[2/6] DWARF Two Level Line Tables: dwarf2_line_debug
Commit Message
Hi.
This patch adds "set debug dwarf2-line".
2015-03-12 Doug Evans <dje@google.com>
* NEWS: Mention "set debug dwarf2-line".
* dwarf2read.c (dwarf2_line_debug): New static global.
(add_include_dir): Add debug dwarf2-line support.
(add_file_name, dwarf_record_line, dwarf_finish_line): Ditto.
(_initialize_dwarf2_read): New parameter "debug dwarf2-line".
doc/
* gdb.texinfo (Debugging Output): Mention set/show debug dwarf2-line.
Comments
> From: Doug Evans <dje@google.com>
> cc: ccoutant@google.com
> Date: Thu, 12 Mar 2015 13:01:49 -0700
>
> --- a/gdb/NEWS
> +++ b/gdb/NEWS
> @@ -30,6 +30,10 @@ record bts
>
> * New options
>
> +set debug dwarf2-line
> +show debug dwarf2-line
> + Control display of debugging info regarding DWARF line processing.
This is OK.
> --- a/gdb/doc/gdb.texinfo
> +++ b/gdb/doc/gdb.texinfo
> @@ -23245,6 +23245,14 @@ The value is the number of nesting levels to print.
> A value of zero turns off the display.
> @item show debug dwarf2-die
> Show the current state of DWARF2 DIE debugging.
> +@item set debug dwarf2-line
> +@cindex DWARF2 Line Tables
> +Turns on or off display of debugging messages related to reading
> +DWARF line tables. The default is 0 (off).
> +A value of 1 provides basic information.
> +A value greater than 1 provides more verbose information.
> +@item show debug dwarf2-line
> +Show the current state of DWARF2 line table debugging.
This is also OK, but how about using "DWARF2" for consistency at least
in this fragment?
(Btw, I find this "DWARF2", "DWARF 4", "DWARF 5" thing utterly
confusing for the reader; the truth is "DWARF version 5" etc. I wish
we started using that throughout.)
Thanks.
On Thu, Mar 12, 2015 at 1:37 PM, Eli Zaretskii <eliz@gnu.org> wrote:
>> From: Doug Evans <dje@google.com>
>> cc: ccoutant@google.com
>> Date: Thu, 12 Mar 2015 13:01:49 -0700
>>
>> --- a/gdb/NEWS
>> +++ b/gdb/NEWS
>> @@ -30,6 +30,10 @@ record bts
>>
>> * New options
>>
>> +set debug dwarf2-line
>> +show debug dwarf2-line
>> + Control display of debugging info regarding DWARF line processing.
>
> This is OK.
>
>> --- a/gdb/doc/gdb.texinfo
>> +++ b/gdb/doc/gdb.texinfo
>> @@ -23245,6 +23245,14 @@ The value is the number of nesting levels to print.
>> A value of zero turns off the display.
>> @item show debug dwarf2-die
>> Show the current state of DWARF2 DIE debugging.
>> +@item set debug dwarf2-line
>> +@cindex DWARF2 Line Tables
>> +Turns on or off display of debugging messages related to reading
>> +DWARF line tables. The default is 0 (off).
>> +A value of 1 provides basic information.
>> +A value greater than 1 provides more verbose information.
>> +@item show debug dwarf2-line
>> +Show the current state of DWARF2 line table debugging.
>
> This is also OK, but how about using "DWARF2" for consistency at least
> in this fragment?
Actually, I'd rather use "DWARF".
There's a lot of history/inertia here that we should really overcome
and put behind us.
We call things DWARF2 to disambiguate from DWARF1 (which no longer
exists in any useful form). dwarf2read.c should *really* be
dwarfread.c IMO (and so on for every other instance of
dwarf2/DWARF2/dw2/etc. in file names, code, docs, etc.)
> (Btw, I find this "DWARF2", "DWARF 4", "DWARF 5" thing utterly
> confusing for the reader; the truth is "DWARF version 5" etc. I wish
> we started using that throughout.)
Agreed. Where the version doesn't matter I propose using "DWARF".
(as in the above)
> Date: Tue, 12 May 2015 13:39:27 -0700
> From: Doug Evans <dje@google.com>
> Cc: gdb-patches <gdb-patches@sourceware.org>, Cary Coutant <ccoutant@google.com>
>
> >> A value of zero turns off the display.
> >> @item show debug dwarf2-die
> >> Show the current state of DWARF2 DIE debugging.
> >> +@item set debug dwarf2-line
> >> +@cindex DWARF2 Line Tables
> >> +Turns on or off display of debugging messages related to reading
> >> +DWARF line tables. The default is 0 (off).
> >> +A value of 1 provides basic information.
