arrayidx.exp failures

arrayidx.exp failures

[Nick Roberts]

In gdb/testsuite/ChangeLog:

2006-10-06  Joel Brobecker  <brobecker@adacore.com>

	* gdb.ada/arrayidx.exp: Remove calls to setup_kfail now that
	the corresponding bug has been fixed in GDB (PR/2018).

They still fail for me with:

GNU gdb 6.5.50.20061013-cvs
This GDB was configured as "i686-pc-linux-gnu".

I can't see any mention of PR/2018 in gdb/ChangeLog.  Has this fix actually
been committed?


-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: arrayidx.exp failures

[Joel Brobecker]

> 2006-10-06  Joel Brobecker  <brobecker@adacore.com>
> 
> 	* gdb.ada/arrayidx.exp: Remove calls to setup_kfail now that
> 	the corresponding bug has been fixed in GDB (PR/2018).
> 
> They still fail for me with:
> 
> GNU gdb 6.5.50.20061013-cvs
> This GDB was configured as "i686-pc-linux-gnu".
> 
> I can't see any mention of PR/2018 in gdb/ChangeLog.  Has this fix actually
> been committed?

Yes: 

2006-10-06 Joel Brobecker  <brobecker@adacore.com>

        * dwarf2read.c (read_array_type): Set the type name if the name
        attribute is present.

I will add a mention of the PR in the changelog, that's a good remark.

In terms of why it fails for you, it is probably because of something
else. I will guess that you are missing some parts of the encodings
necessary to complete the decoding of the array. The encoding is
documented in ada/exp_dbug.ads of GCC. So I'm guessing a malfunction
of GCC rather than a possible bug in GDB.

We're pushing all the changes we made for 41 in that department as fast
as we can so that everyone can enjoy the most of Ada. But long term,
we'll be moving towards pure dwarf data.

-- 
Joel

Re: arrayidx.exp failures

[Daniel Jacobowitz]

On Thu, Oct 12, 2006 at 07:38:39PM -0700, Joel Brobecker wrote:
> In terms of why it fails for you, it is probably because of something
> else. I will guess that you are missing some parts of the encodings
> necessary to complete the decoding of the array. The encoding is
> documented in ada/exp_dbug.ads of GCC. So I'm guessing a malfunction
> of GCC rather than a possible bug in GDB.

FYI, Joel, I have no idea what compilers you test with - but half of
the Ada tests in the GDB testsuite have never passed for me with FSF
gnat, at least 4.0 and 4.1, and possibly 3.4 too.

FAIL: gdb.ada/arrayidx.exp: print e_one_two_three, indexes off
FAIL: gdb.ada/arrayidx.exp: print r_two_three, indexes off
FAIL: gdb.ada/arrayidx.exp: print u_one_two_three, indexes off
FAIL: gdb.ada/arrayidx.exp: print p_one_two_three, indexes off
FAIL: gdb.ada/arrayidx.exp: print e_one_two_three
FAIL: gdb.ada/arrayidx.exp: print r_two_three
FAIL: gdb.ada/arrayidx.exp: print u_one_two_three
FAIL: gdb.ada/arrayidx.exp: print p_one_two_three
FAIL: gdb.ada/null_record.exp: ptype on null record
FAIL: gdb.ada/packed_array.exp: print var

print e_one_two_three
$2 = (0 => 1, 2, 3)
(gdb) FAIL: gdb.ada/arrayidx.exp: print e_one_two_three, indexes off
print r_two_three
$3 = (2, 3)
(gdb) FAIL: gdb.ada/arrayidx.exp: print r_two_three, indexes off
print u_one_two_three
No definition of "u_one_two_three" in current context.
(gdb) FAIL: gdb.ada/arrayidx.exp: print u_one_two_three, indexes off
print p_one_two_three
$4 = warning: could not find bounds information on packed array
can't unpack array

