diff options
Diffstat (limited to 'gnu/usr.bin/as/ChangeLog')
| -rw-r--r-- | gnu/usr.bin/as/ChangeLog | 429 |
1 files changed, 429 insertions, 0 deletions
diff --git a/gnu/usr.bin/as/ChangeLog b/gnu/usr.bin/as/ChangeLog new file mode 100644 index 000000000000..db772344197c --- /dev/null +++ b/gnu/usr.bin/as/ChangeLog @@ -0,0 +1,429 @@ +Sun Mar 1 17:02:06 1992 K. Richard Pixley (rich@cygnus.com) + + * README: updated to 1.92.3, included mail announcement. + +Sat Feb 29 00:53:16 1992 K. Richard Pixley (rich@cygnus.com) + + * tc-sparc.c (md_apply_fix): relocation overflow checks. + + * atof-generic.c (atof_generic): recognize 99e999 as infinity for + older, broken, compilers. + + * version.c: bump to 1.92.3, drop "Cygnus". + + * input-scrub.c (as_where): use myname (which comes from argv[0]) + as part of all error messages. + + * mess-dose renaming: + flonum-copy.c -> flo-copy.c + flonum-const.c -> flo-const.c + config/a.out.gnu.h -> config/aout.h + config/coff.gnu.h -> config/coff.h + + * Makefile.in, obj-aout.h, obj-coff.h: reflect file renaming. + + * output-file.c (output_file_create): add "b" to the fopen to + humor mess-dos. + + * configure.in: tahoe needs atof-tahoe. + + * config/tc-tahoe.[hc], config/atof-tahoe.c, opcode/tahoe.h: new + files. This is kinda blind cause I don't have anything to run + through it or compare against. + + * read.c (read_a_source_file), expr.c (operand): fix a very old + bug in label reading exposed by m88k. Also, m88k can't have a + pseudo "set". + + * config/m88k.[hc]: freshen copyrights, version 2 gpl, update to + current gas. + + * config/m88k-opcode.h moved to opcode/m88k.h + + * read.c: NO_DOT_PSEUDOS from hacks unfinished work. + + * opcode/m68k.h: Sun's JFcc aliases appear to be variable length. + Make them so. + + * opcode/a29k.h: remove rcsid. + + * config/te-sun3.h: remove semicolon typo. + + * config/obj-vms.c: another patch from eric youngdale. + + * write.c: white space only. + + * config/tc-i960.c: change from intel for header flags. + + * config/te-sequent.h, config/obj-aout.h: first cut at building + sequent headers. + + * config/tc-ns32k.c: patches from Jyrki Kuoppala <jkp@cs.hut.fi>. + + * struct-symbol.h: removed redundant decl of N_TYPE_seg. + + * config/tc-sparc.c (sparc_ip), opcode/sparc.h: changes from chris + torek to correct a problem with "neg". some white space. + + * confic/tc-m68k.c: a fix pulled from hack's unfinished work and + my mail archives. Try again to get pcrel working. Fix stupid + botch on cpu_type comparison. + + * config/tc-sparc.c: .empty pseudo-op from + gordoni@cs.adelaide.edu.au. + + * opcode/sparc.h: some new aliases from chris torek. + + * opcode/i386.h: some new aliases and opcodes. also patches from + Steve Bleazard <steve@robobar.co.uk>. + + * config/te-hpux.h: new file. + + * configure.in: when targetting hpux, use te-hpux.h. + + * config/obj-aout.c (obj-pre-write-hook), config/obj-bout.[ch] + (obj-pre-write-hook), config/obj-coff.[ch] (obj-pre-write-hook), + config/obj-generic.h, config/obj-vms.h, write.c + (write_object_file): move magic number fiddling out of write.c + and into obj-pre-write-hook. + + * config/tc-i860.c: gcc -Wall cleanup. + +Fri Feb 28 00:30:36 1992 K. Richard Pixley (rich@rtl.cygnus.com) + + * configure.in: if target is sun3, use te-sun3.h. + + * config/tc-m68k.h, config/te-sun3.h: moved #define of + default_magic_number_for_object_file from former to latter. + + * config/te-sun3.h: removed sun_asm_syntax and te_sun3, they + aren't used. + + * all: white space changes. + " -> " becomes "->" + "foo [" becomes "foo[" + "a . b" becomes "a.b" + "\(if\|for\|while\|switch\)(" become "\\1(" + "\\([^\n]\\)[ \t]*\\([=!