[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Bug gas/1879] FAIL: weakref tests

From: dave at hiauly1 dot hia dot nrc dot ca
Subject: [Bug gas/1879] FAIL: weakref tests
Date: 19 Nov 2005 16:50:47 -0000

------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  
2005-11-19 16:50 -------
Subject: Re:  FAIL: weakref tests

> Hey, it's the same FAILs as for mmix-knuth-mmixware!
> Ever since rth's change from ".set" to "=" for these tests.
> I scratched the "host" and "build" fields, as this was a
> cross from i686-pc-linux-gnu (FC2).

1) On hppa, the operation needs to be indented.  This allows
   weakref1.s to assemble.  It is interesting that this seems to
   work.  Before I saw this, I thought it was impossible to alias
   a symbol under hpux.  The HP assembler doesn't allow it, and
   .equ and .set don't work with GAS.  Never tried .equiv.

2) With some tweaking of the .d files, I can get "relocations",
   "global syms" and "local syms" to pass on hppa2.0w-hp-hpux11.11.
   However, I don't have solution for "strong undefined syms"
   or "weak undefined syms".

I see in dump.o:

         U wa1
         U wb1
         U wc1
         U wh2
         U wh3
         U wh4
         U wh5
         U wh6
         U wh7
         U wm6
         U wm7
         U wm8
         U ww1
         U ww10
         U ww2
         U ww3
         U ww4
         U ww5
         U ww6
         U ww7
         U ww8
         U ww9

With odump I see for wa1:

00000000 00000000 Data  Unsat  0 ......S...  3 00000 wa1

The 'S' indicates a "weak" secondary definition.  On hppa-linux,
we have:

         w wa1
         w wb1
         w wc1
         w wh2

So, the symbols are weak and local.  I think that this is a
problem under hpux as secondary definitions always have have
univeral (global) or unsatisfied scope.  I'm not sure why it's
necessary to output weak locals.  There's no mention of such symbols
in the SYSV ABI as far as I know.

I need to find a solution as to how to handle weak references under
hpux as their introduction is also causing problems in GCC.




------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]