[PATCH 0/4] [gdb/testsuite] Fix some more dwarf assembly test-cases

[PATCH 0/4] [gdb/testsuite] Fix some more dwarf assembly test-cases

[Tom de Vries]

Three more patches fixing warnings with readelf -w or llvm-dwarfdump -a in
dwarf assembly test-cases.

In the process, I found a problem in readelf, fixed by commit 1678d7d3d2e
("[binutils] Handle implicit const in .debug_names").

The last patch doesn't fix a warning, but fixes comments related to implicit
const in generated dwarf assembly.

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=34157

Tom de Vries (4):
  [gdb/testsuite] Remove invalid .debug_line section in
    gdb.dwarf2/dw2-icycle.exp
  [gdb/testsuite] Complete 64-bit port of
    gdb.dwarf2/dw2-skip-prologue.exp
  [gdb/testsuite] Add missing type_offset in
    gdb.dwarf/debug-names-tu.exp
  [gdb/testsuite] Improve dwarf assembly for implicit const in
    .debug_names

 .../gdb.dwarf2/debug-names-bad-cu-index.exp   |  6 ++--
 .../gdb.dwarf2/debug-names-tu.exp.tcl         |  6 ++--
 gdb/testsuite/gdb.dwarf2/dw2-const.S          | 18 +++++++++---
 gdb/testsuite/gdb.dwarf2/dw2-const.exp        | 12 +++++++-
 gdb/testsuite/gdb.dwarf2/dw2-icycle.S         |  4 ---
 gdb/testsuite/gdb.dwarf2/dw2-skip-prologue.S  | 28 ++++++++++---------
 gdb/testsuite/lib/dwarf.exp                   |  5 ++--
 7 files changed, 51 insertions(+), 28 deletions(-)


base-commit: 1678d7d3d2e4cf83cda5a96334a0326573ad0f7c
-- 
2.51.0

[PATCH 1/4] [gdb/testsuite] Remove invalid .debug_line section in gdb.dwarf2/dw2-icycle.exp

[Tom de Vries]

Test-case source gdb.dwarf2/dw2-icycle.S contains an invalid .debug_line
section contribution, causing a readelf -w warning.

It's also unused, so fix the warning by removing it.

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=34157
---
 gdb/testsuite/gdb.dwarf2/dw2-icycle.S | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/gdb/testsuite/gdb.dwarf2/dw2-icycle.S b/gdb/testsuite/gdb.dwarf2/dw2-icycle.S
index 3e79c966aef..4bd67a4cb7d 100644
--- a/gdb/testsuite/gdb.dwarf2/dw2-icycle.S
+++ b/gdb/testsuite/gdb.dwarf2/dw2-icycle.S
@@ -250,10 +250,6 @@ p__top:
 	.byte	0
 	.byte	0
 
-        .section .debug_line
-.Lline1_begin:
-        .byte   0
-
 #ifdef __arm__
 	.section	.note.GNU-stack,"",%progbits
 #else
-- 
2.51.0

Re: [PATCH 1/4] [gdb/testsuite] Remove invalid .debug_line section in gdb.dwarf2/dw2-icycle.exp

[Tom Tromey]

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

Tom> Test-case source gdb.dwarf2/dw2-icycle.S contains an invalid .debug_line
Tom> section contribution, causing a readelf -w warning.

Tom> It's also unused, so fix the warning by removing it.

Looks good.
Approved-By: Tom Tromey <tom@tromey.com>

Tom

[PATCH 2/4] [gdb/testsuite] Complete 64-bit port of gdb.dwarf2/dw2-skip-prologue.exp

[Tom de Vries]

Test-source gdb.dwarf2/dw2-skip-prologue.S was generated using 32-bit, and
subsequently ported to 64-bit using macro PTRBYTE.

When using 64-bit, there are some readelf -w warnings because the port to
64-bit is not complete.  Fix these warnings.

Likewise in gdb.dwarf/dw2-const.exp.

