[Bug tdep/29816] New: [gdb/tdep, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[Bug tdep/29816] New: [gdb/tdep, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

            Bug ID: 29816
           Summary: [gdb/tdep, powerpc64le] FAIL: gdb.ada/float-bits.exp:
                    print 16llf#4000921fb54442d18469898cc51701b8#
           Product: gdb
           Version: 12.1
            Status: NEW
          Severity: normal
          Priority: P2
         Component: tdep
          Assignee: unassigned at sourceware dot org
          Reporter: vries at gcc dot gnu.org
  Target Milestone: ---

In OBS, with a gdb 12.1 based package, on ppc64le, I run into:
...
FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#
FAIL: gdb.ada/float-bits.exp: print
$foo:=16llf#4000921fb54442d18469898cc51701b8#
FAIL: gdb.ada/float-bits.exp: print internal long double variable after
assignment
...

In more detail:
...
(gdb) print 16llf#4000921fb54442d18469898cc51701b8#^M
$9 = <invalid float value>^M
(gdb) FAIL: gdb.ada/float-bits.exp: print
16llf#4000921fb54442d18469898cc51701b8#
print $foo:=16llf#4000921fb54442d18469898cc51701b8#^M
$10 = <invalid float value>^M
(gdb) FAIL: gdb.ada/float-bits.exp: print
$foo:=16llf#4000921fb54442d18469898cc51701b8#
print $foo^M
$11 = <invalid float value>^M
(gdb) FAIL: gdb.ada/float-bits.exp: print internal long double variable after
assignment
...

I have some hope these are no longer reproducible with master, but I'm having
trouble installing an ada compiler on my powerpc64le box (so, outside OBS) to
verify this, so I'm filing this not to lose track of this.

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

[Bug tdep/29816] [gdb/tdep, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|12.1                        |HEAD

--- Comment #1 from Tom de Vries <vries at gcc dot gnu.org> ---
Nope, still fails on master.

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

[Bug tdep/29816] [gdb/tdep, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

--- Comment #2 from Tom de Vries <vries at gcc dot gnu.org> ---
reproducible with:
...
$ gdb -q -batch -ex "set language ada" -ex "print
16llf#4000921fb54442d18469898cc51701b8#"
$1 = <invalid float value>
...

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

[Bug tdep/29816] [gdb/tdep, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

--- Comment #3 from Tom de Vries <vries at gcc dot gnu.org> ---
Hmm, so it seems floatformat_ibm_long_double_is_valid returns 0, because:
...
328       if (bot_exp > top_exp - 53 && bot_exp != 0)
(gdb) n
329         return 0;
...

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[gdb/tdep, powerpc64le]     |[gdb/testsuite,
                   |FAIL:                       |powerpc64le] FAIL:
                   |gdb.ada/float-bits.exp:     |gdb.ada/float-bits.exp:
                   |print                       |print
                   |16llf#4000921fb54442d184698 |16llf#4000921fb54442d184698
                   |98cc51701b8#                |98cc51701b8#
                 CC|                            |cel at us dot ibm.com,
                   |                            |ulrich.weigand at de dot ibm.com
          Component|tdep                        |testsuite

--- Comment #4 from Tom de Vries <vries at gcc dot gnu.org> ---
Downgrading this to a test-case issue.

It's plausible that a hex representation is valid for one float format, and not
for another.

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[tromey at sourceware dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

Tom Tromey <tromey at sourceware dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tromey at sourceware dot org

--- Comment #5 from Tom Tromey <tromey at sourceware dot org> ---
There's already a bit of code in this .exp to check for
non-IEEE floating-point and adapt.  I guess this just needs
to be extended for this target.

My guess would be that it's not set up for little-endian PPC,
though I'm not 100% sure.

