external/binutils.git
30 years agoSat Feb 5 12:30:32 1994 Stan Shebs (shebs@andros.cygnus.com)
Stan Shebs [Sat, 5 Feb 1994 21:45:51 +0000 (21:45 +0000)]
Sat Feb  5 12:30:32 1994  Stan Shebs  (shebs@andros.cygnus.com)

* 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.

30 years ago * gdb.t17/interrupt.c: If we get EINTR, don't print an error message.
Jim Kingdon [Sat, 5 Feb 1994 21:29:45 +0000 (21:29 +0000)]
* gdb.t17/interrupt.c: If we get EINTR, don't print an error message.

* gdb.t17/callfuncs.exp: Add xfails for rs6000.

30 years ago* config.bfd: Put m68*-*-sysv* line after m68*-*-sysv4*.
Jim Kingdon [Sat, 5 Feb 1994 20:41:44 +0000 (20:41 +0000)]
* config.bfd: Put m68*-*-sysv* line after m68*-*-sysv4*.

30 years ago * config/i386/tm-i386aix.h (REGISTER_CONVERT_TO_RAW): Add missing
Jim Kingdon [Sat, 5 Feb 1994 20:38:12 +0000 (20:38 +0000)]
* config/i386/tm-i386aix.h (REGISTER_CONVERT_TO_RAW): Add missing
backslash.
* i386-tdep.c (i386_extract_return_value): Pass TYPE_LENGTH (type)
to store_floating, not nonexistent variable len.

30 years ago * remote-mips.c (mips_insert_breakpoint, mips_remove_breakpoint):
Jim Kingdon [Sat, 5 Feb 1994 20:30:35 +0000 (20:30 +0000)]
* remote-mips.c (mips_insert_breakpoint, mips_remove_breakpoint):
New functions.
(mips_store_word): Change calling convention to return errors, and
to provide old contents if the caller wants it.
(mips_xfer_memory): Deal with errors from mips_store_word.
* config/mips/tm-idt.h, config/mips/tm-idtl.h: Remove BREAKPOINT
define now that remote-mips.c doesn't use BREAKPOINT.

30 years ago * config/tc-hppa.c (pa_ip): addb[tf] should only accept
Jeff Law [Sat, 5 Feb 1994 19:54:51 +0000 (19:54 +0000)]
    * config/tc-hppa.c (pa_ip): addb[tf] should only accept
        non-negated condition completers.  Add support for addb pseudo-op
        which accepts both negated and non-negated completers.

30 years ago * hppa.h (addb): Use '@' for addb and addib pseudo ops.
Jeff Law [Sat, 5 Feb 1994 19:53:05 +0000 (19:53 +0000)]
    * hppa.h (addb): Use '@' for addb and addib pseudo ops.
        (so we can determine valid completers for both addb and addb[tf].)

30 years ago * gdb.disasm/hppa.s (comib_tests): Add missing instruction.
Jeff Law [Sat, 5 Feb 1994 19:16:49 +0000 (19:16 +0000)]
    * gdb.disasm/hppa.s (comib_tests): Add missing instruction.
        (comib_nullified_tests): Fix typo.
        * gdb.disasm/hppa.exp (all comb tests): Fix thinkos in expect
        strings.  No longer expected to fail.

30 years ago * hppa-dis.c (print_insn_hppa): For '?' and '@' only adjust the
Jeff Law [Sat, 5 Feb 1994 19:15:20 +0000 (19:15 +0000)]
    * hppa-dis.c (print_insn_hppa): For '?' and '@' only adjust the
        condition name index if this is for a negated condition.

30 years ago * remote-e7000.c: Replace "snoop" command (e7000_noecho) with
Jim Kingdon [Sat, 5 Feb 1994 18:50:52 +0000 (18:50 +0000)]
* remote-e7000.c: Replace "snoop" command (e7000_noecho) with
remote_debug.

30 years ago * gdb.disasm/hppa.s (addib_tests): Fix typo.
Jeff Law [Sat, 5 Feb 1994 18:40:20 +0000 (18:40 +0000)]
    * gdb.disasm/hppa.s (addib_tests): Fix typo.
        * gdb.disasm/hppa.exp (all addb tests): Fix thinkos in expect
        strings.  No longer expected to fail.

30 years ago * gdb.t24/demangle.exp: Re-write demangle to just call gdb_test,
Jim Kingdon [Sat, 5 Feb 1994 18:34:33 +0000 (18:34 +0000)]
* gdb.t24/demangle.exp: Re-write demangle to just call gdb_test,
properly quoting regexps.  This makes the second argument to
demangle a fixed string, which matches the callers, instead of a
shell-style pattern which is what the previous implementation
used.  Update some of the callers.

30 years ago * gdb.disasm/hppa.exp (fmpy_addsub_tests): Fix typo in expect string.
Jeff Law [Sat, 5 Feb 1994 18:20:02 +0000 (18:20 +0000)]
    * gdb.disasm/hppa.exp (fmpy_addsub_tests): Fix typo in expect string.
        No longer expected to fail.

