[Bug gold/22011] glibc math/test-matherr.c fails with gold

[ian at airs dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=22011

--- Comment #3 from Ian Lance Taylor <ian at airs dot com> ---
I can't recreate this.  I get "matherr is working" from both gold and GNU ld. 
I tested with tip gold and "Ubuntu EGLIBC 2.19-0ubuntu6.13".


That said, I can see the difference.  With GNU ld, readelf -s --wide run on the
linked executable shows

     8: 0000000000601050     4 OBJECT  WEAK   DEFAULT   26
address@hidden (3)
    10: 00000000004006fd    39 FUNC    GLOBAL DEFAULT   14 address@hidden

With gold it shows

     8: 00000000004006ad    39 FUNC    GLOBAL DEFAULT   13 matherr
     9: 0000000000402040     4 OBJECT  GLOBAL DEFAULT   26
address@hidden (2)

I'm not sure, but it may be that gold is doing this because in my libm.so
matherr shows up as

   358: 0000000000008430     3 FUNC    WEAK   DEFAULT   13 matherr@@GLIBC_2.2.5

which is to say that it is the default version.  And, as mentioned above,
everything works as expected.

What are the readelf -s --wide outputs on your system?

-- 
You are receiving this mail because:
You are on the CC list for the bug.