[Bug tdep/29543] [gdb/tdep, ppc] inferior call with long double vararg not handled correctly

["cel at us dot ibm.com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=29543

--- Comment #13 from Carl E Love <cel at us dot ibm.com> ---

I ran the following on a Power 10 system with:
Fedora release 36 (Thirty Six)
gcc (GCC) 12.2.1 20220819 (Red Hat 12.2.1-1)


Initially with the current gdb tree, the top commit was:

commit d647c797b7386dcaca5984a3fbe9fba469771737 (origin/master, origin/HEAD)
Author: Aaron Merey <amerey@redhat.com>
Date:   Fri Sep 2 19:01:56 2022 -0400

    Add debuginfod support for objdump -S
   ...

The   make check RUNTESTFLAGS='GDB=/home/carll/bin/gdb gdb.base/varargs.exp '
test was used.

The results from the test in gdb/testsuite/gdb.log were:

gdb) FAIL: gdb.base/varargs.exp: print find_max_long_double_real(4, ldc1, ldc2,
ldc3, ldc4)
testcase
/home/carll/GDB/binutils-gdb-bug-29247/gdb/testsuite/gdb.base/varargs.exp
completed in 0 seconds

                === gdb Summary ===

# of expected passes            10
# of unexpected failures        1

gdb reported one failure as expected.

-----------------

From Tom's patch, I added just the gdb code change:

--- a/gdb/ppc-sysv-tdep.c
+++ b/gdb/ppc-sysv-tdep.c
@@ -1444,7 +1444,7 @@ ppc64_sysv_abi_push_param (struct gdbarch *gdbarch,
          == floatformats_ieee_quad))
     {
       /* IEEE FLOAT128, args in vector registers.  */
-      ppc64_sysv_abi_push_val (gdbarch, val, TYPE_LENGTH (type), 0, argpos);
+      ppc64_sysv_abi_push_val (gdbarch, val, TYPE_LENGTH (type), 16, argpos);
       ppc64_sysv_abi_push_vreg (gdbarch, val, argpos);
     }

The gdb test now passes with:

               === gdb Summary ===

# of expected passes            11

as expected.

---------------------

Add just the testcase changes for the long double _Complex case, i.e. no
additional gdb functional changes.  The results of the run were:


(gdb) FAIL: gdb.base/varargs.exp: print find_max_struct_long_double_real(4,
sldc1, sldc2, sldc3, sldc4)
testcase
/home/carll/GDB/binutils-gdb-bug-29247/gdb/testsuite/gdb.base/varargs.exp
completed in 1 seconds

                === gdb Summary ===

# of expected passes            11
# of unexpected failures        1

Which generates the error as Ulrich expected.

----------------------

Added the additional fixes from Tom's patch to address the complex case.  The 
results of the gdb test was:


               === gdb Summary ===

# of expected passes            12

With Tom's additional fixes, the updated testcase also passes.


Tom's patch works with elf v2 and the gcc version 12 built to default to
128-bit IEEE long double.

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