30 years ago * hppa-dis.c (print_insn_hppa): No space before 'H' operand.
Jeff Law [Sat, 5 Feb 1994 18:18:49 +0000 (18:18 +0000)]
    * hppa-dis.c (print_insn_hppa): No space before 'H' operand.
        Floating point format for 'H' operand is backwards from normal
        case (0 == double, 1 == single).  For '4', '6', '7', '9', and '8'
        operands (fmpyadd and fmpysub), handle bizarre register translation
        correctly for single precision format.

30 years ago * Makefile.in: Remove obsolete rules for history.info and
Jim Kingdon [Sat, 5 Feb 1994 16:37:44 +0000 (16:37 +0000)]
* Makefile.in: Remove obsolete rules for history.info and
readline.info.

30 years ago * config/rs6000/tm-rs6000.h (STORE_STRUCT_RETURN): Don't cast
Jim Kingdon [Sat, 5 Feb 1994 16:12:18 +0000 (16:12 +0000)]
* config/rs6000/tm-rs6000.h (STORE_STRUCT_RETURN): Don't cast
to unsigned int.

30 years ago * srec.c (srec_write_record): Put CONST keyword for "src" before
Peter Schauer [Sat, 5 Feb 1994 13:38:52 +0000 (13:38 +0000)]
* srec.c (srec_write_record):  Put CONST keyword for "src" before
"unsigned", some compilers don't like it after "unsigned".
* libcoff.h, libcoff-in.h (bfd_perform_slip):  Rename "value" to
"val" in prototype declaration because some compilers don't like
arguments whose names are the same as types.

30 years ago * value.h (print_longest): Rename "value" to "val" in prototype
Peter Schauer [Sat, 5 Feb 1994 13:37:00 +0000 (13:37 +0000)]
* value.h (print_longest):  Rename "value" to "val" in prototype
declaration because some compilers don't like arguments whose
names are the same as types.
* remote.c (remote_xfer_memory):  Cast "myaddr" to unsigned char *
before passing it to remote_*_bytes.

30 years ago * emultempl/lnk960.em (append), emultempl/hppaosf.em
David MacKenzie [Sat, 5 Feb 1994 12:00:15 +0000 (12:00 +0000)]
* emultempl/lnk960.em (append), emultempl/hppaosf.em
(hppaosf_finish): Call xmalloc, not ldmalloc.
* ldmain.c (preserve_output): Function removed.
(main): Do it here instead.

30 years ago * hppa-dis.c (print_insn_hppa): Do not emit a space after 'F'
Jeff Law [Sat, 5 Feb 1994 08:04:45 +0000 (08:04 +0000)]
    * hppa-dis.c (print_insn_hppa): Do not emit a space after 'F'
        or 'I' operands if the next format specifier is 'M' (fcmp
        condition completer).

30 years ago * gdb.disasm/hppa.exp (fcmp_tests): Fix thinko in expect string.
Jeff Law [Sat, 5 Feb 1994 08:03:34 +0000 (08:03 +0000)]
    * gdb.disasm/hppa.exp (fcmp_tests): Fix thinko in expect string.
        No longer expected to fail.

30 years ago * gdb.disasm/hppa.exp (xmpyu_tests): Fix buglet in expect
Jeff Law [Sat, 5 Feb 1994 07:46:20 +0000 (07:46 +0000)]
    * gdb.disasm/hppa.exp (xmpyu_tests): Fix buglet in expect
        string.  No longer expected to fail.

30 years ago * hppa.h (xmpyu): No floating point format specifier for the
Jeff Law [Sat, 5 Feb 1994 07:44:34 +0000 (07:44 +0000)]
    * hppa.h (xmpyu): No floating point format specifier for the
        xmpyu instruction.

30 years ago * gas/hppa/more.parse/xmpyubug.s: New test.
Jeff Law [Sat, 5 Feb 1994 07:43:29 +0000 (07:43 +0000)]
    * gas/hppa/more.parse/xmpyubug.s: New test.
        * gas/hppa/more.parse/parse.exp: Run it.

30 years ago * ldlang.h (LANG_FOR_EACH_{INPUT,OUTPUT}_SECTION): Delete (unused)
Jeff Law [Sat, 5 Feb 1994 07:03:48 +0000 (07:03 +0000)]
    * ldlang.h (LANG_FOR_EACH_{INPUT,OUTPUT}_SECTION): Delete (unused)
        GNU C specific macros.

        * emultempl/hppaosf.em (hppaosf_finish): Expand the only remaining
        call to LANG_FOR_EACH_INPUT_SECTION.

30 years ago * aoutx.h (aout_link_check_ar_symbols): Correct test for whether
Ian Lance Taylor [Sat, 5 Feb 1994 06:17:36 +0000 (06:17 +0000)]
* aoutx.h (aout_link_check_ar_symbols): Correct test for whether
object file defines symbol.  Also, if skipping a symbol, skip the
second symbol of a N_WARNING or N_INDR symbol as well.

30 years ago * configure.in (rs6000-*-*): Build gas.
Ian Lance Taylor [Sat, 5 Feb 1994 06:01:33 +0000 (06:01 +0000)]
* configure.in (rs6000-*-*): Build gas.

30 years ago * configure.in (rs6000*): Use cpu_type ppc.
Ian Lance Taylor [Sat, 5 Feb 1994 05:38:24 +0000 (05:38 +0000)]
* configure.in (rs6000*): Use cpu_type ppc.
(ppc-*-aix*): New target; use coff and force bfd_gas.

