* config/obj-coffbfd.c (obj_coff_init_stab_section): Zero the
initial stab symbol after allocating it.
* config/obj-elf.c (obj_elf_init_stab_section): Ditto.
This fixes a critical bug observed in Q4 for Lynx.
* write.c (cvt_frag_to_fill): Don't assume that fr_var for rs_align or rs_org
frags will be 1.
(relax_segment): For rs_align, if fr_var is not 1, complain if
required padding is not a multiple of the size of the pad pattern.
(fixup_segment): Leave gp-relative relocations alone. For pcrel relocations
referring to the same segment, clear fx_pcrel when clearing fx_addsy.
* as.h: Adjust comments on rs_align.
* write.c, config/obj-{aout,bout,coff*}.c, config/tc-sparc.c: Query the fx_done
field instead of fx_addsy to see if the fixup still needs to be applied. Set
fx_done and clear fx_addsy both, for now. If TC_HANDLES_FX_DONE isn't defined,
assume md_apply_fix will only clear fx_addsy, and set fx_done accordingly after
returning.
* config/tc-sparc.h (TC_HANDLES_FX_DONE): Define.
* config/obj-coff.c (dot_text_symbol, dot_data_symbol, dot_bss_symbol): Defined
here, static.
* config/obj-aout.c [BFD_ASSEMBLER]: Undef NO_RELOC before including aout/aout64.h.
* write.c (write_object_file): If EMIT_SECTION_SYMBOLS is false, don't write
out a section symbol even if it's used in a relocation; assume relocations will
handle section numbers somehow. Rename "punt_it" label to "punt_it_if_unused"
to reflect it's true use.
(EMIT_SECTION_SYMBOLS): Default to 1.
(adjust_reloc_syms): Don't create a new symbol for an absolute
reference; just use the absolute section symbol.
(write_relocs): Make printout of reloc values dependent on flag DEBUG3, not
DEBUG2.
* config/obj-aout.h (EMIT_SECTION_SYMBOLS): Define as 0.
* config/obj-ecoff.h (EMIT_SECTION_SYMBOLS): Ditto.
const.
* config/obj-*.c, config/tc-*.c: Omit superfluous "return" statements at ends
of functions. Don't check for null return from hash_new, since it won't return
at all if there's no memory available. Also, check for null return from
hash_insert, rather than zero-length string, as success indicator.
to stab section name to get the stab string section name. Pass
the full name of the stab string section to get_stab_string_offset.
* config/obj-elf.c (obj_elf_init_stab_section): Likewise.
physical_input_line, logical_input_line): Made static.
(as_where): Return current file name and line number, don't print
them out.
* messages.c (as_show_where): New static function. Other
functions use it instead of as_where.
(as_bad_internal): New static function.
(as_bad): Use as_bad_internal.
(as_bad_where): New function, like as_bad but taking a file name
and line number.
* as.h (as_bad_where): Declare.
(as_where): Change prototype for new arguments.
* write.h (fixS): Added fields fx_file and fx_line.
* write.c (fix_new_internal): Save file and line number in fix.
(fixup_segment): Use as_bad_where, not as_bad.
* input-file.c (f_in, file_name): Made static.
* cond.c (struct file_line): Just use file and line fields.
(s_else): Use as_where and as_bad_where, not get_file_line and
set_file_line.
(get_file_line, set_file_line): Removed.
* listing.c (listing_newline): Use as_where.
* config/obj-coffbfd.c (obj_coff_init_stab_section): Use as_where.
* config/obj-ecoff.c (add_file): Use as_where.
* config/obj-elf.c (obj_elf_init_stab_section): Use as_where.
* config/tc-m68k.c (md_apply_fix_2): Use as_bad_where.
* config/tc-mips.c (tc_gen_reloc): Use as_bad_where, not assert.
* read.c (change_to_section): Removed. This is now done by
subseg_new.
(get_stab_string_offset): Rearranged somewhat. Create the section
using subseg_new. Store the string index in seg_info, rather than
in a static variable. Force the first string to be empty. Use
frag_more rather than FRAG_APPEND_1_CHAR.
(s_stab_generic): Rewrote.
* subsegs.h (segment_info_type): Added stabu union.
* subsegs.c (subseg_new): Initialize stab_string_size to 0.