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=34157
---
 gdb/testsuite/gdb.dwarf2/dw2-const.S         | 18 ++++++++++---
 gdb/testsuite/gdb.dwarf2/dw2-const.exp       | 12 ++++++++-
 gdb/testsuite/gdb.dwarf2/dw2-skip-prologue.S | 28 +++++++++++---------
 3 files changed, 40 insertions(+), 18 deletions(-)

diff --git a/gdb/testsuite/gdb.dwarf2/dw2-const.S b/gdb/testsuite/gdb.dwarf2/dw2-const.S
index 94cf1acbed6..1161633f181 100644
--- a/gdb/testsuite/gdb.dwarf2/dw2-const.S
+++ b/gdb/testsuite/gdb.dwarf2/dw2-const.S
@@ -15,6 +15,16 @@
    You should have received a copy of the GNU General Public License
    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
 
+#if PTRBITS == 64
+# define PTRBYTE .8byte
+# define PTRBYTES 8
+#elif PTRBITS == 32
+# define PTRBYTE .4byte
+# define PTRBYTES 4
+#else
+# error "Unsupported pointer size"
+#endif
+
 /* Dummy function to provide debug information for.  */
 
 	.text
@@ -228,9 +238,9 @@ func_cu1:
 
 .Lline1_lines:
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
-	.4byte		.Lbegin_func_cu1
+	PTRBYTE		.Lbegin_func_cu1
 
 	.byte		3	/* DW_LNS_advance_line */
 	.sleb128	3	/* ... to 4 */
@@ -240,9 +250,9 @@ func_cu1:
 	.byte		1	/* DW_LNS_copy (second time as an end-of-prologue marker) */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
-	.4byte		.Lend_func_cu1
+	PTRBYTE		.Lend_func_cu1
 
 	.byte		0	/* DW_LNE_end_of_sequence */
 	.uleb128	1
diff --git a/gdb/testsuite/gdb.dwarf2/dw2-const.exp b/gdb/testsuite/gdb.dwarf2/dw2-const.exp
index 93ddfaa0f2f..f855659f8bd 100644
--- a/gdb/testsuite/gdb.dwarf2/dw2-const.exp
+++ b/gdb/testsuite/gdb.dwarf2/dw2-const.exp
@@ -19,8 +19,18 @@ require dwarf2_support
 
 standard_testfile .S
 
+if {[is_ilp32_target]} {
+    set ptrbits 32
+} else {
+    set ptrbits 64
+}
+
+set srcfile_opts {}
+lappend srcfile_opts nodebug
+lappend srcfile_opts additional_flags=-DPTRBITS=$ptrbits
+
 if {[prepare_for_testing_full "failed to prepare" \
-	 [list $testfile debug main.c debug $srcfile nodebug]]} {
+	 [list $testfile debug main.c debug  $srcfile $srcfile_opts]]} {
     return
 }
 
diff --git a/gdb/testsuite/gdb.dwarf2/dw2-skip-prologue.S b/gdb/testsuite/gdb.dwarf2/dw2-skip-prologue.S
index 19165a513a4..11d794bc412 100644
--- a/gdb/testsuite/gdb.dwarf2/dw2-skip-prologue.S
+++ b/gdb/testsuite/gdb.dwarf2/dw2-skip-prologue.S
@@ -17,8 +17,10 @@
 
 #if PTRBITS == 64
 # define PTRBYTE .8byte
+# define PTRBYTES 8
 #elif PTRBITS == 32
 # define PTRBYTE .4byte
+# define PTRBYTES 4
 #else
 # error "Unsupported pointer size"
 #endif
@@ -130,7 +132,7 @@ loclist:
 1:	.byte		0x50	/* DW_OP_reg0 */
 2:
 	/* Location list end.  */
-	.4byte	0, 0
+	PTRBYTE		0, 0
 
 /* Abbrev table */
 	.section .debug_abbrev
@@ -283,7 +285,7 @@ loclist:
 
 .Lline1_lines:
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		func_start
 	.byte		3	/* DW_LNS_advance_line */
@@ -291,7 +293,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		func0
 	.byte		4	/* DW_LNS_set_file */
@@ -301,7 +303,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		func1
 	.byte		4	/* DW_LNS_set_file */
@@ -311,7 +313,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		func2
 	.byte		4	/* DW_LNS_set_file */
@@ -321,7 +323,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		func3
 	.byte		4	/* DW_LNS_set_file */
@@ -331,14 +333,14 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		func_end
 
 /* Equivalent copy but renamed s/func/fund/.  */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		fund_start
 	.byte		3	/* DW_LNS_advance_line */
@@ -346,7 +348,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		fund0
 	.byte		4	/* DW_LNS_set_file */
@@ -356,7 +358,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		fund1
 	.byte		4	/* DW_LNS_set_file */
@@ -366,7 +368,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		fund2
 	.byte		4	/* DW_LNS_set_file */
@@ -376,7 +378,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		fund3
 	.byte		4	/* DW_LNS_set_file */
@@ -386,7 +388,7 @@ loclist:
 	.byte		1	/* DW_LNS_copy */
 
 	.byte		0	/* DW_LNE_set_address */
-	.uleb128	5
+	.uleb128	1 + PTRBYTES
 	.byte		2
 	PTRBYTE		fund_end
 
-- 
2.51.0

Re: [PATCH 2/4] [gdb/testsuite] Complete 64-bit port of gdb.dwarf2/dw2-skip-prologue.exp

[Tom Tromey]

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

Tom> Test-source gdb.dwarf2/dw2-skip-prologue.S was generated using 32-bit, and
Tom> subsequently ported to 64-bit using macro PTRBYTE.