30 years agoNew files for PowerPC/POWER (RS/6000) support.
Ian Lance Taylor [Sat, 5 Feb 1994 05:35:11 +0000 (05:35 +0000)]
New files for PowerPC/POWER (RS/6000) support.

30 years ago Changed relocs to be based on subsegments (when BFD_ASSEMBLER).
Ian Lance Taylor [Sat, 5 Feb 1994 05:29:42 +0000 (05:29 +0000)]
Changed relocs to be based on subsegments (when BFD_ASSEMBLER).
* subsegs.h (struct frchain): If BFD_ASSEMBLER, added new fields
fix_root and fix_tail.
(segment_info_type): If BFD_ASSEMBLER, don't define fix_tail
field.
* write.c (fix_new_internal): If BFD_ASSEMBLER, set fix_rootP and
fix_tailP based on frchain_now, not seg_info (now_seg).
(chain_frchains_together_1): Chain the subsegment relocs together.
* subsegs.c (subseg_change): Don't clear fix_tail field.
(subseg_get): Likewise.
* literal.c (add_to_literal_pool): Look through the relocs via
frchain_now, not seginfo.

* write.c (write_object_file): Simplified usage of obj_frob_symbol
and tc_frob_symbol.  Always call both if the symbol is going to be
output.

* write.c (relax_segment): Use %ld rather than %d when printing
fragP->fr_var, and cast it to long.

30 years ago * reloc.c (bfd_perform_relocation): Work around one gross hack
Ian Lance Taylor [Sat, 5 Feb 1994 05:13:59 +0000 (05:13 +0000)]
* reloc.c (bfd_perform_relocation): Work around one gross hack
with another: actually look at the target name to avoid the broken
COFF check.
(bfd_reloc_code_real_type): Add BFD_RELOC_PPC_B26,
BFD_RELOC_PPC_BA26 and BFD_RELOC_PPC_TOC16.
* bfd-in2.h: Rebuilt.

30 years ago * coffcode.h (combined_entry_type): Changed fix_tag and fix_end
Ian Lance Taylor [Sat, 5 Feb 1994 05:11:46 +0000 (05:11 +0000)]
* coffcode.h (combined_entry_type): Changed fix_tag and fix_end
fields to bitfields.  Added fields fix_value and fix_scnlen.
(sec_to_styp_flags): If STYP_DEBUG is defined, use it rather than
STYP_INFO for the type of a section named .debug.
(coff_add_missing_symbols): Don't define if RS6000COFF_C.
(coff_write_object_contents): If RS6000COFF_C, don't call
coff_add_missing_symbols.
(coff_slurp_symbol_table): If RS6000COFF_C, then if the last aux
entry has type STY_LD change the x_scnlen into a pointer to a
symbol and set fix_scnlen.  Also, for a C_BSTAT symbol, change the
value into a pointer to a symbol and set fix_value.
* libcoff.h: Rebuilt.
* coffgen.c (coff_mangle_symbols): Reindent.  If fix_value is set,
get the symbol offset. Likewise for fix_scnlen.
(string_size): Change type to bfd_size_type.
(debug_string_size, debug_string_section): New static variables.
(coff_fix_symbol_name): If bfd_coff_symname_in_debug returns true,
write the symbol name into the .debug section; assume that the
section has already been created with the right size.
(coff_write_symbols): Initialize debug_string_size to 0.  If
bfd_coff_symname_in_debug returns true, don't put symbol name in
usual string table.  After writing out all symbols, if
debug_string_size is not 0, check that it matches the size of the
.debug section.
(coff_get_normalized_symtab): Clear new fix_value and fix_scnlen
fields.  If the string offset is 0, always use an empty string as
the name.
(coff_make_empty_symbol): Zero out the symbol structure.

30 years ago * ppc-opc.c (powerpc_operands): New operand type MBE to handle a
Ian Lance Taylor [Sat, 5 Feb 1994 04:46:34 +0000 (04:46 +0000)]
* ppc-opc.c (powerpc_operands): New operand type MBE to handle a
single number giving a bitmask for the MB and ME fields of an M
form instruction.  Change NB to accept 32, and turn it into 0;
also turn 0 into 32 when disassembling.  Seperated SH from NB.
(insert_mbe, extract_mbe): New functions.
(insert_nb, extract_nb): New functions.
(SC_MASK): Mask out SA and LK bits.
(powerpc_opcodes): Change "cal" to use RT, D, RA rather than RT,
RA, SI.  Change "liu" and "cau" to use UI rather than SI.  Mark
"bctr" and "bctrl" as accepted by POWER.  Change "rlwimi",
"rlimi", "rlwimi.", "rlimi.", "rlwinm", "rlinm", "rlwinm.",
"rlinm.", "rlmi", "rlmi.", "rlwnm", "rlnm", "rlwnm.", "rlnm." to
use MBE rather than MB.  Add "mfmq" and "mtmq" POWER instructions.
(powerpc_macros): Define table of macro definitions.
(powerpc_num_macros): Define.

