Re: 2.12 update

Re: 2.12 update

[H . J . Lu]

> On Sat, 19 Jan 2002, Daniel Jacobowitz wrote:
> > > > So - are there any more issues that you want fixed before release?

Here is one:

http://sources.redhat.com/ml/binutils/2002-01/msg00353.html


H.J.

Re: 2.12 update

[Alan Modra]

On Sat, Jan 19, 2002 at 11:18:42AM -0800, H . J . Lu wrote:
> 
> http://sources.redhat.com/ml/binutils/2002-01/msg00353.html

Please commit.

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre

2.12 update

[Daniel Jacobowitz]

At the moment, I've got most of the want-fixed-by-next-release bugs
checked off as fixed - there's the HP/UX libtool situation, for which
Steve will hopefully have a patch next week; an SH bug report; one
linking bug which I'm not sure is a bug; and some testsuite failures
that should be cleaned up soonish.

So - are there any more issues that you want fixed before release? 
Otherwise, I'd ask that we slow active development on the trunk for a
week or two, in preparation for branching.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

Re: 2.12 update

[Hans-Peter Nilsson]

On Sat, 19 Jan 2002, Daniel Jacobowitz wrote:
> So - are there any more issues that you want fixed before release?

I'd like to:

- Finish on-demand-linker-allocated-global-registers for MMIX (a
  day or two; more testing).
- Autoconf away egcs-1.1.2 for 64-bit bfd and a i386-type host
  (because long long has too many bugs there).
- Work around a gcc-2.95-series bug with tail recursion in
  objdump.
- Fix all binutils subdirs so "make install" implies "make
  install-info" as expected (not to mention GNU standards).

But I don't have more than that on my "things to implement"
list, I think.

brgds, H-P

Re: 2.12 update

[Daniel Jacobowitz]

On Sat, Jan 19, 2002 at 01:53:19AM -0500, Hans-Peter Nilsson wrote:
> On Sat, 19 Jan 2002, Daniel Jacobowitz wrote:
> > So - are there any more issues that you want fixed before release?
> 
> I'd like to:
> 
> - Finish on-demand-linker-allocated-global-registers for MMIX (a
>   day or two; more testing).
> - Autoconf away egcs-1.1.2 for 64-bit bfd and a i386-type host
>   (because long long has too many bugs there).
> - Work around a gcc-2.95-series bug with tail recursion in
>   objdump.
> - Fix all binutils subdirs so "make install" implies "make
>   install-info" as expected (not to mention GNU standards).
> 
> But I don't have more than that on my "things to implement"
> list, I think.

These all sound good.  And we don't even need to hold up the branch for
them :)

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

Re: 2.12 update

[Hans-Peter Nilsson]

On Sat, 19 Jan 2002, Daniel Jacobowitz wrote:

> At the moment, I've got most of the want-fixed-by-next-release bugs
> checked off as fixed - there's the HP/UX libtool situation, for which
> Steve will hopefully have a patch next week; an SH bug report; one
> linking bug which I'm not sure is a bug; and some testsuite failures
> that should be cleaned up soonish.
>
> So - are there any more issues that you want fixed before release?

Two more:
- Move "relax" from ld.h:args_type to struct bfd_link_info.
- Add PTR target_data to struct bfd_link_info.

These and the previous (except the MMIX stuff) issues as
previously mentioned here.

brgds, H-P

Re: 2.12 update

[Daniel Jacobowitz]

On Sat, Jan 19, 2002 at 02:13:53AM -0500, Hans-Peter Nilsson wrote:
> 
> 
> On Sat, 19 Jan 2002, Daniel Jacobowitz wrote:
> 
> > At the moment, I've got most of the want-fixed-by-next-release bugs
> > checked off as fixed - there's the HP/UX libtool situation, for which
> > Steve will hopefully have a patch next week; an SH bug report; one
> > linking bug which I'm not sure is a bug; and some testsuite failures
> > that should be cleaned up soonish.
> >
> > So - are there any more issues that you want fixed before release?
> 
> Two more:
> - Move "relax" from ld.h:args_type to struct bfd_link_info.
> - Add PTR target_data to struct bfd_link_info.
> 
> These and the previous (except the MMIX stuff) issues as
> previously mentioned here.

These two I'm not so sure about.  Do they solve problems?

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

Re: 2.12 update

[Hans-Peter Nilsson]

On Sat, 19 Jan 2002, Daniel Jacobowitz wrote:
> > > So - are there any more issues that you want fixed before release?
> On Sat, Jan 19, 2002 at 02:13:53AM -0500, Hans-Peter Nilsson wrote:
> > - Move "relax" from ld.h:args_type to struct bfd_link_info.
> > - Add PTR target_data to struct bfd_link_info.
 than
> These two I'm not so sure about.  Do they solve problems?

The latter is needed for the MMIX stuff I mentioned, or else I
have to continue to use the coff-specific (so it seems; not
documented) base_file member in struct bfd_link_info.  Having
"relax" in bfd_link_info isn't really necessary, but the
relaxation machinery is more a bfd thing than a ld thing, so
it's definitely more convenient to have it available in bfd.
As I mentioned for example, a target can set it (only) when
necessary, like when some presumably costly relaxation only
needs to be done if certain relocs are present.  So yes they
solve problems, but it is possible to handle them other ways and
I think they're not worth holding up a release for.  But that's
not really what you asked; you asked for "issues [I] want fixed
before the release".  That's what they are, nothing more or
less.  FWIW, Nick though those changes would be positive.

brgds, H-P

Re: 2.12 update

[Andreas Jaeger]

I still see the following on x86-64:
/opt/x86-64/x86_64-unknown-linux/bin/ld: tclExecute.o: access beyond end of merged section (-4 + 0)

But haven't tracked it down - and expect it's - as HJ mentioned - a
generic bug.

