[PATCH][gdb] Fix gdb.dwarf2/varval.exp with -fPIE/-pie

[PATCH][gdb] Fix gdb.dwarf2/varval.exp with -fPIE/-pie

[Tom de Vries]

Hi,

With target board unix/-fPIE/-pie, we get:
...
FAIL: gdb.dwarf2/varval.exp: print varval2
...

This is due comparing a get_frame_pc result (which includes the for PIE
non-zero relocation offset) with pc_high and pc_low obtained using
get_scope_pc_bounds (which do not include the relocation offset).

Fix this by adjusting pc_high and pc_low with the relocation offset.

Tested on x86_64-linux with target board unix/-fPIE/-pie.

OK for trunk?

Thanks,
- Tom

[gdb] Fix gdb.dwarf2/varval.exp with -fPIE/-pie

gdb/ChangeLog:

2019-08-08  Tom de Vries  <tdevries@suse.de>

	PR gdb/24591
	* dwarf2read.c (dwarf2_fetch_die_loc_sect_off): Adjust pc_high and
	pc_low with relocation offset.

---
 gdb/dwarf2read.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
index 1c26f8355f..fde08047ca 100644
--- a/gdb/dwarf2read.c
+++ b/gdb/dwarf2read.c
@@ -23206,6 +23206,9 @@ dwarf2_fetch_die_loc_sect_off (sect_offset sect_off,
 	  != dwarf2_per_objfile->abstract_to_concrete.end ()))
     {
       CORE_ADDR pc = (*get_frame_pc) (baton);
+      CORE_ADDR baseaddr
+	= ANOFFSET (objfile->section_offsets, SECT_OFF_TEXT (objfile));
+      struct gdbarch *gdbarch = get_objfile_arch (objfile);
 
       for (const auto &cand_off
 	     : dwarf2_per_objfile->abstract_to_concrete[die->sect_off])
@@ -23220,6 +23223,8 @@ dwarf2_fetch_die_loc_sect_off (sect_offset sect_off,
 
 	  CORE_ADDR pc_low, pc_high;
 	  get_scope_pc_bounds (cand->parent, &pc_low, &pc_high, cu);
+	  pc_low = gdbarch_adjust_dwarf2_addr (gdbarch, pc_low + baseaddr);
+	  pc_high = gdbarch_adjust_dwarf2_addr (gdbarch, pc_high + baseaddr);
 	  if (pc_low == ((CORE_ADDR) -1)
 	      || !(pc_low <= pc && pc < pc_high))
 	    continue;

Re: [PATCH][gdb] Fix gdb.dwarf2/varval.exp with -fPIE/-pie

[Tom Tromey]

>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:

Thanks for the patch.

Tom> +      CORE_ADDR baseaddr
Tom> +	= ANOFFSET (objfile->section_offsets, SECT_OFF_TEXT (objfile));
Tom> +      struct gdbarch *gdbarch = get_objfile_arch (objfile);
 
Tom>        for (const auto &cand_off
Tom>  	     : dwarf2_per_objfile->abstract_to_concrete[die->sect_off])
Tom> @@ -23220,6 +23223,8 @@ dwarf2_fetch_die_loc_sect_off (sect_offset sect_off,
 
Tom>  	  CORE_ADDR pc_low, pc_high;
Tom>  	  get_scope_pc_bounds (cand->parent, &pc_low, &pc_high, cu);
Tom> +	  pc_low = gdbarch_adjust_dwarf2_addr (gdbarch, pc_low + baseaddr);
Tom> +	  pc_high = gdbarch_adjust_dwarf2_addr (gdbarch, pc_high + baseaddr);
Tom>  	  if (pc_low == ((CORE_ADDR) -1)

I think this test has to precede the offsetting.

Tom

Re: [PATCH][gdb] Fix gdb.dwarf2/varval.exp with -fPIE/-pie

[Kevin Buettner]

On Thu, 08 Aug 2019 11:16:44 -0600
Tom Tromey <tom@tromey.com> wrote:

> >>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:  
> 
> Thanks for the patch.
> 
> Tom> +      CORE_ADDR baseaddr
> Tom> +	= ANOFFSET (objfile->section_offsets, SECT_OFF_TEXT (objfile));
> Tom> +      struct gdbarch *gdbarch = get_objfile_arch (objfile);  
>  
> Tom>        for (const auto &cand_off
> Tom>  	     : dwarf2_per_objfile->abstract_to_concrete[die->sect_off])
> Tom> @@ -23220,6 +23223,8 @@ dwarf2_fetch_die_loc_sect_off (sect_offset sect_off,  
>  
> Tom>  	  CORE_ADDR pc_low, pc_high;
> Tom>  	  get_scope_pc_bounds (cand->parent, &pc_low, &pc_high, cu);
> Tom> +	  pc_low = gdbarch_adjust_dwarf2_addr (gdbarch, pc_low + baseaddr);
> Tom> +	  pc_high = gdbarch_adjust_dwarf2_addr (gdbarch, pc_high + baseaddr);
> Tom>  	  if (pc_low == ((CORE_ADDR) -1)  
> 
> I think this test has to precede the offsetting.