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

--- Comment #6 from Tom de Vries <vries at gcc dot gnu.org> ---
This makes the test pass for me:
...
diff --git a/gdb/testsuite/gdb.ada/float-bits.exp
b/gdb/testsuite/gdb.ada/float-bits.exp
index 55a0566580c..3bf22637910 100644
--- a/gdb/testsuite/gdb.ada/float-bits.exp
+++ b/gdb/testsuite/gdb.ada/float-bits.exp
@@ -93,6 +93,10 @@ if { [istarget x86_64-*-* ] || [istarget i?86-*-*] } {
        set invalid_long_double "16llf#9019a5c800007ffff7ff4054#"
        set printed_long_double "5.0e\\+25"
     }
+} elseif { [istarget powerpc*-*-*] } {
+    # 5.0e+25 in floatformat_ibm_long_double.
+    set valid_long_double "16llf#c1e1c000000000004544adf4b7320335#"
+    set printed_long_double "50000000000000000000000000.0"
 }

 # Exercise GDB-side long doubles.
...
but I suppose it'll fail for a compiler defaulting to floatformats_ieee_quad.

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[cel at us dot ibm.com]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

--- Comment #7 from Carl E Love <cel at us dot ibm.com> ---
I tried the fix on Power 9 and Power 10.  I see the same results with and
without the patch. 

make check RUNTESTFLAGS='GDB=/home/carll/bin/gdb gdb.ada/float-bits.exp '

The contents of gdb/testsuite/gdb.log:

print val_double^M
$8 = -2.0e-19^M
(gdb) PASS: gdb.ada/float-bits.exp: print val_double after assignment
ptype long_long_float^M
type = <8-byte float>^M
(gdb) ptype 16llf#0#^M
type = <16-byte float>^M
(gdb) print 16llf#4000921fb54442d18469898cc51701b8#^M
$9 = <invalid float value>^M
(gdb) FAIL: gdb.ada/float-bits.exp: print
16llf#4000921fb54442d18469898cc51701b8#
print $foo:=16llf#4000921fb54442d18469898cc51701b8#^M
$10 = <invalid float value>^M
(gdb) FAIL: gdb.ada/float-bits.exp: print
$foo:=16llf#4000921fb54442d18469898cc51701b8#
print $foo^M
$11 = <invalid float value>^M
(gdb) FAIL: gdb.ada/float-bits.exp: print internal long double variable after
assignment
testcase
/home/carll/GDB/build-current/gdb/testsuite/../../../binutils-gdb-current/gdb/testsuite/gdb.ada/float-bits.exp
c\
ompleted in 0 seconds

                === gdb Summary ===

# of expected passes            9
# of unexpected failures        3



You mentioned OBS, what is that? Perhaps there is something different in our
environments??

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

--- Comment #8 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Carl E Love from comment #7)
> I tried the fix on Power 9 and Power 10.  I see the same results with and
> without the patch. 
> 
> make check RUNTESTFLAGS='GDB=/home/carll/bin/gdb gdb.ada/float-bits.exp '
> 
> The contents of gdb/testsuite/gdb.log:
> 
> print val_double^M
> $8 = -2.0e-19^M
> (gdb) PASS: gdb.ada/float-bits.exp: print val_double after assignment
> ptype long_long_float^M
> type = <8-byte float>^M
> (gdb) ptype 16llf#0#^M
> type = <16-byte float>^M
> (gdb) print 16llf#4000921fb54442d18469898cc51701b8#^M

Is this a sign of the patch not being applied, or of the patch not working?

The patch attempts to print a different hex string here.

> $9 = <invalid float value>^M
> (gdb) FAIL: gdb.ada/float-bits.exp: print
> 16llf#4000921fb54442d18469898cc51701b8#
> print $foo:=16llf#4000921fb54442d18469898cc51701b8#^M
> $10 = <invalid float value>^M
> (gdb) FAIL: gdb.ada/float-bits.exp: print
> $foo:=16llf#4000921fb54442d18469898cc51701b8#
> print $foo^M
> $11 = <invalid float value>^M
> (gdb) FAIL: gdb.ada/float-bits.exp: print internal long double variable
> after assignment
> testcase
> /home/carll/GDB/build-current/gdb/testsuite/../../../binutils-gdb-current/
> gdb/testsuite/gdb.ada/float-bits.exp c\
> ompleted in 0 seconds
> 
>                 === gdb Summary ===
> 
> # of expected passes            9
> # of unexpected failures        3
> 
> 
> 
> You mentioned OBS, what is that? Perhaps there is something different in our
> environments??