30 years ago * ppc-dis.c (print_insn_powerpc): Don't skip optional operands
Ian Lance Taylor [Sat, 5 Feb 1994 04:45:24 +0000 (04:45 +0000)]
* ppc-dis.c (print_insn_powerpc): Don't skip optional operands
if PPC_OPERAND_NEXT is set.

30 years ago * ppc.h (PPC_OPERAND_NEXT): Define.
Ian Lance Taylor [Sat, 5 Feb 1994 04:38:05 +0000 (04:38 +0000)]
* ppc.h (PPC_OPERAND_NEXT): Define.
(PPC_OPERAND_NEGATIVE): Change value to make room for above.
(struct powerpc_macro): Define.
(powerpc_macros, powerpc_num_macros): Declare.

30 years ago Eliminate arg from bfd_xmalloc_by_size_t macro, allowing
David MacKenzie [Sat, 5 Feb 1994 03:02:32 +0000 (03:02 +0000)]
Eliminate arg from bfd_xmalloc_by_size_t macro, allowing
undoing of obstack parts of last change.

30 years ago * libbfd.c (bfd_zmalloc): Call bfd_xmalloc instead of malloc.
David MacKenzie [Sat, 5 Feb 1994 02:46:12 +0000 (02:46 +0000)]
* libbfd.c (bfd_zmalloc): Call bfd_xmalloc instead of malloc.
(bfd_xmalloc, bfd_xmalloc_by_size_t): Functions deleted.
* libbfd-in.h: Define them as macros calling xmalloc and declare
xmalloc.
* libbfd.h: Rebuilt.
* ecofflink.c hash.c ieee.c opncls.c (obstack_chunk_alloc): Define
to be xmalloc, not bfd_xmalloc_by_size_t.

30 years ago * ldmisc.c (ldmalloc, xmalloc, ldrealloc, xrealloc): Functions
David MacKenzie [Sat, 5 Feb 1994 01:14:56 +0000 (01:14 +0000)]
* ldmisc.c (ldmalloc, xmalloc, ldrealloc, xrealloc): Functions
deleted; will use libiberty versions instead.
* ldctor.c ldfile.c ldlang.c ldmain.c ldmisc.c ldmisc.h lexsup.c
mri.c Makefile.in: Change callers.

* ldmisc.c (vfinfo): Remove cleanup code.
* ldmain.c (remove_output): Put it here (new function).
(preserve_output): New function.
(main): Register remove_output and preserve_output with atexit.
* ldmain.c ldgram.y: Call xexit instead of exit.
* ldmisc.h: Declare xexit.

30 years agoREGISTER_TYPE is obsolete.
Steve Chamberlain [Sat, 5 Feb 1994 01:07:48 +0000 (01:07 +0000)]
REGISTER_TYPE is obsolete.

30 years ago(hms_fetch_register): Take out REGISTER_TYPE.
Steve Chamberlain [Sat, 5 Feb 1994 01:06:21 +0000 (01:06 +0000)]
(hms_fetch_register): Take out REGISTER_TYPE.

30 years agoFix REGISTER_TYPE and order of register names.
Steve Chamberlain [Sat, 5 Feb 1994 00:39:09 +0000 (00:39 +0000)]
Fix REGISTER_TYPE and order of register names.

30 years agoFix REGISTER_TYPE.
Steve Chamberlain [Sat, 5 Feb 1994 00:38:29 +0000 (00:38 +0000)]
Fix REGISTER_TYPE.

30 years agoGet REGISTER_TYPE right.
Steve Chamberlain [Sat, 5 Feb 1994 00:36:32 +0000 (00:36 +0000)]
Get REGISTER_TYPE right.

30 years agoFri Feb 4 15:53:18 1994 Steve Chamberlain (sac@cygnus.com)
Steve Chamberlain [Sat, 5 Feb 1994 00:22:08 +0000 (00:22 +0000)]
Fri Feb  4 15:53:18 1994  Steve Chamberlain  (sac@cygnus.com)

* h8500-tdep.c (saved_pc_after_call): The size of the
pc is memory model dependent. (segmented_command,
unsegmented_command, _initialize_h8500_tdep): New commands to
change memory model.
* remote-7000.c (initialize_remote_e7000): Change name of snoop
command.
* remote-hms.c (hms_load): Remove breakpoints when loaded.
(hms_wait): Use new status structure
(hms_open): Push the target here. (hms_before_main_loop): Not
here.  (supply_val, hms_fetch_register, hms_store_register):  Cope
with H8/500 names too.
* sh-tdep.c (show_regs, initialize_sh_tdep): New command to print
all registers in a compact way.

30 years ago * Makefile.in: Lots of new H8/500 memory models.
Steve Chamberlain [Fri, 4 Feb 1994 23:19:33 +0000 (23:19 +0000)]
* Makefile.in: Lots of new H8/500 memory models.

30 years agoNew memory model.
Steve Chamberlain [Fri, 4 Feb 1994 23:18:19 +0000 (23:18 +0000)]
New memory model.

30 years agoMore sicko segmentation support.
Steve Chamberlain [Fri, 4 Feb 1994 23:08:00 +0000 (23:08 +0000)]
More sicko segmentation support.

30 years agoSupport for various sicko segmented memory models.
Steve Chamberlain [Fri, 4 Feb 1994 22:57:26 +0000 (22:57 +0000)]
Support for various sicko segmented memory models.