+-*/<>]\\)=[ \t]*" become "\\1 \\2= " + + * read.c, write.c, config/tc-i386.c: white space and comments + only. + + * config/obj-vms.c: convert PUT_LONG and PUT_SHORT to squirt byte + swapped numbers. + + * as.c, flonum-const.c, hex-value.c, input-file.c, version.c, + config/obj-aout.h, config/obj-vms.c: VMS -> HO_VMS. + + * config/ho-vms.h: added HO_VMS. + +Thu Feb 27 18:25:11 1992 K. Richard Pixley (rich@rtl.cygnus.com) + + * config/ChangeLog: removed. entries merged into this file. + + * config/ho-vms.h: new file. Move the VMS stuff out of ho-vax.h + into ho-vms.h. + + * configure.in: use ho-i386v4 for i386-sysvr4. + + * config/ho-i386v4: new file. + +Tue Feb 25 19:54:04 1992 (Eric Youngdale at youngdale@v6550c.nrl.navy.mil) + + * config/obj-vms.c (VMS_write_object_file): Add work-around + for g++ compiler bug involving external vtables. + +Mon Feb 24 22:19:10 1992 (Eric Youngdale at youngdale@v6550c.nrl.navy.mil) + + * README-vms: Describe how to get a VMS obj file to a vms machine + via NFS. + + * configure.in: For i386-sysv*, use gas_host=i386. + + * Makefile.in: Remove continuation line markers when the next line + is blank. + + * read.c (line_comment_chars): Make external. + + * input-file.c: Remove redundant include of <assert.h>. + + * config/ho-vax.h [VMS]: Include <ctype.h> and <perror.h>. + + * config/obj-vms.h: Remove said includes. Add RELOC_32 to + reloc_type to prevent compilation error. + + * config/obj-vms.c: Change bcopy to memcpy throughout. + (VMS_local_stab_Parse): Fix typo. + (VMS_local_stab_Parse, VMS_RSYM_Parse, Define_Local_Symbols, + Define_Routine, VMS_write_object_file): Allow 'f' for functions + as well as 'F'. + +Mon Feb 24 03:48:04 1992 K. Richard Pixley (rich@cygnus.com) + + * README: updated to reflect current testing status. + + * README.rich, NOTES, NOTES.config: updated slightly, marked as + "under construction". + + * CONTRIBUTORS: new file. + + * README-vms: options to configure are now -options=, not + +options=. + + * version.c: bumped version to 1.92.2. + +Mon Feb 24 03:27:00 1992 Eric Youngdale (youngdale at v6550c.nrl.navy.mil) + + * config.sub: Added vms as a target system. (So people do not + have to try to figure out that "vax-dec-vms" would work). + + * configure.in: Added vms as a target os, and object file format. + (Useless on a vms system, but this is for people who want to + cross assemble). + + * config-gas.com: New file. Script for VMS systems to set up the + configuration to build gas for VMS, and create config.status. + + * make-gas.com: Redone to work with the bfd-gas scheme. + + * as.c: Add const modifier to version_string. + + * atof-vax.c: Remove redundant include of flonum.h. (This is also + included via as.h). + + * expr.c: Add "const" modifier to hex_value. + + * read.c: Add "const" modifier to line_comment_chars, and + line_separator_chars. Make use of the -1 switch for backward + compatibility with gcc 1.nn. + (s_ignore): remove redundant declaration of is_end_of_line. + + * symbols.c: Finish conversion to S_* macros in the VMS only + parts of the program. Add "const" modifier to + md_[long,short]_jump_size. Remove declaration of const_flag + (which will be declared in obj-vms.h). + + * write.c: Add "const" modifier to md_[long,short]_jump_size. + Fix arguments to VMS_write_object_file. + + * obj-vms.h: New file (sort of). Mostly canibalized from other + files, using: + + - objrecdef.h: Removed structure definition that we do not use, + and removed