On 3/12/21 3:20 AM, Jakub Jelinek wrote:
Hi!
For -gdwarf-4 -gsplit-dwarf we used to emit .debug_ranges section
(so in the binaries/shared libraries) with DW_AT_ranges from skeleton
units as well as .debug_info.dwo pointing to it through DW_FORM_sec_offset
(and DW_AT_GNU_ranges_base pointing into section, not sure for what
reason exactly).
When DWARF5 support was being added, we've started using .debug_rnglists
section, added DW_AT_rnglists_base to the DW_TAG_skeleton_unit, kept
DW_AT_ranges with DW_FORM_sec_offset in the skeleton and switched
over to DW_FORM_rnglistx for DW_AT_ranges in .debug_info.dwo.
But the DWARF5 spec actually means for the ranges section (at least
everything for those DW_AT_ranges in .debug_info.dwo) to sit
in .debug_rnglists.dwo section next to the .debug_info.dwo, rather than
having consumers look it up in the binary/shared library instead.
Based on some discussions in the DWARF discuss mailing list:
http://lists.dwarfstd.org/pipermail/dwarf-discuss-dwarfstd.org/2021-March/thread.html#4765
this patch mostly follows what LLVM emits for that right now:
1) small .debug_rnglists section (when needed) just to cover the
skeleton DW_AT_ranges (if present); the content of the section
uses the Split DWARFy DW_RLE_* codes with addrx encodings where
possible
2) DW_AT_ranges in the skeleton uses DW_FORM_sec_offset (difference
from LLVM which uses DW_FORM_rnglistx, which makes it larger
and ambiguous)
3) DW_AT_rnglists_base attribute is gone from the skeleton (again,
unlike LLVM where it is just confusing what exactly it means because
it is inherited; it would make sense if we emitted DW_FORM_rnglistx
in non-split DWARF, but unless ranges are shared, I'm afraid we'd
make DWARF larger with fewer relocations by that)
4) usually big .debug_rnglists.dwo section again with using DW_RLE_*x*
where possible
5) DW_AT_ranges with DW_FORM_rnglistx from .debug_info.dwo referring to
that .debug_rnglists.dwo ranges
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2021-03-12 Jakub Jelinek <ja...@redhat.com>
PR debug/99490
* dwarf2out.c (debug_ranges_dwo_section): New variable.
(DW_RANGES_IDX_SKELETON): Define.
(struct dw_ranges): Add begin_entry and end_entry members.
(DEBUG_DWO_RNGLISTS_SECTION): Define.
(add_ranges_num): Adjust r initializer for addition of *_entry
members.
(add_ranges_by_labels): For -gsplit-dwarf and force_direct,
set idx to DW_RANGES_IDX_SKELETON.
(index_rnglists): Don't set r->idx if it is equal to
DW_RANGES_IDX_SKELETON. Initialize r->begin_entry and
r->end_entry for -gsplit-dwarf if those will be needed by
output_rnglists.
(output_rnglists): Add DWO argument. If true, switch to
debug_ranges_dwo_section rather than debug_ranges_section.
Adjust l1/l2 label indexes. Only output the offset table when
dwo is true and don't include in there the skeleton range
entry if present. For -gsplit-dwarf, skip ranges that belong
to the other rnglists section. Change return type from void
to bool and return true if there are any range entries for
the other section. For dwarf_split_debug_info use
DW_RLE_startx_endx, DW_RLE_startx_length and DW_RLE_base_addressx
entries instead of DW_RLE_start_end, DW_RLE_start_length and
DW_RLE_base_address.
(init_sections_and_labels): Initialize debug_ranges_dwo_section
if -gsplit-dwarf and DWARF >= 5. Adjust ranges_section_label
and range_base_label indexes.
(dwarf2out_finish): Call index_rnglists earlier before finalizing
.debug_addr. Never emit DW_AT_rnglists_base attribute. For
-gsplit-dwarf and DWARF >= 5 call output_rnglists up to twice
with different dwo arguments.
(dwarf2out_c_finalize): Clear debug_ranges_dwo_section.
