Re: [RFA]dwarf reader: Avoid complaint on const type

[Pedro Alves <pedro@codesourcery.com>]

On Friday 21 May 2010 21:35:37, Pierre Muller wrote:

> > Pierre> 2010-05-21  Pierre Muller  <muller@ics.u-strasbg.fr>
> > Pierre> 	* dwarf2read.c (process_die): Also allow DW_TAG_const_type
> > Pierre> 	and DW_TAG_volatile_type.
> > Pierre> 	(new_symbol): Likewise.

This caused

 @@ -7385,16 +7385,16 @@ PASS: gdb.base/sigaltstack.exp: handle S
  PASS: gdb.base/sigaltstack.exp: handle SIGVTALRM print pass nostop
  PASS: gdb.base/sigaltstack.exp: handle SIGPROF print pass nostop
  PASS: gdb.base/sigaltstack.exp: break catcher if level == INNER
 -PASS: gdb.base/sigaltstack.exp: continue to catch
 +FAIL: gdb.base/sigaltstack.exp: continue to catch (the program exited)
  PASS: gdb.base/sigaltstack.exp: next
 -PASS: gdb.base/sigaltstack.exp: backtrace
 -PASS: gdb.base/sigaltstack.exp: finish from catch LEAF
 -PASS: gdb.base/sigaltstack.exp: finish to throw INNER
 -PASS: gdb.base/sigaltstack.exp: finish to catch INNER
 -PASS: gdb.base/sigaltstack.exp: finish from catch INNER
 -PASS: gdb.base/sigaltstack.exp: finish to OUTER
 -PASS: gdb.base/sigaltstack.exp: finish to catch MAIN
 -PASS: gdb.base/sigaltstack.exp: finish to MAIN
 +FAIL: gdb.base/sigaltstack.exp: backtrace (pattern 1)
 +FAIL: gdb.base/sigaltstack.exp: finish from catch LEAF (the program is no longer running)
 +FAIL: gdb.base/sigaltstack.exp: finish to throw INNER (the program is no longer running)
 +FAIL: gdb.base/sigaltstack.exp: finish to catch INNER (the program is no longer running)
 +FAIL: gdb.base/sigaltstack.exp: finish from catch INNER (the program is no longer running)
 +FAIL: gdb.base/sigaltstack.exp: finish to OUTER (the program is no longer running)
 +FAIL: gdb.base/sigaltstack.exp: finish to catch MAIN (the program is no longer running)
 +FAIL: gdb.base/sigaltstack.exp: finish to MAIN (the program is no longer running)

for me.

The problem is in the `level' variable in the test.


 (gdb) break catcher if level == INNER
 A syntax error in expression, near `== INNER'.
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 (gdb) PASS: gdb.base/sigaltstack.exp: break catcher if level == INNER
 continue
 Continuing.

 Program received signal SIGALRM, Alarm clock.

 Program received signal SIGVTALRM, Virtual timer expired.

 Program exited with code 03.
 (gdb) FAIL: gdb.base/sigaltstack.exp: continue to catch (the program exited)
 ...

Note that the variable has the same name as the enum:

 enum level { MAIN, OUTER, INNER, LEAF, NR_LEVELS };

 /* Levels completed flag.  */
 volatile enum level level = NR_LEVELS;


It boils down to this --- before the patch, running sigaltstack
under gdb (run to main):

 (gdb) p level
 $1 = NR_LEVELS
 (gdb) ptype level
 type = volatile enum level {MAIN, OUTER, INNER, LEAF, NR_LEVELS}

and after the patch:

 (gdb) p level
 Attempt to use a type name as an expression
 (gdb) ptype level
 type = volatile enum level {MAIN, OUTER, INNER, LEAF, NR_LEVELS}



Here's the relevant dwarf:

 <1><44d>: Abbrev Number: 20 (DW_TAG_enumeration_type)
    <44e>   DW_AT_name        : (indirect string, offset: 0xe3): level	
    <452>   DW_AT_byte_size   : 4	
    <453>   DW_AT_decl_file   : 1	
    <454>   DW_AT_decl_line   : 24	
    <455>   DW_AT_sibling     : <0x478>	
 <2><459>: Abbrev Number: 21 (DW_TAG_enumerator)
    <45a>   DW_AT_name        : (indirect string, offset: 0x2be): MAIN	
    <45e>   DW_AT_const_value : 0	
 <2><45f>: Abbrev Number: 21 (DW_TAG_enumerator)
    <460>   DW_AT_name        : (indirect string, offset: 0x281): OUTER	
    <464>   DW_AT_const_value : 1	
...
 <1><5b7>: Abbrev Number: 31 (DW_TAG_variable)
    <5b8>   DW_AT_name        : (indirect string, offset: 0xe3): level	
    <5bc>   DW_AT_decl_file   : 1	
    <5bd>   DW_AT_decl_line   : 27	
    <5be>   DW_AT_type        : <0x5cd>	
    <5c2>   DW_AT_external    : 1	
    <5c3>   DW_AT_location    : 9 byte block: 3 48 10 60 0 0 0 0 0 	(DW_OP_addr: 601048)
 <1><5cd>: Abbrev Number: 32 (DW_TAG_volatile_type)
    <5ce>   DW_AT_name        : (indirect string, offset: 0xe3): level	
    <5d2>   DW_AT_type        : <0x44d>	

Below's the patch again:  It's not clear to me why is it correct to
treat DW_TAG_const_type and DW_TAG_volatile_type qualifiers as
typedefs, though I haven't investigated this much yet.  FYI, or
in case it rings a bell...

> Index: src/gdb/dwarf2read.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/dwarf2read.c,v
> retrieving revision 1.386
> diff -u -p -r1.386 dwarf2read.c
> --- src/gdb/dwarf2read.c        17 May 2010 15:55:01 -0000      1.386
> +++ src/gdb/dwarf2read.c        21 May 2010 14:40:59 -0000
> @@ -3194,6 +3194,8 @@ process_die (struct die_info *die, struc
>      case DW_TAG_base_type:
>      case DW_TAG_subrange_type:
>      case DW_TAG_typedef:
> +    case DW_TAG_const_type:
> +    case DW_TAG_volatile_type:
>        /* Add a typedef symbol for the type definition, if it has a
>           DW_AT_name.  */
>        new_symbol (die, read_type_die (die, cu), cu);
> @@ -8742,6 +8744,8 @@ new_symbol (struct die_info *die, struct
>           break;
>         case DW_TAG_base_type:
>          case DW_TAG_subrange_type:
> +        case DW_TAG_const_type:
> +        case DW_TAG_volatile_type:
>           SYMBOL_CLASS (sym) = LOC_TYPEDEF;
>           SYMBOL_DOMAIN (sym) = VAR_DOMAIN;
>           add_symbol_to_list (sym, cu->list_in_scope);

-- 
Pedro Alves