It's Open Build Service, specifically the place where I maintain a gdb package
for openSUSE/SLE ( https://build.opensuse.org/package/show/devel:gcc/gdb ).

But I debugged the test-case and wrote the fix outside of OBS.

A possibility for the FAILs with the patch that you mentioned, is that it's the
effect of having a different long double format.

The compiler I'm using is gcc 7.5.0, with the long double format defaulting to
--with-long-double-format=ibm.

But well, there could also be endian issues, I'm using powerpc64le.

The way I produced the hex string is by compiling:
...
long double a = 5.0e+25L;

int
main (void)
{
  return 0;
}
...
and in gdb doing "p /x a".

So, which hex string are you seeing when you do that, and which long double
format are you using?

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

--- Comment #9 from Tom de Vries <vries at gcc dot gnu.org> ---
I start to wonder whether we should give up trying to hardcode these hex
numbers, and just work with whatever is printed for "p /x Val_Long_Double".

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

--- Comment #10 from Tom de Vries <vries at gcc dot gnu.org> ---
https://sourceware.org/pipermail/gdb-patches/2022-November/194353.html

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[cvs-commit at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

--- Comment #11 from cvs-commit at gcc dot gnu.org <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Tom de Vries <vries@sourceware.org>:

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=32a5aa2625686a709b22db19300a578d96285dd1

commit 32a5aa2625686a709b22db19300a578d96285dd1
Author: Tom de Vries <tdevries@suse.de>
Date:   Thu Dec 1 07:25:04 2022 +0100

    [gdb/testsuite] Fix gdb.ada/float-bits.exp for powerpc64le

    On powerpc64le-linux, I run into:
    ...
    (gdb) print 16llf#4000921fb54442d18469898cc51701b8#^M
    $9 = <invalid float value>^M
    (gdb) FAIL: gdb.ada/float-bits.exp: print \
      16llf#4000921fb54442d18469898cc51701b8#
    ...

    The problem is that we're using a hex string for the 128-bit IEEE quad long
    double format, but the actual long double float format is:
    ...
    gdbarch_dump: long_double_format = floatformat_ibm_long_double_little^M
    ...

    Fix this by using the hex string obtained by compiling test.c:
    ...
    long double a = 5.0e+25L;
    ...
    like so:
    ...
    $ gcc -mlittle test.c -c -g
    ...
    and running gdb:
    ...
    $ gdb -q -batch test.o -ex "p /x a"
    $1 = 0xc1e1c000000000004544adf4b7320335
    ...
    and likewise for -mbig:
    ...
    $ gdb -q -batch test.o -ex "p /x a"
    $1 = 0x4544adf4b7320335c1e1c00000000000
    ...

    Tested on powerpc64le-linux.

    I excercised the case of floatformat_ibm_long_double_big by
    using "set endian big" in the test-case.

    Note that for this patch to work correctly, recent commit aaa79cd62b8
("[gdb]
    Improve printing of float formats") is required.

    PR testsuite/29816
    Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29816
    Approved-By: Tom Tromey <tom@tromey.com>

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

[Bug testsuite/29816] [gdb/testsuite, powerpc64le] FAIL: gdb.ada/float-bits.exp: print 16llf#4000921fb54442d18469898cc51701b8#

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29816

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
   Target Milestone|---                         |13.1
         Resolution|---                         |FIXED

--- Comment #12 from Tom de Vries <vries at gcc dot gnu.org> ---
Fixed by above commit.

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