--- gcc/dwarf2out.c.jj 2021-03-10 17:36:37.037537129 +0100
+++ gcc/dwarf2out.c 2021-03-11 12:50:00.402418873 +0100
@@ -171,6 +171,7 @@ static GTY(()) section *debug_line_str_s
static GTY(()) section *debug_str_dwo_section;
static GTY(()) section *debug_str_offsets_section;
static GTY(()) section *debug_ranges_section;
+static GTY(()) section *debug_ranges_dwo_section;
static GTY(()) section *debug_frame_section;
/* Maximum size (in bytes) of an artificially generated label. */
@@ -3152,11 +3153,17 @@ struct GTY(()) dw_ranges {
/* If this is positive, it's a block number, otherwise it's a
bitwise-negated index into dw_ranges_by_label. */
int num;
+ /* If idx is equal to DW_RANGES_IDX_SKELETON, it should be emitted
+ into .debug_rnglists section rather than .debug_rnglists.dwo
+ for -gsplit-dwarf and DWARF >= 5. */
+#define DW_RANGES_IDX_SKELETON ((1U << 31) - 1)
/* Index for the range list for DW_FORM_rnglistx. */
unsigned int idx : 31;
/* True if this range might be possibly in a different section
from previous entry. */
unsigned int maybe_new_sec : 1;
+ addr_table_entry *begin_entry;
+ addr_table_entry *end_entry;
};
/* A structure to hold a macinfo entry. */
@@ -4099,6 +4106,9 @@ new_addr_loc_descr (rtx addr, enum dtpre
#ifndef DEBUG_RNGLISTS_SECTION
#define DEBUG_RNGLISTS_SECTION ".debug_rnglists"
#endif
+#ifndef DEBUG_DWO_RNGLISTS_SECTION
+#define DEBUG_DWO_RNGLISTS_SECTION ".debug_rnglists.dwo"
+#endif
#ifndef DEBUG_LINE_STR_SECTION
#define DEBUG_LINE_STR_SECTION ".debug_line_str"
#endif
@@ -11751,7 +11761,7 @@ output_aranges (void)
static unsigned int
add_ranges_num (int num, bool maybe_new_sec)
{
- dw_ranges r = { NULL, num, 0, maybe_new_sec };
+ dw_ranges r = { NULL, num, 0, maybe_new_sec, NULL, NULL };
vec_safe_push (ranges_table, r);
return vec_safe_length (ranges_table) - 1;
}
@@ -11796,6 +11806,8 @@ add_ranges_by_labels (dw_die_ref die, co
add_AT_range_list (die, DW_AT_ranges, offset, force_direct);
*added = true;
note_rnglist_head (offset);
+ if (dwarf_split_debug_info && force_direct)
+ (*ranges_table)[offset].idx = DW_RANGES_IDX_SKELETON;
}
}
@@ -11921,23 +11933,87 @@ asm_outputs_debug_line_str (void)
}
-/* Assign .debug_rnglists indexes. */
+/* Assign .debug_rnglists indexes and unique indexes into the debug_addr
+ section when needed. */
static void
index_rnglists (void)
{
unsigned i;
dw_ranges *r;
+ bool base = false;
+ unsigned int len = vec_safe_length (ranges_table);
FOR_EACH_VEC_SAFE_ELT (ranges_table, i, r)
- if (r->label)
- r->idx = rnglist_idx++;
+ {
+ if (r->label && r->idx != DW_RANGES_IDX_SKELETON)
+ r->idx = rnglist_idx++;
+
+ if (!have_multiple_function_sections)
+ continue;
+ int block_num = r->num;
+ if (HAVE_AS_LEB128 && (r->label || r->maybe_new_sec))
+ base = false;
+ if (block_num > 0)
+ {
+ char blabel[MAX_ARTIFICIAL_LABEL_BYTES];
+ char elabel[MAX_ARTIFICIAL_LABEL_BYTES];
+
+ ASM_GENERATE_INTERNAL_LABEL (blabel, BLOCK_BEGIN_LABEL, block_num);
+ ASM_GENERATE_INTERNAL_LABEL (elabel, BLOCK_END_LABEL, block_num);
+
+ if (HAVE_AS_LEB128)
+ {
+ if (!base)
+ {
+ dw_ranges *r2 = NULL;
+ if (i < len - 1)
+ r2 = &(*ranges_table)[i + 1];
+ if (r2
+ && r2->num != 0
+ && r2->label == NULL
+ && !r2->maybe_new_sec)
Looks like this logic is copied from output_rnglists; can we factor it
out? Say, range_needs_distinct_base_address (r2) or (i+1)?
It would also be good to have a comment about why there's no end_entry
in this case.