* config/obj-aout.c: Don't include aout/stab_gnu.h.
(obj_aout_stab, obj_aout_desc): Removed.
(obj_pseudo_table): Removed desc and stabX entries.
* config/obj-bout.c: Same changes as config/obj-aout.c.
* config/obj-bout.h (S_SET_TYPE): Define.
(tc_bout_fix_to_chars): Declare.
* config/obj-coff.c (obj_coff_stab): Removed.
(obj_pseudo_table): Removed desc and stabX entries.
* config/obj-coff.h (SEPARATE_STAB_SECTIONS): Define.
* config/obj-coffbfd.c (current_stab_symbol): Removed.
* config/obj-coffbfd.h (obj_symbol_type): Removed n_strx, n_type,
n_other, n_desc and n_value fields.
(S_{S,G}ET_{OFFSET,OTHER,TYPE,DESC}): Removed.
(MAKE_STAB_SYMBOL): Removed.
* config/obj-ecoff.c (obj_ecoff_stab): Renamed to ecoff_stab.
Changed arguments and removed parsing code.
(obj_pseudo_table): Removed stabX entries.
* config/obj-ecoff.h (ecoff_stab): Declare.
(OBJ_PROCESS_STAB): Define.
* config/obj-elf.c: Don't include aout/stab_gnu.h.
(obj_elf_stab, obj_elf_xstab, obj_elf_desc,
elf_stab_symbol_string, elf_stab_symbol, obj_elf_stab_generic):
Removed.
(obj_pseudo_table): Removed desc, stabX and xstabs entries.
(obj_elf_version): Use subseg_new, not bfd_make_section. Don't
set SEC_LOAD for .note section.
(adjust_stab_sections): Get frag pointer from seg_info, rather
than looking through frags.
* config/obj-elf.h (S_{S,G}ET_{OTHER,TYPE,DESC}): Removed.
(SEPARATE_STAB_SECTIONS, INIT_STAB_SECTION, OBJ_PROCESS_STAB):
Define.
* config/obj-vms.c (obj_aout_stab): Removed.
(obj_pseudo_table): Removed stabX entries.
* config/obj-vms.h (S_SET_TYPE): Define.
Wrote non-BFD_ASSEMBLER subseg_new. Now subseg_new always takes a
section name, and subseg_set always takes a segT. Changed all
callers as appropriate.
* config/obj-coffbfd.c (change_to_section): Renamed to
obj_coff_add_segment. Corrected. Made callers use subseg_new.
* config/obj-coffbfd.h (obj_segment_name, obj_add_segment):
Define.
Also some more gcc warning removal.
up opcodes as pseudo-ops even if they don't start with '.'.
* config/tc-m88k.h (NO_PSEUDO_DOT): Define.
* config/tc-m88k.c (md_assemble): Removed special pseudo-op
handling.
(md_apply_fix): Set fx_offset to the upper 16 bits of the reloc.
Output the low 16 bits for RELOC_HI16, not the high 16 bits.
* config/obj-coffbfd.c (do_relocs_for): If TC_M88K, set the
r_offset field of the reloc to the fixup offset.
(fixup_segments): If TC_M88K, don't warn about fixup overflows.
* doc/as.texinfo: Minor updates.
undefined symbols.
Fixes PR 2997.
* read.c (s_data), config/obj-coffbfd.c (obj_coff_data): If -R,
switch to text section rather than data section.
Fixes PR 2971.
by md_pcrel_from if we are relocating against a symbol (we still
need md_pcrel_from for a PC relative relocation within the same
file).
* config/tc-m88k.c (md_pcrel_from): Corrected return value.
* configure.in (m88k-*-coff*): New target. Use coffbfd and
m88kcoff.
* config/m88kcoff.mt: New file.
* read.c (lex_type): New macro LEX_AT to set lex type of '@'.
(pseudo_set): Handle difference of symbols in different fragments
by saving the entire expression as the value of the symbol.
* symbols.c (resolve_symbol_value): Resolve difference
expressions.
* config/obj-coffbfd.c (obj_pseudo_table): If TC_M88K, accept
"sdef" as a synonym for "def".
* config/obj-coffbfd.h: If TC_M88K, include coff/m88k.h and set
TARGET_FORMAT.