30 years ago* lib/gas-defs.exp (regexp_diff): New procedure, vaguely based on
Ken Raeburn [Fri, 4 Feb 1994 22:19:31 +0000 (22:19 +0000)]
* lib/gas-defs.exp (regexp_diff): New procedure, vaguely based on
"simple_diff" from linker test suite.
* gas/sun4/addend.exp: Use it.
* gas/sun4/addend.d: New file.

30 years ago * objcopy.c (filter_bytes): Make MEMHUNK a char *, not PTR, so we
David MacKenzie [Fri, 4 Feb 1994 18:47:33 +0000 (18:47 +0000)]
* objcopy.c (filter_bytes): Make MEMHUNK a char *, not PTR, so we
can do arithmetic on it.

30 years ago * gdb.t24/demangle.exp: Remove a bunch more tests which differ
Jim Kingdon [Fri, 4 Feb 1994 18:45:39 +0000 (18:45 +0000)]
* gdb.t24/demangle.exp: Remove a bunch more tests which differ
only in the names (I probably got most of them).  I was even
generous with some which differed in underscores in names, and
probably missed a few, and I *still* was able to remove almost a
third of the file with almost no impact on testsuite coverage.

30 years ago * config/rs6000/tm-rs6000.h: Declare rs6000_struct_return_address
Jim Kingdon [Fri, 4 Feb 1994 15:55:04 +0000 (15:55 +0000)]
* config/rs6000/tm-rs6000.h: Declare rs6000_struct_return_address
as CORE_ADDR to match definition in rs6000-tdep.c.

30 years ago * gdb.texinfo (Continuing and Stepping): When talking about "step"
Jim Kingdon [Fri, 4 Feb 1994 14:40:56 +0000 (14:40 +0000)]
* gdb.texinfo (Continuing and Stepping): When talking about "step"
versus functions without line numbers, also mention stepping into
them as well as "step" when you are in them.  Tell the user how to
deal with the situation.  Add comment about "debugging information".

30 years ago * dwarfread.c (process_dies): Skip nested TAG_compile_unit DIEs.
Peter Schauer [Fri, 4 Feb 1994 09:19:35 +0000 (09:19 +0000)]
* dwarfread.c (process_dies):  Skip nested TAG_compile_unit DIEs.
* dwarfread.c (add_partial_symbol):  Do not enter opaque aggregate
definitions into the psymtab.

30 years ago * ecofflink.c (bfd_ecoff_debug_externals): If a small undefined
Ian Lance Taylor [Fri, 4 Feb 1994 04:18:28 +0000 (04:18 +0000)]
* ecofflink.c (bfd_ecoff_debug_externals): If a small undefined
symbol has a value in the ECOFF symbol but not in the BFD symbol,
keep the value in the ECOFF symbol.  This helps gas.

30 years ago * ecoff.c (ecoff_symbol_new_hook): Clear ecoff_extern_size, not
Ian Lance Taylor [Fri, 4 Feb 1994 04:14:24 +0000 (04:14 +0000)]
* ecoff.c (ecoff_symbol_new_hook): Clear ecoff_extern_size, not
ecoff_undefined.
(ecoff_frob_symbol): Don't check ecoff_undefined.
(ecoff_build_symbols): Get size of an undefined symbol from
sym->ecoff_extern_size, not S_GET_VALUE (sym).

30 years agoRevert last change.
Ian Lance Taylor [Fri, 4 Feb 1994 03:17:30 +0000 (03:17 +0000)]
Revert last change.

30 years ago * config/tc-mips.c (md_apply_fix): If this is an undefined symbol,
Ian Lance Taylor [Fri, 4 Feb 1994 02:32:54 +0000 (02:32 +0000)]
* config/tc-mips.c (md_apply_fix): If this is an undefined symbol,
subtract the value from fx_addnumber to since
bfd_perform_relocation is going to add it in.

30 years ago * gdb.t00/default.exp: Match udi_attach error message as well as
Jim Kingdon [Fri, 4 Feb 1994 02:25:33 +0000 (02:25 +0000)]
* gdb.t00/default.exp: Match udi_attach error message as well as
the one from child_attach.

* gdb.t00/default.exp: Use "." instead of "?" to match a question mark.

30 years ago* remote-udi.c (udi_attach): If no arguments, print error.
Jim Kingdon [Fri, 4 Feb 1994 02:22:53 +0000 (02:22 +0000)]
* remote-udi.c (udi_attach): If no arguments, print error.

30 years ago * Makefile.in (VERSION): Bump to 4.12.1
Fred Fish [Fri, 4 Feb 1994 01:37:20 +0000 (01:37 +0000)]
* Makefile.in (VERSION):  Bump to 4.12.1
* NEWS, README:  Update to match 4.12 release.

30 years ago * linker.c (_bfd_generic_link_output_symbols,
Ian Lance Taylor [Thu, 3 Feb 1994 23:01:49 +0000 (23:01 +0000)]
* linker.c (_bfd_generic_link_output_symbols,
_bfd_generic_link_write_global_symbol): Don't require that all
references to a common symbol be themselves common symbols.

