[PATCH] Print die_hash2 for --odr --devel-dump-dies

[PATCH] Print die_hash2 for --odr --devel-dump-dies

[Tom de Vries]

Hi,

For --odr, there's an additional hash value, die->u.p1.die_hash2.

Print this with --devel-dump-dies, such that we have:
...
$ dwz --devel-dump-dies --odr odr-struct
  114 O f56004ac(cd46c7ab) f56004ac aaa structure_type
  ...
  1ba O f56004ac(cd46c7ab) f56004ac aaa structure_type
...

Used to debug PR27400.

Any comments?

Thanks,
- Tom

Print die_hash2 for --odr --devel-dump-dies

2021-02-15  Tom de Vries  <tdevries@suse.de>

	* dwz.c (dump_die_with_indent): Print die->u.p1.die_hash2.

---
 dwz.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/dwz.c b/dwz.c
index 17921f0..80bc1e7 100644
--- a/dwz.c
+++ b/dwz.c
@@ -5452,11 +5452,13 @@ dump_die_with_indent (int indent, dw_die_ref die)
   else
     {
       const char *name = get_name (die);
-      fprintf (stderr, "%*s %x %c %x %x %s %s", indent, "", die->die_offset,
+      fprintf (stderr, "%*s %x %c %x", indent, "", die->die_offset,
 	       die->die_ck_state == CK_KNOWN ? 'O' : 'X',
-	       (unsigned) die->u.p1.die_hash,
-	       (unsigned) die->u.p1.die_ref_hash, name ? name : "",
-	       get_DW_TAG_name (die->die_tag) + 7);
+	       (unsigned) die->u.p1.die_hash);
+      if (odr && die->die_odr_state != ODR_NONE)
+	  fprintf (stderr, "(%x)", (unsigned) die->u.p1.die_hash2);
+      fprintf (stderr, " %x %s %s", (unsigned) die->u.p1.die_ref_hash,
+	       name ? name : "", get_DW_TAG_name (die->die_tag) + 7);
       dump_type (die);
     }
   fprintf (stderr, "\n");

Re: [PATCH] Print die_hash2 for --odr --devel-dump-dies

[Mark Wielaard]

Hi Tom,

On Mon, 2021-02-15 at 21:34 +0100, Tom de Vries wrote:
> For --odr, there's an additional hash value, die->u.p1.die_hash2.
> 
> Print this with --devel-dump-dies, such that we have:
> ...
> $ dwz --devel-dump-dies --odr odr-struct
>   114 O f56004ac(cd46c7ab) f56004ac aaa structure_type
>   ...
>   1ba O f56004ac(cd46c7ab) f56004ac aaa structure_type
> ...
> 
> Used to debug PR27400.
> 
> Any comments?

Obviously this is useful.
But I always forget how to use these --devel flags.

They are only in the dwz-for-test binary, which isn't build by default,
and seems to have some subtle other differences, like undefining
__GNUC__?

Is there more documentation on the --devel flags/binary or is this just
meant for people directly hacking on dwz in the git repo?

Cheers,

Mark

Re: [PATCH] Print die_hash2 for --odr --devel-dump-dies

[Tom de Vries]

On 2/18/21 4:03 PM, Mark Wielaard wrote:
> Hi Tom,
> 
> On Mon, 2021-02-15 at 21:34 +0100, Tom de Vries wrote:
>> For --odr, there's an additional hash value, die->u.p1.die_hash2.
>>
>> Print this with --devel-dump-dies, such that we have:
>> ...
>> $ dwz --devel-dump-dies --odr odr-struct
>>   114 O f56004ac(cd46c7ab) f56004ac aaa structure_type
>>   ...
>>   1ba O f56004ac(cd46c7ab) f56004ac aaa structure_type
>> ...
>>
>> Used to debug PR27400.
>>
>> Any comments?
> 
> Obviously this is useful.
> But I always forget how to use these --devel flags.
> 
> They are only in the dwz-for-test binary, which isn't build by default,

Hi Mark,

they're only defined when building with -DDEVEL.  Which is enabled in
dwz-for-test (which is enabled by default in the testsuite), but also in
my usual build command.

> and seems to have some subtle other differences, like undefining
> __GNUC__?
> 

That's just a ploy to get both the -D__GNUC__ and -U__GNUC__ paths
exercised.  It's not really specific to -DDEVEL.

> Is there more documentation on the --devel flags/binary

Nope.

> or is this just
> meant for people directly hacking on dwz in the git repo?

Yep.

The idea was to move all hacky stuff (like f.i -DDEBUG_DUMP_DIES) to
command line switches under -DDEVEL, such that you have two versions:
- the official one, -UDEVEL, without anything slowing dwz down, or
  exposing unstable interfaces to users, and
- the development one, -DDEVEL, with the perks of extra traces
  and switches to influence behaviour, but possible slower and
  experimental.

HTH,
- Tom