(S_IS_LOCAL): Any symbol which includes \001 in the name is local.
* config/tc-m88k.c, config/tc-m88k.h: Numerous changes to bring
m88k port up to date, and to add COFF support.
as the value of a symbol.
* struc-symbol.h (struct symbol): Removed sy_forward field. Added
sy_resolved and sy_resolving single bit fields.
* symbols.c (symbol_new): Don't initialize sy_forward field.
(resolve_symbol_value): New function to adjust symbol value by
fragment address, using recursion to resolve forward symbols.
* symbols.h: Added prototype for new function.
* read.c (pseudo_set): Set symbolP->sy_value to an undefined
expression rather than setting symbolP->sy_forward.
* write.c (write_object_file): Use resolve_symbol_value on
symbols, keeping the common case (the old behaviour) inline.
* config/obj-aout.c (obj_aout_frob_symbol): Removed sy_forward
handling (subsumed by write.c change).
* config/obj-coff.c, config/obj-coffbfd.c (obj_coff_val): Set
sy_value rather than sy_forward.
* config/obj-coffbfd.c (obj_coff_endef, yank_symbols): Check
expression segment rather than sy_forward.
(yank_symbols): Use resolve_symbol_value.
(crawl_symbols): Removed extra pass over symbols.
* config/obj-aout.c, config/obj-bout.c, config/obj-coff.c,
config/obj-vms.c (obj_crawl_symbol_chain): Removed extra pass over
symbols which handled sy_forward; use resolve_symbol_value
instead.
* config/obj-coff.h, config/obj-coffbfd.h (obj_frob_forward_symbol):
Define.
* config/obj-elf.c (obj_elf_stab_generic): Check expression
segment rather than sy_forward.
* config/obj-vms.c (VMS_Check_For_Main): Don't initialize
sy_forward; do initialize sy_resolved and sy_resolving.
* config/tc-hppa.h (STAB_FIXUP): Use sy_value, not sy_forward.
just a longword:
* struc-symbol.h: New field sy_value.
* as.h: Include expr.h before struc-symbol.h.
* expr.h: Use struct symbol rather than symbolS.
* symbols.c (S_GET_VALUE, S_SET_VALUE): Rewrote to retrieve value
of sy_value field; compile unconditionally, not just if
BFD_ASSEMBLER.
* symbols.h: Compile S_{SG}ET_VALUE prototypes unconditionally.
* write.c (write_object_file): Set BFD symbol value to gas symbol
value.
* config/obj-aout.h, config/obj-bout.h, config/obj-coff.h,
config/obj-coffbfd.h, config/obj-generic.h, config/obj-vms.h
(S_GET_VALUE, S_SET_VALUE): Removed macro definitions.
* config/obj-ieee.c (S_GET_VALUE, S_SET_VALUE): Removed.
* config/obj-coff.h, obj-coffbfd.h: Rewrote several macros to use
S_GET_VALUE rather than ost_entry.n_value.
* config/obj-aout.c (obj_symbol_to_chars), config/obj-bout.c
(obj_symbol_to_chars), config/obj-coff.c (obj_symbol_to_chars),
config/obj-coffbfd.c (symbol_to_chars): Get value to write out
using S_GET_VALUE--don't assume it is already set.
* config/obj-ieee.c (do_symbols): Set BFD symbol value to gas
symbol value.
* config/obj-vms.c (various): Don't assign directly to
S_GET_VALUE; use S_SET_VALUE instead.
comment_chars.
(do_scrub_next_char): If a line comment character is not at the
start of a line, treat it as a comment character if it is one.
For a CPP line comment use pseudo-op .appline rather than .line.
* input-scrub.c (logical_input_line): Make int rather than
unsigned.
(input_scrub_push, input_scrub_begin): Initialize
logical_input_line to -1 rather than 0.
(bump_line_counters): Increment logical_input_line.
(new_logical_line): If line_number is -2, decrement
logical_input_line.
(as_where): Use logical_input_line even if it is 0.
* read.h (s_app_file prototype): Now takes an int argument.
* read.c (potable): Make .appfile call s_app_file with 1. New
.appline pseudo-op calls s_app_line.
(s_app_file): If .appfile, call new_logical_line with -2 to
account for newline inserted by do_scrub_next_char. If listing,
call listing_source_file.