30 years ago * objdump.c (dump_section_stabs, read_section_stabs,
David MacKenzie [Thu, 3 Feb 1994 22:21:28 +0000 (22:21 +0000)]
* objdump.c (dump_section_stabs, read_section_stabs,
print_section_stabs): Functions broken out of dump_stabs_1.
Free the stabs and strings when done with them.

30 years ago * aoutx.h (aout_reloc_index_to_section): Handle N_UNDF.
Ian Lance Taylor [Thu, 3 Feb 1994 21:52:06 +0000 (21:52 +0000)]
* aoutx.h (aout_reloc_index_to_section): Handle N_UNDF.

30 years ago * command.c (empty_sfunc): New function.
Jim Kingdon [Thu, 3 Feb 1994 19:39:47 +0000 (19:39 +0000)]
* command.c (empty_sfunc): New function.
(add_set_cmd): Use it instead of not_just_help_class_command.
(not_just_help_class_command): Change calling convention back to
what it was before yesterday's change.

30 years ago * stabsread.c (read_sun_builtin_type): Skip the semicolon at the end
Jim Kingdon [Thu, 3 Feb 1994 19:28:30 +0000 (19:28 +0000)]
* stabsread.c (read_sun_builtin_type): Skip the semicolon at the end
of the type if present.

30 years ago * Move test with enums and partial symbols back to
Jim Kingdon [Thu, 3 Feb 1994 17:25:43 +0000 (17:25 +0000)]
* Move test with enums and partial symbols back to
gdb.t03/{ptype.exp,gdbme.c} from gdb.stabs/{weird.def,weird.exp}.
Reverts the change of 26 May 93.  Per today's change to
stabs.texinfo, the behavior that weird.exp was expecting was
unreasonable; what is important is that the compiler+gdb get
things right, which happens with a recent gcc.  Also fix the test
to deal with native compilers which put out the stab gdb can't
deal with.

30 years ago * stabs.texinfo (Enumerations): Document restriction on where
Jim Kingdon [Thu, 3 Feb 1994 16:44:13 +0000 (16:44 +0000)]
* stabs.texinfo (Enumerations): Document restriction on where
enumeration types can appear and still win with GDB.

30 years ago More changes to print ambiguous matches and use the
David MacKenzie [Thu, 3 Feb 1994 13:08:29 +0000 (13:08 +0000)]
More changes to print ambiguous matches and use the
abstracting macros.

30 years agoFormatting adjustments, for better smallbook formatting
Roland Pesch [Thu, 3 Feb 1994 04:38:58 +0000 (04:38 +0000)]
Formatting adjustments, for better smallbook formatting
and to agree better w/FSF (RMS) indentation conventions.

30 years agoDescribe -retain-symbols-file.
Roland Pesch [Thu, 3 Feb 1994 04:18:11 +0000 (04:18 +0000)]
Describe -retain-symbols-file.

30 years ago* bfd-in2.h: Rebuilt.
Jim Kingdon [Thu, 3 Feb 1994 04:03:11 +0000 (04:03 +0000)]
* bfd-in2.h: Rebuilt.

30 years ago * valarith.c (value_binop): Figure out type ourself based on
Jim Kingdon [Thu, 3 Feb 1994 03:58:26 +0000 (03:58 +0000)]
* valarith.c (value_binop): Figure out type ourself based on
sizeof (LONGEST) rather than relying on BUILTIN_TYPE_LONGEST.  The
point is that we don't depend on CC_HAS_LONG_LONG anymore.

30 years ago * language.h, c-lang.c, ch-lang.c, m2-lang.c, language.c: Remove
Jim Kingdon [Thu, 3 Feb 1994 03:44:15 +0000 (03:44 +0000)]
* language.h, c-lang.c, ch-lang.c, m2-lang.c, language.c: Remove
longest_int and longest_unsigned_int.

30 years ago* bfd-in2.h: Rebuilt.
Jim Kingdon [Thu, 3 Feb 1994 03:33:20 +0000 (03:33 +0000)]
* bfd-in2.h: Rebuilt.

30 years ago * libbfd.c, bfd-in.h, hosts/alphaosf.h, hosts/sparc-ll.h, aoutf1.h,
Jim Kingdon [Thu, 3 Feb 1994 03:31:22 +0000 (03:31 +0000)]
* libbfd.c, bfd-in.h, hosts/alphaosf.h, hosts/sparc-ll.h, aoutf1.h,
sparclynx.c, Makefile.in: Change HOST_64_BIT to BFD_HOST_64_BIT.

30 years ago * printcmd.c (decode_format): Don't blithely set the size for
Jim Kingdon [Thu, 3 Feb 1994 03:18:58 +0000 (03:18 +0000)]
* printcmd.c (decode_format): Don't blithely set the size for
an address to 'w'.  Make it 'g' or 'h', depending on TARGET_PTR_BIT.