> >> +A value greater than 1 provides more verbose information.
> >> +@item show debug dwarf2-line
> >> +Show the current state of DWARF2 line table debugging.
> >
> > This is also OK, but how about using "DWARF2" for consistency at least
> > in this fragment?
>
> Actually, I'd rather use "DWARF".
I don't mind, as long as it's consistent.
> Agreed. Where the version doesn't matter I propose using "DWARF".
> (as in the above)
The above used both DWARF and DWARF2.
@@ -30,6 +30,10 @@ record bts
* New options
+set debug dwarf2-line
+show debug dwarf2-line
+ Control display of debugging info regarding DWARF line processing.
+
set max-completions
show max-completions
Set the maximum number of candidates to be considered during
@@ -23245,6 +23245,14 @@ The value is the number of nesting levels to print.
A value of zero turns off the display.
@item show debug dwarf2-die
Show the current state of DWARF2 DIE debugging.
+@item set debug dwarf2-line
+@cindex DWARF2 Line Tables
+Turns on or off display of debugging messages related to reading
+DWARF line tables. The default is 0 (off).
+A value of 1 provides basic information.
+A value greater than 1 provides more verbose information.
+@item show debug dwarf2-line
+Show the current state of DWARF2 line table debugging.
@item set debug dwarf2-read
@cindex DWARF2 Reading
Turns on or off display of debugging messages related to reading
@@ -84,6 +84,9 @@ static unsigned int dwarf2_read_debug = 0;
/* When non-zero, dump DIEs after they are read in. */
static unsigned int dwarf2_die_debug = 0;
+/* When non-zero, dump line number entries as they are read in. */
+static unsigned int dwarf2_line_debug = 0;
+
/* When non-zero, cross-check physname against demangler. */
static int check_physname = 0;
@@ -17106,6 +17109,10 @@ free_line_header_voidp (void *arg)
static void
add_include_dir (struct line_header *lh, const char *include_dir)
{
+ if (dwarf2_line_debug >= 2)
+ fprintf_unfiltered (gdb_stdlog, "Adding dir %u: %s\n",
+ lh->num_include_dirs + 1, include_dir);
+
/* Grow the array if necessary. */
if (lh->include_dirs_size == 0)
{
@@ -17135,6 +17142,10 @@ add_file_name (struct line_header *lh,
{
struct file_entry *fe;
+ if (dwarf2_line_debug >= 2)
+ fprintf_unfiltered (gdb_stdlog, "Adding file %u: %s\n",
+ lh->num_file_names + 1, name);
+
/* Grow the array if necessary. */
if (lh->file_names_size == 0)
{
@@ -17483,6 +17494,14 @@ dwarf_record_line (struct gdbarch *gdbarch, struct subfile *subfile,
{
CORE_ADDR addr = gdbarch_addr_bits_remove (gdbarch, address);
+ if (dwarf2_line_debug)
+ {
+ fprintf_unfiltered (gdb_stdlog,
+ "Recording line %u, file %s, address %s\n",
+ line, lbasename (subfile->name),
+ paddress (gdbarch, address));
+ }
+
(*p_record_line) (subfile, line, addr);
}
@@ -17495,8 +17514,18 @@ static void
dwarf_finish_line (struct gdbarch *gdbarch, struct subfile *subfile,
CORE_ADDR address, record_line_ftype p_record_line)
{
- if (subfile != NULL)
- dwarf_record_line (gdbarch, subfile, 0, address, p_record_line);
+ if (subfile == NULL)
+ return;
+
+ if (dwarf2_line_debug)
+ {
+ fprintf_unfiltered (gdb_stdlog,
+ "Finishing current line, file %s, address %s\n",
+ lbasename (subfile->name),
+ paddress (gdbarch, address));
+ }
+
+ dwarf_record_line (gdbarch, subfile, 0, address, p_record_line);
}
/* Subroutine of dwarf_decode_lines to simplify it.
@@ -23278,6 +23307,16 @@ The value is the maximum depth to print."),
NULL,
&setdebuglist, &showdebuglist);
+ add_setshow_zuinteger_cmd ("dwarf2-line", no_class, &dwarf2_line_debug, _("\
+Set debugging of the dwarf2 line reader."), _("\
+Show debugging of the dwarf2 line reader."), _("\
+When enabled (non-zero), line number entries are dumped as they are read in.\n\
+A value of 1 (one) provides basic information.\n\
+A value greater than 1 provides more verbose information."),
+ NULL,
+ NULL,
+ &setdebuglist, &showdebuglist);
+
add_setshow_boolean_cmd ("check-physname", no_class, &check_physname, _("\
Set cross-checking of \"physname\" code against demangler."), _("\
Show cross-checking of \"physname\" code against demangler."), _("\