Tom> When using 64-bit, there are some readelf -w warnings because the port to
Tom> 64-bit is not complete.  Fix these warnings.

Tom> Likewise in gdb.dwarf/dw2-const.exp.

This looks good.

Approved-By: Tom Tromey <tom@tromey.com>

I may try converting dw2-const.exp to the DWARF assembler.
dw2-skip-prologue sounds harder, based on the name at least.

Tom

[PATCH 3/4] [gdb/testsuite] Add missing type_offset in gdb.dwarf/debug-names-tu.exp

[Tom de Vries]

With llvm-dwarfdump we run into this warning:
...
$ llvm-dwarfdump -a -v debug-names-tu > /dev/null
warning: DWARF type unit at offset 0x00000028 has its relocated type_offset \
  0x00000028 pointing inside the header
...

Fix this by adding the missing type_offset.

Likewise in gdb.dwarf2/debug-names-bad-cu-index.exp.

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=34157
---
 gdb/testsuite/gdb.dwarf2/debug-names-bad-cu-index.exp | 6 ++++--
 gdb/testsuite/gdb.dwarf2/debug-names-tu.exp.tcl       | 6 ++++--
 2 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/gdb/testsuite/gdb.dwarf2/debug-names-bad-cu-index.exp b/gdb/testsuite/gdb.dwarf2/debug-names-bad-cu-index.exp
index 10e649f06af..5d31414ba29 100644
--- a/gdb/testsuite/gdb.dwarf2/debug-names-bad-cu-index.exp
+++ b/gdb/testsuite/gdb.dwarf2/debug-names-bad-cu-index.exp
@@ -34,6 +34,8 @@ Dwarf::assemble {
 	global $var
     }
 
+    declare_labels type_offset
+
     cu { label cu_label } {
 	compile_unit {
 	    DW_AT_language @DW_LANG_C
@@ -47,11 +49,11 @@ Dwarf::assemble {
 	}
     }
 
-    tu { label tu_label } 0x8ece66f4224fddb3 "" {
+    tu { label tu_label } 0x8ece66f4224fddb3 type_offset {
 	type_unit {} {
 	    declare_labels int_type
 
-	    structure_type {
+	    type_offset: structure_type {
 		DW_AT_name struct_with_int_member
 		DW_AT_byte_size 4 sdata
 	    } {
diff --git a/gdb/testsuite/gdb.dwarf2/debug-names-tu.exp.tcl b/gdb/testsuite/gdb.dwarf2/debug-names-tu.exp.tcl
index af3f263a273..bd48df6a894 100644
--- a/gdb/testsuite/gdb.dwarf2/debug-names-tu.exp.tcl
+++ b/gdb/testsuite/gdb.dwarf2/debug-names-tu.exp.tcl
@@ -39,6 +39,8 @@ Dwarf::assemble {
 	global $var
     }
 
+    declare_labels type_label
+
     cu { label cu_label version $dwarf_version } {
 	compile_unit {
 	    DW_AT_language @DW_LANG_C
@@ -52,13 +54,13 @@ Dwarf::assemble {
 	}
     }
 
-    tu { label tu_label version $dwarf_version } 0x8ece66f4224fddb3 "" {
+    tu { label tu_label version $dwarf_version } 0x8ece66f4224fddb3 type_label {
 	type_unit {
 	    DW_AT_language @DW_LANG_C
 	} {
 	    declare_labels int_type
 
-	    structure_type {
+	    type_label: structure_type {
 		DW_AT_name struct_with_int_member
 		DW_AT_byte_size 4 sdata
 	    } {
-- 
2.51.0

Re: [PATCH 3/4] [gdb/testsuite] Add missing type_offset in gdb.dwarf/debug-names-tu.exp

[Tom Tromey]

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

Tom> With llvm-dwarfdump we run into this warning:
Tom> ...
Tom> $ llvm-dwarfdump -a -v debug-names-tu > /dev/null
Tom> warning: DWARF type unit at offset 0x00000028 has its relocated type_offset \
Tom>   0x00000028 pointing inside the header
Tom> ...

Tom> Fix this by adding the missing type_offset.

Tom> Likewise in gdb.dwarf2/debug-names-bad-cu-index.exp.