+ {
+ r->begin_entry
+ = add_addr_table_entry (xstrdup (blabel),
+ ate_kind_label);
+ base = true;
+ }
+ }
+ if (base)
+ continue;
+ r->begin_entry
+ = add_addr_table_entry (xstrdup (blabel), ate_kind_label);
+ }
+ else
+ {
+ r->begin_entry
+ = add_addr_table_entry (xstrdup (blabel), ate_kind_label);
+ r->end_entry
+ = add_addr_table_entry (xstrdup (elabel), ate_kind_label);
+ }
+ }
+
+ /* Negative block_num stands for an index into ranges_by_label. */
+ else if (block_num < 0)
+ {
+ int lab_idx = - block_num - 1;
+ const char *blabel = (*ranges_by_label)[lab_idx].begin;
+ const char *elabel = (*ranges_by_label)[lab_idx].end;
+
+ r->begin_entry
+ = add_addr_table_entry (xstrdup (blabel), ate_kind_label);
+ if (!HAVE_AS_LEB128)
+ r->end_entry
+ = add_addr_table_entry (xstrdup (elabel), ate_kind_label);
+ }
+ }
}
-/* Emit .debug_rnglists section. */
+/* Emit .debug_rnglists or (when DWO is true) .debug_rnglists.dwo section. */
-static void
-output_rnglists (unsigned generation)
+static bool
+output_rnglists (unsigned generation, bool dwo)
{
unsigned i;
dw_ranges *r;
@@ -11945,14 +12021,19 @@ output_rnglists (unsigned generation)
char l2[MAX_ARTIFICIAL_LABEL_BYTES];
char basebuf[MAX_ARTIFICIAL_LABEL_BYTES];
- switch_to_section (debug_ranges_section);
- ASM_OUTPUT_LABEL (asm_out_file, ranges_section_label);
+ if (dwo)
+ switch_to_section (debug_ranges_dwo_section);
+ else
+ {
+ switch_to_section (debug_ranges_section);
+ ASM_OUTPUT_LABEL (asm_out_file, ranges_section_label);
+ }
/* There are up to 4 unique ranges labels per generation.
See also init_sections_and_labels. */
ASM_GENERATE_INTERNAL_LABEL (l1, DEBUG_RANGES_SECTION_LABEL,
- 2 + generation * 4);
+ 2 + 2 * dwo + generation * 6);
ASM_GENERATE_INTERNAL_LABEL (l2, DEBUG_RANGES_SECTION_LABEL,
- 3 + generation * 4);
+ 3 + 2 * dwo + generation * 6);
if (DWARF_INITIAL_LENGTH_SIZE - dwarf_offset_size == 4)
dw2_asm_output_data (4, 0xffffffff,
"Initial length escape value indicating "
@@ -11969,13 +12050,13 @@ output_rnglists (unsigned generation)
the offset table plus corresponding DW_FORM_rnglistx uleb128 indexes
into it are usually larger than just DW_FORM_sec_offset offsets
into the .debug_rnglists section. */
- dw2_asm_output_data (4, dwarf_split_debug_info ? rnglist_idx : 0,
+ dw2_asm_output_data (4, dwo ? rnglist_idx : 0,
"Offset Entry Count");
- if (dwarf_split_debug_info)
+ if (dwo)
{
ASM_OUTPUT_LABEL (asm_out_file, ranges_base_label);
FOR_EACH_VEC_SAFE_ELT (ranges_table, i, r)
- if (r->label)
+ if (r->label && r->idx != DW_RANGES_IDX_SKELETON)
dw2_asm_output_delta (dwarf_offset_size, r->label,
ranges_base_label, NULL);
}
@@ -11983,15 +12064,32 @@ output_rnglists (unsigned generation)
const char *lab = "";
unsigned int len = vec_safe_length (ranges_table);
const char *base = NULL;
+ bool skipping = false;
+ bool ret = false;
FOR_EACH_VEC_SAFE_ELT (ranges_table, i, r)
{
int block_num = r->num;
if (r->label)
{
+ if (dwarf_split_debug_info)
+ {
+ if ((r->idx == DW_RANGES_IDX_SKELETON) == dwo)
These two ifs could be combined.