(gdb) FAIL: gdb.ada/arrayidx.exp: print p_one_two_three, indexes off

print e_one_two_three
$9 = (0 => 1, 1 => 2, 2 => 3)
(gdb) FAIL: gdb.ada/arrayidx.exp: print e_one_two_three
print r_two_three
$10 = (1 => 2, 2 => 3)
(gdb) FAIL: gdb.ada/arrayidx.exp: print r_two_three
print u_one_two_three
No definition of "u_one_two_three" in current context.
(gdb) FAIL: gdb.ada/arrayidx.exp: print u_one_two_three
print p_one_two_three
$11 = warning: could not find bounds information on packed array
can't unpack array

(gdb) FAIL: gdb.ada/arrayidx.exp: print p_one_two_three


ptype empty
type = function return int
(gdb) FAIL: gdb.ada/null_record.exp: ptype on null record


(gdb) print var
$1 = warning: could not find bounds information on packed array
can't unpack array

(gdb) FAIL: gdb.ada/packed_array.exp: print var


-- 
Daniel Jacobowitz
CodeSourcery

Re: arrayidx.exp failures

[Joel Brobecker]

> FYI, Joel, I have no idea what compilers you test with - but half of
> the Ada tests in the GDB testsuite have never passed for me with FSF
> gnat, at least 4.0 and 4.1, and possibly 3.4 too.

It's good you told me, I will try to talk to some of my collegues
who work on the compiler and see where we are.

I use exclusively GNAT Pro, which is the compiler we maintain at
AdaCore. Right now, it's based on GCC 3.4.6, but are actively moving
to a GCC 4.1 backend on most of our platforms. That's when we found
a lot of issues in debug info generation that apparently appear only
with Ada - we have fixed several of them and I believe that they have
been submitted although I will admit that I do not follow compiler
work very closely.

If I started doing my testing with various flavours of GCC, the number
of combinations would explode. I already run our in-house testsuite (not
based on dejagnu) together with the GDB testsuite, both with our
GDB tree and the FSF head.

All I can say is that we're getting better at pushing our changes
to the FSF tree both on the GCC side as well as the GDB side, but
it's a slow process because first we have to catch up and then we
have to learn how to best do it.

In the meantime, are these fails an issue? For me, I have the same sort
of issue in a smaller scale with C++ testcases, but the way I test is
by using a smart-diff between the testcase results before and after.

-- 
Joel

Re: arrayidx.exp failures

[Nick Roberts]

 > In the meantime, are these fails an issue? For me, I have the same sort
 > of issue in a smaller scale with C++ testcases, but the way I test is
 > by using a smart-diff between the testcase results before and after.

It's not an issue for those of us who don't use Ada but it might be one for
you, in that we might not notice our changes have broken the Ada tests if they
are already broken for us.

-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: arrayidx.exp failures

[Joel Brobecker]

> It's not an issue for those of us who don't use Ada but it might be
> one for you, in that we might not notice our changes have broken the
> Ada tests if they are already broken for us.

That's perfectly fine. We all do our best, and even when the FSF GCC is
better, some people may not have it installed and stil potentially break
some of the Ada testcases. I try to check the results on a regular
basis to catch such unavoidable mishaps.

-- 
Joel

Re: arrayidx.exp failures

[Daniel Jacobowitz]

On Fri, Oct 13, 2006 at 09:33:52AM -0700, Joel Brobecker wrote:
> In the meantime, are these fails an issue? For me, I have the same sort
> of issue in a smaller scale with C++ testcases, but the way I test is
> by using a smart-diff between the testcase results before and after.

Not at present.

Two or three years ago, I pushed i386-linux down almost to zero
failures.  It's crept back up ever since, and x86_64-linux currently
has a whole bunch more, but I'm still working on it when I can find
time.

At work we require zero-failure testsuites for most components;
right now we're looking at an exception for GDB, which makes me quite
sad.

-- 
Daniel Jacobowitz
CodeSourcery