Tom> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=34157

This looks good but I wonder if the type offset should simply be
mandatory to 'proc tu'?  Like should it reject an empty string and
require a real label?

Approved-By: Tom Tromey <tom@tromey.com>

Tom

Re: [PATCH 3/4] [gdb/testsuite] Add missing type_offset in gdb.dwarf/debug-names-tu.exp

[Tom de Vries]

On 5/22/26 5:03 PM, Tom Tromey wrote:
>>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:
> 
> Tom> With llvm-dwarfdump we run into this warning:
> Tom> ...
> Tom> $ llvm-dwarfdump -a -v debug-names-tu > /dev/null
> Tom> warning: DWARF type unit at offset 0x00000028 has its relocated type_offset \
> Tom>   0x00000028 pointing inside the header
> Tom> ...
> 
> Tom> Fix this by adding the missing type_offset.
> 
> Tom> Likewise in gdb.dwarf2/debug-names-bad-cu-index.exp.
> 
> Tom> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=34157
> 
> This looks good but I wonder if the type offset should simply be
> mandatory to 'proc tu'?  Like should it reject an empty string and
> require a real label?
> 

I came across fission-type-unit-locexpr.exp, which has:
...
     # Stub for "int" type unit. 

     tu {
         version 4
     } 0xCAFE1 "" {
         type_unit {
             DW_AT_GNU_dwo_name ${::gdb_test_file_name}.dwo DW_FORM_strp
             DW_AT_comp_dir ${outdir}
	} {
         }
     }
...

We could fix this by doing:
...
+    declare_labels none1
      # Stub for "int" type unit.
      tu {
  	version 4
-    } 0xCAFE1 "" {
+    } 0xCAFE1 none1 {
  	type_unit {
  	    DW_AT_GNU_dwo_name ${::gdb_test_file_name}.dwo DW_FORM_strp
  	    DW_AT_comp_dir ${outdir}
  	} {
+	    none1:
  	}
      }
...
and indeed that fixes the warning, but eventually there might be another 
warning complaining about that (points to abbrev number 0 instead of type).

So I left this as is.

Thanks,
- Tom

> Approved-By: Tom Tromey <tom@tromey.com>
> 
> Tom

[PATCH 4/4] [gdb/testsuite] Improve dwarf assembly for implicit const in .debug_names

[Tom de Vries]

In the assembly generated for the dwarf assembly for test-case
gdb.dwarf2/debug-names-tu-dwarf5.exp, there's this bit:
...
.uleb128        0x2003          /* DW_IDX_GNU_language */
.uleb128        0x21            /* DW_FORM_flag_present */
.sleb128        0x0002          /* DW_LANG_C */
...

As it happens, 0x21 is not DW_FORM_flag_present, but DW_FORM_implicit_const,
and the following entry is the value of the attribute.

Fix the comment, and make the relation between the two entries more clear by
printing:
...
.uleb128        0x2003          /* DW_IDX_GNU_language */
.uleb128        0x21            /* DW_FORM_implicit_const */
.sleb128        0x0002          /* DW_FORM_implicit_const value: DW_LANG_C */
...
---
 gdb/testsuite/lib/dwarf.exp | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/gdb/testsuite/lib/dwarf.exp b/gdb/testsuite/lib/dwarf.exp
index 5358dc317a1..9575f4b0b1c 100644
--- a/gdb/testsuite/lib/dwarf.exp
+++ b/gdb/testsuite/lib/dwarf.exp
@@ -3636,8 +3636,9 @@ namespace eval Dwarf {
 		    _op .uleb128 $_constants(DW_IDX_GNU_language) \
 			"DW_IDX_GNU_language"
 		    _op .uleb128 $_constants(DW_FORM_implicit_const) \
-			"DW_FORM_flag_present"
-		    _op .sleb128 $_constants($word) $word
+			"DW_FORM_implicit_const"
+		    _op .sleb128 $_constants($word) \
+			"DW_FORM_implicit_const value: $word"
 		} else {
 		    error "unrecognized extra keyword $word"
 		}
-- 
2.51.0

Re: [PATCH 4/4] [gdb/testsuite] Improve dwarf assembly for implicit const in .debug_names

[Tom Tromey]

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

Tom> Fix the comment, and make the relation between the two entries more clear by
Tom> printing:
Tom> ...
Tom> .uleb128        0x2003          /* DW_IDX_GNU_language */
Tom> .uleb128        0x21            /* DW_FORM_implicit_const */
Tom> .sleb128        0x0002          /* DW_FORM_implicit_const value: DW_LANG_C */

Thank you.
Approved-By: Tom Tromey <tom@tromey.com>

Tom