Yes, I agree.

It appears to me that the test / continue in its entirety...

	  if (pc_low == ((CORE_ADDR) -1)
	      || !(pc_low <= pc && pc < pc_high))
	    continue;

...will need to be split with the -1 part being performed before the
pc_low/pc_high adjustment and the other part being performed after.

Kevin

Re: [PATCH][gdb] Fix gdb.dwarf2/varval.exp with -fPIE/-pie

[Tom de Vries]

[-- Attachment #1: Type: text/plain, Size: 1379 bytes --]

On 08-08-19 20:04, Kevin Buettner wrote:
> On Thu, 08 Aug 2019 11:16:44 -0600
> Tom Tromey <tom@tromey.com> wrote:
> 
>>>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:  
>>
>> Thanks for the patch.
>>
>> Tom> +      CORE_ADDR baseaddr
>> Tom> +	= ANOFFSET (objfile->section_offsets, SECT_OFF_TEXT (objfile));
>> Tom> +      struct gdbarch *gdbarch = get_objfile_arch (objfile);  
>>  
>> Tom>        for (const auto &cand_off
>> Tom>  	     : dwarf2_per_objfile->abstract_to_concrete[die->sect_off])
>> Tom> @@ -23220,6 +23223,8 @@ dwarf2_fetch_die_loc_sect_off (sect_offset sect_off,  
>>  
>> Tom>  	  CORE_ADDR pc_low, pc_high;
>> Tom>  	  get_scope_pc_bounds (cand->parent, &pc_low, &pc_high, cu);
>> Tom> +	  pc_low = gdbarch_adjust_dwarf2_addr (gdbarch, pc_low + baseaddr);
>> Tom> +	  pc_high = gdbarch_adjust_dwarf2_addr (gdbarch, pc_high + baseaddr);
>> Tom>  	  if (pc_low == ((CORE_ADDR) -1)  
>>
>> I think this test has to precede the offsetting.

Ouch, thanks for catching that.

> 
> Yes, I agree.
> 
> It appears to me that the test / continue in its entirety...
> 
> 	  if (pc_low == ((CORE_ADDR) -1)
> 	      || !(pc_low <= pc && pc < pc_high))
> 	    continue;
> 
> ...will need to be split with the -1 part being performed before the
> pc_low/pc_high adjustment and the other part being performed after.

Done.

Committed as attached.

Thanks,
- Tom



[-- Attachment #2: 0001-gdb-Fix-gdb.dwarf2-varval.exp-with-fPIE-pie.patch --]
[-- Type: text/x-patch, Size: 1839 bytes --]

[gdb] Fix gdb.dwarf2/varval.exp with -fPIE/-pie

With target board unix/-fPIE/-pie, we get:
...
FAIL: gdb.dwarf2/varval.exp: print varval2
...

This is due comparing a get_frame_pc result (which includes the for PIE
non-zero relocation offset) with pc_high and pc_low obtained using
get_scope_pc_bounds (which do not include the relocation offset).

Fix this by adjusting pc_high and pc_low with the relocation offset.

Tested on x86_64-linux with target board unix/-fPIE/-pie.

gdb/ChangeLog:

2019-08-08  Tom de Vries  <tdevries@suse.de>

	PR gdb/24591
	* dwarf2read.c (dwarf2_fetch_die_loc_sect_off): Adjust pc_high and
	pc_low with relocation offset.

---
 gdb/dwarf2read.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
index 1c26f8355f..de9755f6ce 100644
--- a/gdb/dwarf2read.c
+++ b/gdb/dwarf2read.c
@@ -23206,6 +23206,9 @@ dwarf2_fetch_die_loc_sect_off (sect_offset sect_off,
 	  != dwarf2_per_objfile->abstract_to_concrete.end ()))
     {
       CORE_ADDR pc = (*get_frame_pc) (baton);
+      CORE_ADDR baseaddr
+	= ANOFFSET (objfile->section_offsets, SECT_OFF_TEXT (objfile));
+      struct gdbarch *gdbarch = get_objfile_arch (objfile);
 
       for (const auto &cand_off
 	     : dwarf2_per_objfile->abstract_to_concrete[die->sect_off])
@@ -23220,8 +23223,11 @@ dwarf2_fetch_die_loc_sect_off (sect_offset sect_off,
 
 	  CORE_ADDR pc_low, pc_high;
 	  get_scope_pc_bounds (cand->parent, &pc_low, &pc_high, cu);
-	  if (pc_low == ((CORE_ADDR) -1)
-	      || !(pc_low <= pc && pc < pc_high))
+	  if (pc_low == ((CORE_ADDR) -1))
+	    continue;
+	  pc_low = gdbarch_adjust_dwarf2_addr (gdbarch, pc_low + baseaddr);
+	  pc_high = gdbarch_adjust_dwarf2_addr (gdbarch, pc_high + baseaddr);
+	  if (!(pc_low <= pc && pc < pc_high))
 	    continue;
 
 	  die = cand;