Testsuite failure in check-localplt

Testsuite failure in check-localplt

[Andreas Jaeger]

[-- Attachment #1: Type: text/plain, Size: 686 bytes --]



On x86-64 I get the following testsuite failure for elf/check-localplt.out:

--- ../scripts/data/localplt-generic.data       2007-08-07 07:35:58.000000000 +0200
+++ -   2007-10-08 16:14:22.828168685 +0200
@@ -2,5 +2,7 @@
 libc.so: free
 libc.so: malloc
 libc.so: memalign
+libc.so: read
 libc.so: realloc
+libc.so: write
 libm.so: matherr

Should we adjust the testsuite or is there something else to change?

Andreas
-- 
 Andreas Jaeger, Director Platform/openSUSE, aj@suse.de
  SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
   Maxfeldstr. 5, 90409 Nürnberg, Germany
    GPG fingerprint = 93A3 365E CE47 B889 DF7F  FED1 389A 563C C272 A126

[-- Attachment #2: Type: application/pgp-signature, Size: 188 bytes --]

Re: Testsuite failure in check-localplt

[Roland McGrath]

Look for the relocs in libc.so that refer to those symbols.  Probably they
should be calling __read, which already has libc_hidden_proto.  If they are
in code never reached by ISO C-only, then they could legitimately stick to
read for aesthetic purposes if one really cared, but then you'd need to add
libc_hidden_proto for read in include/unistd.h, which we haven't needed yet.
(If they are in low-level linux-specific code, they might want to be using
INLINE_SYSCALL anyway.)


Thanks,
Roland

Re: Testsuite failure in check-localplt

[Andreas Jaeger]

[-- Attachment #1: Type: text/plain, Size: 963 bytes --]

Roland McGrath <roland@redhat.com> writes:

> Look for the relocs in libc.so that refer to those symbols.  Probably they
> should be calling __read, which already has libc_hidden_proto.  If they are
> in code never reached by ISO C-only, then they could legitimately stick to
> read for aesthetic purposes if one really cared, but then you'd need to add
> libc_hidden_proto for read in include/unistd.h, which we haven't needed yet.
> (If they are in low-level linux-specific code, they might want to be using
> INLINE_SYSCALL anyway.)

Thanks - that helped.  It was in eventfd_read and write - and I just
noticed that Uli already committed the obvious fix.

The testsuite passes now for me,
Andreas
-- 
 Andreas Jaeger, Director Platform/openSUSE, aj@suse.de
  SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
   Maxfeldstr. 5, 90409 Nürnberg, Germany
    GPG fingerprint = 93A3 365E CE47 B889 DF7F  FED1 389A 563C C272 A126

[-- Attachment #2: Type: application/pgp-signature, Size: 188 bytes --]