->st_value;
Stefan Liebler April 7, 2022, 1:51 p.m. UTC | #2 
On 07/04/2022 15:36, Adhemerval Zanella via Libc-alpha wrote:
> 
> 
> On 07/04/2022 08:59, Stefan Liebler via Libc-alpha wrote:
>> If glibc is configured with --disable-default-pie and build on
>> s390 with -O3, the tests elf/tst-audit25a and elf/tst-audit25b are
>> failing as there are additional la_symbind lines for free and malloc.
>> It turns out that those belong to the executable. In fact those are
>> the PLT-stubs. Furthermore la_symbind is also called for calloc and
>> realloc symbols, but those belong to libc.
>>
>> Those functions are not called at all, but dlsym'ed in
>> elf/dl-minimal.c:
>> __rtld_malloc_init_real (struct link_map *main_map)
>> {
>> ...
>>   void *new_calloc = lookup_malloc_symbol (main_map, "calloc", &version);
>>   void *new_free = lookup_malloc_symbol (main_map, "free", &version);
>>   void *new_malloc = lookup_malloc_symbol (main_map, "malloc", &version);
>>   void *new_realloc = lookup_malloc_symbol (main_map, "realloc", &version);
>> ...
>> }
>>
>> Therefore, this commit just ignored symbols with LA_SYMB_DLSYM flag.
> 
> LGTM, thank.  Why don't we see in other configuration and/or architecture?
> What s390 is doing different here?
> 
> Reviewed-by: Adheemrval Zanella  <adhemerval.zanella@linaro.org>
> 

Thanks.
Then I'll commit the patch.

You are right I have not seen it on x86_64. Perhaps it triggers
something in s390x-code in ld regarding -Bsymbolic and
pointer-equality-reasons. But this is only a guess.

Thanks,
Stefan
diff mbox series

Patch

diff --git a/elf/tst-auditmod25.c b/elf/tst-auditmod25.c
index 20640a8daf..0524c5aab1 100644
--- a/elf/tst-auditmod25.c
+++ b/elf/tst-auditmod25.c
@@ -72,7 +72,8 @@  la_symbind32 (Elf32_Sym *sym, unsigned int ndx,
 	      unsigned int *flags, const char *symname)
 #endif
 {
-  if (*refcook != -1 && *defcook != -1 && symname[0] != '\0')
+  if (*refcook != -1 && *defcook != -1 && symname[0] != '\0'
+      && (*flags & LA_SYMB_DLSYM) == 0)
     fprintf (stderr, "la_symbind: %s %u\n", symname,
 	     *flags & (LA_SYMB_NOPLTENTER | LA_SYMB_NOPLTEXIT) ? 1 : 0);
   return sym->st_value;