+ {
+ ret = true;
+ skipping = true;
+ continue;
+ }
+ }
ASM_OUTPUT_LABEL (asm_out_file, r->label);
lab = r->label;
}
+ if (skipping)
+ {
+ if (block_num == 0)
+ skipping = false;
+ continue;
+ }
if (HAVE_AS_LEB128 && (r->label || r->maybe_new_sec))
base = NULL;
if (block_num > 0)
@@ -12027,10 +12125,23 @@ output_rnglists (unsigned generation)
&& r2->label == NULL
&& !r2->maybe_new_sec)
{
- dw2_asm_output_data (1, DW_RLE_base_address,
- "DW_RLE_base_address (%s)", lab);
- dw2_asm_output_addr (DWARF2_ADDR_SIZE, blabel,
- "Base address (%s)", lab);
+ if (dwarf_split_debug_info)
+ {
+ dw2_asm_output_data (1, DW_RLE_base_addressx,
+ "DW_RLE_base_addressx (%s)",
+ lab);
+ dw2_asm_output_data_uleb128 (r->begin_entry->index,
+ "Base address index "
+ "(%s)", blabel);
+ }
+ else
+ {
+ dw2_asm_output_data (1, DW_RLE_base_address,
+ "DW_RLE_base_address (%s)",
+ lab);
+ dw2_asm_output_addr (DWARF2_ADDR_SIZE, blabel,
+ "Base address (%s)", lab);
+ }
strcpy (basebuf, blabel);
base = basebuf;
}
@@ -12045,13 +12156,35 @@ output_rnglists (unsigned generation)
"Range end address (%s)", lab);
continue;
}
- dw2_asm_output_data (1, DW_RLE_start_length,
- "DW_RLE_start_length (%s)", lab);
- dw2_asm_output_addr (DWARF2_ADDR_SIZE, blabel,
- "Range begin address (%s)", lab);
+ if (dwarf_split_debug_info)
+ {
+ dw2_asm_output_data (1, DW_RLE_startx_length,
+ "DW_RLE_startx_length (%s)", lab);
+ dw2_asm_output_data_uleb128 (r->begin_entry->index,
+ "Range begin address index "
+ "(%s)", blabel);
+ }
+ else
+ {
+ dw2_asm_output_data (1, DW_RLE_start_length,
+ "DW_RLE_start_length (%s)", lab);
+ dw2_asm_output_addr (DWARF2_ADDR_SIZE, blabel,
+ "Range begin address (%s)", lab);
+ }
dw2_asm_output_delta_uleb128 (elabel, blabel,
"Range length (%s)", lab);
}
+ else if (dwarf_split_debug_info)
+ {
+ dw2_asm_output_data (1, DW_RLE_startx_endx,
+ "DW_RLE_startx_endx (%s)", lab);
+ dw2_asm_output_data_uleb128 (r->begin_entry->index,
+ "Range begin address index "
+ "(%s)", blabel);
+ dw2_asm_output_data_uleb128 (r->end_entry->index,
+ "Range end address index "
+ "(%s)", elabel);
+ }
else
{
dw2_asm_output_data (1, DW_RLE_start_end,
@@ -12074,13 +12207,35 @@ output_rnglists (unsigned generation)
gcc_unreachable ();
if (HAVE_AS_LEB128)
{
- dw2_asm_output_data (1, DW_RLE_start_length,
- "DW_RLE_start_length (%s)", lab);
- dw2_asm_output_addr (DWARF2_ADDR_SIZE, blabel,
- "Range begin address (%s)", lab);
+ if (dwarf_split_debug_info)
+ {
+ dw2_asm_output_data (1, DW_RLE_startx_length,
+ "DW_RLE_startx_length (%s)", lab);
+ dw2_asm_output_data_uleb128 (r->begin_entry->index,
+ "Range begin address index "
+ "(%s)", blabel);
+ }
+ else
+ {
+ dw2_asm_output_data (1, DW_RLE_start_length,
+ "DW_RLE_start_length (%s)", lab);
+ dw2_asm_output_addr (DWARF2_ADDR_SIZE, blabel,
+ "Range begin address (%s)", lab);
+ }
dw2_asm_output_delta_uleb128 (elabel, blabel,
"Range length (%s)", lab);
}
+ else if (dwarf_split_debug_info)
+ {
+ dw2_asm_output_data (1, DW_RLE_startx_endx,
+ "DW_RLE_startx_endx (%s)", lab);
+ dw2_asm_output_data_uleb128 (r->begin_entry->index,
+ "Range begin address index "
+ "(%s)", blabel);
+ dw2_asm_output_data_uleb128 (r->end_entry->index,
+ "Range end address index "
+ "(%s)", elabel);
+ }
else
{
dw2_asm_output_data (1, DW_RLE_start_end,
@@ -12096,6 +12251,7 @@ output_rnglists (unsigned generation)
"DW_RLE_end_of_list (%s)", lab);
}
ASM_OUTPUT_LABEL (asm_out_file, l2);
+ return ret;
}
/* Data structure containing information about input files. */
@@ -28823,6 +28979,10 @@ init_sections_and_labels (bool early_lto
debug_macinfo_section = get_section (debug_macinfo_section_name,
SECTION_DEBUG | SECTION_EXCLUDE,
NULL);
+ if (dwarf_version >= 5)
+ debug_ranges_dwo_section
+ = get_section (DEBUG_DWO_RNGLISTS_SECTION,
+ SECTION_DEBUG | SECTION_EXCLUDE, NULL);
}
debug_aranges_section = get_section (DEBUG_ARANGES_SECTION,
SECTION_DEBUG, NULL);
@@ -28857,15 +29017,15 @@ init_sections_and_labels (bool early_lto
ASM_GENERATE_INTERNAL_LABEL (debug_line_section_label,
DEBUG_LINE_SECTION_LABEL,
init_sections_and_labels_generation);
- /* There are up to 4 unique ranges labels per generation.
