[Bug symtab/14605] New: Display of partially optimised out variables from registers, using dwarf piece info.

["aburgess at broadcom dot com" <sourceware-bugzilla@sourceware.org>]

http://sourceware.org/bugzilla/show_bug.cgi?id=14605

             Bug #: 14605
           Summary: Display of partially optimised out variables from
                    registers, using dwarf piece info.
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: symtab
        AssignedTo: unassigned@sourceware.org
        ReportedBy: aburgess@broadcom.com
    Classification: Unclassified


Created attachment 6645
  --> http://sourceware.org/bugzilla/attachment.cgi?id=6645
An x86-64 assembler test file for this bug.

Raising defect as requested in this gdb-patches mailing list thread:
  http://sourceware.org/ml/gdb-patches/2012-08/msg00715.html

This bug is for a failure in the test gdb.dwarf2/dw2-op-out-param.exp that
was added at the end of the above mailing list thread.

If a variable is described by the DWARF as being split over multiple
registers using DWARF piece information, but some of those register are
marked as optimised out by a later frame then gdb incorrectly believes the
value of those pieces is zero.

Having a variable help in a register that is marked as optimised out by a
later frame should probably be fixed in the dwarf producer, however, it is
interesting in this case that gdb, without error or warning displays
the *wrong* answer.  However, no example of this bug has yet been seen in
the wild.

The test creates two call stacks:
  main --> struct_param_two_reg_pieces --> breakpt
  main --> int_param_two_reg_pieces --> breakpt

In each function "struct_param_two_reg_pieces" and
"int_param_two_reg_pieces" variables are placed in multiple register using
piece information, and the function "breakpt" marks some of these
registers as optimised out.

The test file is x86-64 specific, download test.S, and assemble as:
gcc -o test test.S

Then run gdb as:
> gdb test
(gdb) break breakpt 
Breakpoint 1 at 0x4004fb
(gdb) r
Starting program: /projects/firepath_work/aburgess/tmp/gdb/test 
warning: no loadable sections found in added symbol-file system-supplied DSO at
0x2aaaaaaab000

Breakpoint 1, 0x00000000004004fb in breakpt ()
(gdb) c
Continuing.

Breakpoint 1, 0x00000000004004fb in breakpt ()
(gdb) c
Continuing.

Breakpoint 1, 0x00000000004004fb in breakpt ()
(gdb) set output-radix 16
Output radix now set to decimal 16, hex 10, octal 20.
(gdb) set print frame-arguments all
(gdb) bt
#0  0x00000000004004fb in breakpt ()
#1  0x000000000040052e in struct_param_two_reg_pieces (operand0=
      {a = 0xdeadbe04deadbe05, b = 0x0}, operand1=
      {a = 0x0, b = 0xdeadbe00deadbe01}, operand2={a = 0x0, b = 0x0})
#2  0x00000000004004ef in main ()
(gdb) c
Continuing.

Breakpoint 1, 0x00000000004004fb in breakpt ()
(gdb) bt
#0  0x00000000004004fb in breakpt ()
#1  0x000000000040053e in int_param_two_reg_pieces (operand0=0xdeadbe05, 
    operand1=0xdeadbe0100000000, operand2=0x0)
#2  0x00000000004004f5 in main ()
(gdb) quit

## END ##

In both backtraces the values for operand0, operand1, and operand2 are
wrong.

In the first backtrace the correct values are:
operand0={a = 0xdeadbe04deadbe05, b = <optimized out>}
operand1={a = <optimized out>, b = 0xdeadbe00deadbe01}
operand2=<optimized out>

In the second backtrace the correct values are:
operand0=<optimized out>
operand1=<optimized out>
operand2=<optimized out>

In the second case we must mark the whole value as optimised out as the
variable is an integer type, I can't imagine how we'd show a partially
optimised out integer variable.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.