(s_app_line): New function to handle fake pseudo-op .appline.
* config/obj-coff.c (obj_pseudo_table): Make .appline call
obj_coff_ln.
(obj_coff_ln): Added argument to indicate whether .appline.
* config/obj-coffbfd.c (obj_pseudo_table): Make .appline call
obj_coff_ln.
(obj_coff_ln): Added argument to indicate whether .appline.
* config/tc-mips.c (s_file): Pass argument to s_app_file.
wasn't being stored into scnhdr.
* config/obj-coffbfd.h: Add prototype of s_get_segment.
* read.c (TC_START_LABEL): Default definition.
(read_a_source_file): Use TC_START_LABEL macro to work out
if a label has been seen.
* config/obj-coffbfd.c (fill_section): Don't set vaddr here.
(write_object_file): Set it here instead, so that fixup_segment
can see the correct value.
* listing.c: Call xmalloc, not malloc; don't declare malloc.
* Changes to support SCO 3.2v4:
* read.c (s_align_bytes, s_align_ptwo): If not SEG_DATA or
SEG_BSS, fill with NOP_OPCODE.
* config/i386coff.mt: Add opcode/i386.h to TARG_CPU_DEPENDENTS.
* config/obj-coffbfd.c (do_relocs_for): Increment addr even if not
using ZERO_BASED_SEGMENTS.
(fill_section): If ZERO_BASED_SEGMENTS, set segment addresses, but
never set segment address for SEG_E2 (.bss) and don't write out
SEG_E2 contents. Set .init and .fini sections to STYP_TEXT.
(obj_coff_endef): Don't merge labels or symbols awaiting forward
definitions, and don't merge tags with non-tags. Check for .bf
rather than just checking whether the second character is b and
the third character is f.
(obj_coff_val): gcc can generate values which we don't handle
correctly; discard information for now, since it only affects the
debugging information.
(tag_find_or_name): Don't insert tags in the symbol table.
(yank_symbols): Don't merge labels.
(write_object_file): Don't define SUB_SEGMENT_ALIGN if it is
already defined. Fill subsegments with NOP_OPCODE, not 0. Don't
set segment address if ZERO_BASED_SEGMENTS.
(obj_coff_section): Accept and ignore a trailing quoted string, as
used in AT&T i386 syntax.
(fixup_segment): Take segment as argument. On the i386, adjust PC
relative addends by the segment vaddr.
* tc-i386.h: Define SUB_SEGMENT_ALIGN.
* tc-a29k.h: Define ZERO_BASED_SEGMENTS.
* tc-i386.c: (i386_operand): If I386COFF, accept any segment type.
Removed some unneeded files.
obj-coff.c (obj_coff_endef): Use as_warn, not fprintf.
tc-m68k.c (md_assemble): 68000+68881 is okay -- could be emulating.
support when using /opt/SUNWspro/SC2.0/acc.
Tue Nov 10 09:50:25 1992 Ian Lance Taylor (ian@cygnus.com)
* obj-coffbfd.c (fill_section): set STYP_NOLOAD bit for .bss
section.
* atof-ieee.c, atof-ns32k.c, tc-*.c: made EXP_CHARS, FLT_CHARS,
comment_chars, line_comment_chars and line_seperator_chars
consistently const, and always initialized them. Included read.h.
* Support for i386-sysv.
obj-coffbfd.c (do_relocs_for, write_object_file): set segment
addresses to reasonable sizes. New define ZERO_BASED_SEGMENTS can
be used to set them all to zero as was done before.
(fill_section): segment addresses now set in write_object_file.
(fill_section): Don't set STYP_NOLOAD for .bss section.
(fixup_segment): 386 uses strange common symbol format.
tc-i386.c (tc_coff_fix2rtype): use R_DIR32, not R_RELLONG, for
compatibility with SVR3.2 linker.
* obj-coffbfd.c (do_relocs_for,fill_section): now allocate all
sections starting from zero, rather than making them consecutive.
This makes subsequent reloc calculations easier, esp if the
object format doesn't understand addends. (obj_coff_lcomm): (maybe
temporarily) allocate lcomm in .data rather than in .bss. It
seems that some tools can't cope with a non-zero sized bss
before linkage.