+ /* There are up to 6 unique ranges labels per generation.
See also output_rnglists. */
ASM_GENERATE_INTERNAL_LABEL (ranges_section_label,
DEBUG_RANGES_SECTION_LABEL,
- init_sections_and_labels_generation * 4);
+ init_sections_and_labels_generation * 6);
if (dwarf_version >= 5 && dwarf_split_debug_info)
ASM_GENERATE_INTERNAL_LABEL (ranges_base_label,
DEBUG_RANGES_SECTION_LABEL,
- 1 + init_sections_and_labels_generation * 4);
+ 1 + init_sections_and_labels_generation * 6);
ASM_GENERATE_INTERNAL_LABEL (debug_addr_section_label,
DEBUG_ADDR_SECTION_LABEL,
init_sections_and_labels_generation);
@@ -31665,6 +31825,9 @@ dwarf2out_finish (const char *filename)
index_location_lists (comp_unit_die ());
}
+ if (dwarf_version >= 5 && !vec_safe_is_empty (ranges_table))
+ index_rnglists ();
+
if (addr_index_table != NULL)
{
unsigned int index = 0;
@@ -31730,9 +31893,6 @@ dwarf2out_finish (const char *filename)
int mark;
struct md5_ctx ctx;
- if (dwarf_version >= 5 && !vec_safe_is_empty (ranges_table))
- index_rnglists ();
-
/* Compute a checksum of the comp_unit to use as the dwo_id. */
md5_init_ctx (&ctx);
mark = 0;
@@ -31752,10 +31912,7 @@ dwarf2out_finish (const char *filename)
comp-unit DIE. */
if (!vec_safe_is_empty (ranges_table))
{
- if (dwarf_version >= 5)
- add_AT_lineptr (main_comp_unit_die, DW_AT_rnglists_base,
- ranges_base_label);
- else
+ if (dwarf_version < 5)
add_AT_lineptr (main_comp_unit_die, DW_AT_GNU_ranges_base,
ranges_section_label);
}
@@ -31833,7 +31990,26 @@ dwarf2out_finish (const char *filename)
if (!vec_safe_is_empty (ranges_table))
{
if (dwarf_version >= 5)
- output_rnglists (generation);
+ {
+ if (dwarf_split_debug_info)
+ {
+ /* We don't know right now whether there are any
+ ranges for .debug_rnglists and any for .debug_rnglists.dwo.
+ Depending on into which of those two belongs the first
+ ranges_table entry, emit that section first and that
+ output_rnglists call will return if the other kind of
Maybe "will return true if..."
+ ranges needs to be emitted as well. */
+ if ((*ranges_table)[0].idx == DW_RANGES_IDX_SKELETON)
+ {
+ if (output_rnglists (generation, false))
+ output_rnglists (generation, true);
+ }
+ else if (output_rnglists (generation, true))
+ output_rnglists (generation, false);
Maybe
bool dwo = (*ranges_table)[0].idx != DW_RANGES_IDX_SKELETON;
if (output_rnglists (generation, dwo))
output_rnglists (generation, !dwo);
OK with these changes.
Jason