* defs.h: Just typedef CORE_ADDR to bfd_vma.  Include bfd.h.
If BFD64, make a LONGEST a BFD_HOST_64_BIT.
* defs.h (longest_to_int): Don't depend on CC_HAS_LONG_LONG; instead
always just check against INT_MIN and INT_MAX (this also fixes things
if sizeof (long) > sizeof (int), e.g. Alpha).
* config/pa/tm-hppa.h, config/i386/sun386.h, config/rs6000/tm-rs6000.h:
Don't define LONGEST or BUILTIN_TYPE_LONGEST.
* gdbtypes.h: Remove BUILTIN_TYPE_LONGEST and
BUILTIN_TYPE_UNSIGNED_LONGEST.
* language.h, c-lang.c, ch-lang.c, m2-lang.c: Remove longest_int
and longest_unsigned_int.
* value.h (struct value): Just align to LONGEST, rather than worrying
about CC_HAS_LONG_LONG.
* valarith.c (value_binop, _initialize_valarith): Figure out type
ourself based on sizeof (LONGEST) rather than relying on
BUILTIN_TYPE_LONGEST.  The point is that we don't depend on
CC_HAS_LONG_LONG anymore.
* valprint.c (val_print_type_code_int): Just call
extract_unsigned_integer directly, rather than going through
unpack_long.
* printcmd.c (decode_format): Remove code which would sometimes
change 'g' size to 'w' for integers.  print_scalar_formatted handles
printing huge integers well enough, thank you.

30 years agomisc updates:
Roland Pesch [Thu, 3 Feb 1994 03:12:38 +0000 (03:12 +0000)]
misc updates:
(1) show `-y SYMBOL' with space
(2) weaken multi-target stuff, to avoid false expectations now that
linker now often config'd with only one target
(3) describe -relax support on i960
(4) correct description of hyphen chars in symbols
(5) describe CONSTRUCTORS linker script cmd

30 years ago * nlmconv.c (main): Use bfd_get_flavour instead of struct member.
David MacKenzie [Thu, 3 Feb 1994 01:39:10 +0000 (01:39 +0000)]
* nlmconv.c (main): Use bfd_get_flavour instead of struct member.
* ar.c (print_contents, extract_file), size.c
(print_berkeley_format): Use bfd_get_filename and
bfd_my_archive instead of dereferencing the structs directly.

30 years ago * ar.c: Use bfd_fatal and bfd_nonfatal instead of bfd_perror and exit.
David MacKenzie [Thu, 3 Feb 1994 00:42:37 +0000 (00:42 +0000)]
* ar.c: Use bfd_fatal and bfd_nonfatal instead of bfd_perror and exit.
Indent.  Remove DEFUNs.

30 years ago * nlmconv.c (main), objcopy.c (copy_file): Print matching formats
David MacKenzie [Thu, 3 Feb 1994 00:25:30 +0000 (00:25 +0000)]
* nlmconv.c (main), objcopy.c (copy_file): Print matching formats
if ambiguous match.
* nm.c (display_file), size.c (display_bfd): Eliminate gotos.
Print matching formats if there is an ambiguous match.  Use
bfd_nonfatal instead of hardcoded error message if nothing matches.

* arsup.c, ar.c, objdump.c: Use bfd_get_filename instead of
abfd->filename.

* nm.c (display_archive): New function, from code in display_file.
(display_rel_file): Renamed from do_one_rel_file.

* size.c: Indent.
(display_archive): New function from code in display_file.
(display_file): Check bfd_close error return.

* strings.c (strings_object_file): Check bfd_check_format
error return.

* strings.c, objdump.c, size.c: Use bfd_nonfatal instead of bfd_perror.

* bucomm.c: Delete references to exit_handler.  It wasn't set
anywhere, and now that we're using the libiberty xmalloc, it
wouldn't always get called before exiting.
(list_matching_formats): Function moved from objdump.c.
* bucomm.h: Declare it.

* objdump.c (disassemble_data): Move some variable decls closer to
their use.  Add some comments.  Replace a nested block with a
return.

30 years ago * Makefile.in: Avoid bug in losing hpux sed.
Jeff Law [Wed, 2 Feb 1994 21:58:54 +0000 (21:58 +0000)]
    * Makefile.in:  Avoid bug in losing hpux sed.

30 years ago * command.c (add_set_cmd, no_just_help_class_command): Change
Jim Kingdon [Wed, 2 Feb 1994 20:20:59 +0000 (20:20 +0000)]
* command.c (add_set_cmd, no_just_help_class_command): Change
to make this the sfunc, not cfunc, since that is how we call it.
* command.h: Comment difference between sfunc and cfunc.
* demangle.c (set_demangling_command): Add third arg since that
is how it is called.
(_initialize_demangler): Use sfunc, not cfunc, for
set_demangling_command, since that is how it is called.
Remove show_demangling_command; it has no effect.

30 years ago * command.c (shell_escape): Report errors correctly (with error
Jim Kingdon [Wed, 2 Feb 1994 19:58:31 +0000 (19:58 +0000)]
* command.c (shell_escape): Report errors correctly (with error
message from strerror).

30 years ago * Makefile.in, test-build.mk: Remove MUNCH_NM; it was only needed
Jim Kingdon [Wed, 2 Feb 1994 19:56:10 +0000 (19:56 +0000)]
* Makefile.in, test-build.mk: Remove MUNCH_NM; it was only needed
for GDB and GDB has been fixed to not need it.

30 years ago * xcoffread.c (read_xcoff_symtab): Change CSECT_LEN to use
Ian Lance Taylor [Wed, 2 Feb 1994 19:37:52 +0000 (19:37 +0000)]
* xcoffread.c (read_xcoff_symtab): Change CSECT_LEN to use
x_scnlen.l rather than x_scnlen to match corresponding change in
coff/internal.h.