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj

Re: 2.12 update

[H . J . Lu]

On Sun, Jan 20, 2002 at 11:28:51AM +0100, Andreas Jaeger wrote:
> 
> I still see the following on x86-64:
> /opt/x86-64/x86_64-unknown-linux/bin/ld: tclExecute.o: access beyond end of merged section (-4 + 0)
> 
> But haven't tracked it down - and expect it's - as HJ mentioned - a
> generic bug.

Can you duplicate it on x86?


H.J.

Re: 2.12 update

[Andreas Jaeger]

"H . J . Lu" <hjl@lucon.org> writes:

> On Sun, Jan 20, 2002 at 11:28:51AM +0100, Andreas Jaeger wrote:
>> 
>> I still see the following on x86-64:
>> /opt/x86-64/x86_64-unknown-linux/bin/ld: tclExecute.o: access beyond end of merged section (-4 + 0)
>> 
>> But haven't tracked it down - and expect it's - as HJ mentioned - a
>> generic bug.
>
> Can you duplicate it on x86?

No :-(

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj

Re: 2.12 update

[H . J . Lu]

On Mon, Jan 21, 2002 at 02:56:51PM +0100, Andreas Jaeger wrote:
> "H . J . Lu" <hjl@lucon.org> writes:
> 
> > On Sun, Jan 20, 2002 at 11:28:51AM +0100, Andreas Jaeger wrote:
> >> 
> >> I still see the following on x86-64:
> >> /opt/x86-64/x86_64-unknown-linux/bin/ld: tclExecute.o: access beyond end of merged section (-4 + 0)
> >> 
> >> But haven't tracked it down - and expect it's - as HJ mentioned - a
> >> generic bug.
> >
> > Can you duplicate it on x86?

Than that may be an x86-64 specific bug.


H.J.

Re: 2.12 update

[Christian Jönsson]

On Sat, Jan 19, 2002 at 12:26:45AM -0500, Daniel Jacobowitz wrote:
> 
> So - are there any more issues that you want fixed before release? 
> 

Well, I'm not sure this is important, I'm new to this, but what about
the test results I sent in the other day, i.e., 

http://sources.redhat.com/ml/binutils/2002-01/msg00351.html



Native configuration is sparc-unknown-linux-gnu

		=== binutils tests ===


Running target unix

		=== binutils Summary ===

# of expected passes		31
# of expected failures		1
		=== gas tests ===


Running target unix

		=== gas Summary ===

# of expected passes		91

		=== ld tests ===


Running target unix
FAIL: visibility (hidden_weak) (non PIC)
FAIL: visibility (hidden_weak) (non PIC, load offset)
FAIL: visibility (hidden_weak) (PIC main, non PIC so)
FAIL: visibility (protected_weak) (non PIC)
FAIL: visibility (protected_weak) (non PIC, load offset)
FAIL: visibility (protected_weak) (PIC main, non PIC so)

		=== ld Summary ===

# of expected passes		121
# of unexpected failures	6
# of expected failures		13


and the (binutils) warnings

Counting all warnings,
there are 158 warnings in stage0 of this bootstrap.

Number of warnings per file:
     48	libiberty/md5.c
     38	libiberty/regex.c
     16	../../binutils/unwind-ia64.c
      5	libjava/java/lang/natClassLoader.cc
      5	../../gprof/vax.c
      5	../../gprof/tahoe.c
      5	../../gprof/i386.c
      4	libjava/java/lang/strtod.c
      4	libjava/java/lang/dtoa.c
      4	../../gprof/sparc.c
      4	../../gprof/alpha.c
      3	libjava/java/lang/e_exp.c
      2	include/getopt.h
      1	libjava/name-finder.cc
      1	libjava/java/lang/e_rem_pio2.c
      1	libjava/java/lang/e_log.c
      1	libjava/java/lang/e_asin.c
      1	libjava/gnu/gcj/runtime/natSharedLibLoader.cc
      1	libjava/gnu/gcj/io/shs.cc
      1	libjava/gnu/gcj/convert/natIconv.cc
      1	libf2c/libI77/open.c
      1	lex.yy.c
      1	include/xregex2.h
      1	include/sys/cdefs.h
      1	config.h
      1	../../ld/lexsup.c
      1	../../ld/ldmain.c
      1	../../../boehm-gc/mark_rts.c

Number of warning types:
     32	traditional C rejects string concatenation
     16	initialization from incompatible pointer type
     16	function-like macro \`FI' must be used with arguments in traditional C
     16	function-like macro \`FH' must be used with arguments in traditional C
     16	function-like macro \`FG' must be used with arguments in traditional C
     16	cast from pointer to integer of different size
     13	\`???' might be used uninitialized in this function
      7	cast to pointer from integer of different size
      5	unused parameter \`
      4	signed and unsigned type in conditional expression
      4	function declaration isn't a prototype
      2	unused parameter \`???'
      2	this is the location of the previous definition
      1	unused 
      1	function returns address of local variable
      1	division: function singularity
      1	comparison between 
      1	cast from pointer to integer of 
      1	\`__restrict_arr' redefined
      1	\`_XOPEN_SOURCE' redefined
      1	\`???' defined but not used
      1	\`???' declared 

Are they what to expect? Could I do something to "report" more
on this?

Cheers,

/ChJ