30 years ago * coffswap.h (coff_swap_reloc_out): If RS6000COFF_C, handle type
Ian Lance Taylor [Wed, 2 Feb 1994 19:35:14 +0000 (19:35 +0000)]
* coffswap.h (coff_swap_reloc_out): If RS6000COFF_C, handle type
and size correctly.
(coff_swap_aux_in): If RS6000COFF_C, change x_csect.x_scnlen to
x_csect.x_scnlen.l to match change in coff/internal.h.
(coff_swap_aux_out): Likewise.

30 years ago * internal.h (union internal_auxent): Change x_csect.x_scnlen into
Ian Lance Taylor [Wed, 2 Feb 1994 19:33:19 +0000 (19:33 +0000)]
* internal.h (union internal_auxent): Change x_csect.x_scnlen into
a union of a long and a pointer to a symbol.  XCOFF sometimes uses
this field as a symbol index.

30 years ago * common.h: Add comments regarding value of EM_HPPA and how to
Jim Kingdon [Wed, 2 Feb 1994 19:21:25 +0000 (19:21 +0000)]
* common.h: Add comments regarding value of EM_HPPA and how to
pick an unofficial value.

30 years ago * coff-mips.c (mips_ecoff_backend_data), coff-alpha.c
Ian Lance Taylor [Wed, 2 Feb 1994 17:40:46 +0000 (17:40 +0000)]
* coff-mips.c (mips_ecoff_backend_data), coff-alpha.c
(alpha_ecoff_backend_data): Change casts of aux_in and aux_out
fields to match yesterday's changes.

30 years ago * coffcode.h (coff_write_relocs): If SELECT_RELOC is defined, pass
Ian Lance Taylor [Wed, 2 Feb 1994 17:36:17 +0000 (17:36 +0000)]
* coffcode.h (coff_write_relocs): If SELECT_RELOC is defined, pass
in the internal_reloc itself, not the type.
* coff-apollo.c, coff-h8300.c, coff-h8500.c, coff-i386.c,
coff-m68k.c, coff-sh.c, coff-we32k.c, coff-z8k.c: Changed
definition of SELECT_RELOC accordingly.

30 years ago * gdbtypes.h, ch-typeprint.c, ch-valprint.c:
Jim Kingdon [Wed, 2 Feb 1994 17:20:45 +0000 (17:20 +0000)]
* gdbtypes.h, ch-typeprint.c, ch-valprint.c:
Change comments regarding TYPE_CODE_BOOL.
* language.c (boolean_type): Always return 1 for TYPE_CODE_BOOL,
regardless of the language.
(value_true): Just call value_logical_not regardless of language.
* coffread.c (coff_read_enum_type), stabsread.c (read_enum_type):
Remove #if 0'd code which makes some enums TYPE_CODE_BOOL.
* language.h: Improve comment for la_builtin_type_vector.
* m2-lang.c (_initialize_m2_language): Don't add any fields to
builtin_type_m2_bool.

30 years ago * config/tc-mips.c (file_mips_isa): New static variable.
Ian Lance Taylor [Wed, 2 Feb 1994 17:06:24 +0000 (17:06 +0000)]
* config/tc-mips.c (file_mips_isa): New static variable.
(md_begin): Set it.
(s_mipsset): Add support for .set mipN to set the ISA level.

30 years ago * gasp.c (kinfo): Fully bracket initializer.
Ian Lance Taylor [Wed, 2 Feb 1994 16:43:00 +0000 (16:43 +0000)]
* gasp.c (kinfo): Fully bracket initializer.

30 years ago * stabs.texinfo (Negative Type Numbers): Document format for type
Jim Kingdon [Wed, 2 Feb 1994 16:33:09 +0000 (16:33 +0000)]
* stabs.texinfo (Negative Type Numbers): Document format for type
-16.

30 years agoStop claiming MIPS li.d and li.s insns are not supported, now that they are.
Roland Pesch [Wed, 2 Feb 1994 03:44:39 +0000 (03:44 +0000)]
Stop claiming MIPS li.d and li.s insns are not supported, now that they are.

30 years agoThree scattered fixes (nearly lost!) from early suggestions by Jeff Law.
Roland Pesch [Wed, 2 Feb 1994 03:32:31 +0000 (03:32 +0000)]
Three scattered fixes (nearly lost!) from early suggestions by Jeff Law.

30 years agoDescribe AT option of SECTIONS command, at long last.
Roland Pesch [Wed, 2 Feb 1994 02:27:55 +0000 (02:27 +0000)]
Describe AT option of SECTIONS command, at long last.

30 years agoBetter organization, indexing, for MIPS-dep section.
Roland Pesch [Wed, 2 Feb 1994 01:43:26 +0000 (01:43 +0000)]
Better organization, indexing, for MIPS-dep section.

30 years agogasp.c, atof-ieee.c, ho-i386aix.h changes
Ken Raeburn [Wed, 2 Feb 1994 00:40:29 +0000 (00:40 +0000)]
gasp.c, atof-ieee.c, ho-i386aix.h changes