dollar signs from identifiers, since Unix System 5 + does not like them. + + - obj-aout.h: Took S_*, some H_* macros, and a number of + symbol definitions. + + - a.out.hp.h: Took nlist structure. We do not really use this + per se, but it is easiest to let gas think that we do. When we + write the object file, we just pick out the parts that we need. + + - stab.h: Just included it, since on non VMS and non a.out systems + we have no guarantee of having it. (Define N_* symbols). + + *obj-vms.c: Renamed from vms.c. Did the following: + + - Reworked to use the S_* macros. + + - Add "const" modifier to version_string. + + - Added global[ref,def,value] support + + - (VMS_Store_PIC_Symbol_Reference):fix a bug with static constants. + + - Remove a few redunant includes - all are now included through as.h. + + - (obj_crawl_symbol_chain): Clean up (a lot), and remove non-VMS + code. Add definition for obj_read_begin_hook. + + - Borrow the stab[s,d,n] routines from obj-aout.c. + + - Borrow the seg_N_TYPE and N_TYPE_seg arrays from aout.c + + - Use <fab.h>,<rab.h> and <xab.h> instead of <vms/fabdef.h> + <vms/rabdef.h> and <vms/xabdef.h>, for more consistent results. + (Some peoples <vms/*.h> files are different than others). + + - Merged vms-dbg.c into obj-vms.c. Modified to use + the S_* macros. Added code to remove the psect hack from + variable names before writing them to the debugger records. + + + + The following patches make cross assembly possible. + + * as.c, read.c, symbols.c, write.c: Change "ifdef VMS" to + "ifdef OBJ_VMS". + + * vms.c: + + - Wrap the #include of some VMS system dependent headers + with "ifdef VMS". + + - (get_VMS_time_on_unix): Add new routine. Generates current + time in VMS format to be written into object file. + + - (Write_VMS_MHD_Records): Use get_VMS_time_on_unix if we are not + running on a VMS system. + + - (Flush_VMS_Object_Record_Buffer): Add code to write correct + record format when running on a non-VMS system. + + - (Create_VMS_Object_File): Use different mode if running under + unix. + + - (VMS_TBT_Source_File): If we are not running on a VMS system, + write a source file record for the debugger that looks reasonable. + +Mon Feb 24 02:06:00 1992 K. Richard Pixley (rich@cygnus.com) + + * Makefile.in: remove $(srcdir)/../include from INCLUDES. It + isn't needed. + + * README: updated with current state. + + * read.c (stringer): read arbitrary expressions between the commas + and treat them as ".byte" values. At least some i860 assembler + does this so now we do too. Also white space throughout. + + * expr.c, expr.h, frags.c, symbols.c, write.c: white space only. + +Mon Feb 24 01:45:40 1992 K. Richard Pixley (rich@cygnus.com) + + * config/te-sequent.h, config/tc-ns32k.h, config/tc-ns32k.c: + SEQUENT_COMPATIBILITY -> TE_SEQUENT. + + * config/obj-aout.c: if OLD_GAS and i386, then screw up the magic + number. + + * config/obj-bout.c: do not include aout/stab_gnu.h if NO_LISTING. + + * config/obj-bout.h: added enum reloc_type. + + * config/tc-i386.c: on OLD_GAS, .align is power of two, rather + than bytes. + + * config/tc-i386.h: on OLD_GAS, the filler byte should be zero + rather than NOOP. + + * config/tc-i860.c: relocs are 12bytes on this target. Also white + space. + + * config/tc-m68kmote.c: removed. Not ready yet. + + * config/a.out.gnu.h, config/tc-a29k.c, config/tc-m68k.c, + config/tc-ns32k.c: white space only. + + * config/tc-a29k.h, config/tc-i860.h, config/tc-i960.h, + config/tc-m68k.h, config/tc-ns32k.h, config/tc-sparc.h, + config/tc-vax.h: NO_LISTING + + * config/tc-m68k.h, config/tc-i860.h, config/tc-vax.h: + REVERSE_SORT_RELOCS if OLD_GAS. + + * config/mt-m68k: removed. not needed. + +Fri Feb 21 06:22:15 1992 K. Richard Pixley (rich@rtl.cygnus.com) + + * config/obj-aout.c: do not include stab.gnu.h if NO_LISTING. + + * config/tc-i860.c, config/a.out.gnu.h: move i860 relocs to a proper place. + + * config/a.out.h: removed. + +Fri Feb 21 06:21:07 1992 K. Richard Pixley (rich@rtl.cygnus.com) + + * Makefile.in: put header files before C source for TAGS; remove + references to non-existent syscalls.h. + + * read.c, write.c subsegs.c: back out the .bss changes. + +Fri Feb 21 02:17:22 1992 Minh Tran-Le (TRANLE@INTELLICORP.COM) + + * config/tc-i386.c: config/tc-i386.c: added handling of the + following opcodes: i/o opcodes - inb, inw, outb and outw. + string manipulation with att syntax - scmp, slod, smov, ssca, + ssto. + +Fri Feb 21 01:53:50 1992 Minh Tran-Le (TRANLE@INTELLICORP.COM) + + * config/obj-coff.c: (for aix386) Moved the symbols .text, .data + and .bss to just after .file . + + In obj_crawl_symbol_chain() where it tries to put the external + symbols apart, with the condition: + (!S_IS_DEFINED(symbolP) && + !S_IS_DEBUG(symbolP) && + !SF_GET_STATICS(symbolP)) + it was moving too many symbols out. So I switch it back to the + condition: + (S_GET_STORAGE_CLASS(symbolP) == C_EXT && !SF_GET_FUNCTION(symbolP)) + + In obj_emit_relocations() added the conditional on KEEP_RELOC_INFO + so that we don't use the F_RELFLG which make the linker complain + that somebody has stripped the relocation info. + + Also, the AIX ld program require that the relocation table + is sorted by r_vaddr like the standard ATT assembler does. + + [he also changed the sizeof(struct ...)'s into the coff + style FOOSZ macros. I'm not sure this is right, but I can't + remember why. xoxorich.] + +Fri Feb 21 01:08:48 1992 Minh Tran-Le (TRANLE@INTELLICORP.COM) + + * symbols.c (local_label_name): symbols now start with ^A. + + * read.c, subsegs.c, write.c obj-coff.c: added handling of + `.bss` pseudo op for unitialized data. The new gcc (1.37.9x) + generate these sections. .align: will use NOP_OPCODE or 0 + for padding. This is just for being nice to the + disassembler. + + * expr.c (operand): changed to generate local label "\001L0" + starting with a ^A so that it is recognized as a local label. + + * as.c (perform_an_assembly_pass): zero bss_fix_root, too. + +Fri Feb 21 01:08:48 1992 K. Richard Pixley (rich@cygnus.com) + + * Makefile.in, configure.in, doc: use the doc. Build it, install + it, clean it, etc. + +Tue Feb 18 02:21:25 1992 K. Richard Pixley (rich at cygnus.com) + + * read.c: white space and comments only. + + * configure.in: use the new atof-ns32.c for ns32k. + + * write.c: comment change only. + +Tue Feb 18 02:11:10 1992 K. Richard Pixley (rich at cygnus.com) + + * config/tc-m88k.[hc]: pulled in from hack's unfinished work. These + aren't yet integrated. + + * config/tc-i860.[hc]: blew off the dust. Something must still be + done about conflicting relocation types. + + * config/tc-ns32k.c: Replaced previous tc_aout_fix_to_chars stub + with the real thing. + + * config/tc-i960.c, tc-sparc.c: white space and comments only. + + * config/tc-a29k.h: delete duplicate macro definition. + + * new file config/atof-ns32k.c copied from hack's last unreleased + gas. + +Mon Feb 17 07:51:06 1992 K. Richard Pixley (rich at cygnus.com) + + * config/tc-ns32k.c: actually make tc_aout_fix_to_chars work + rather than abort. + + * nearly everything. flush ChangeLog, package as gas-1.92.1. + ChangeLog's prior to this are sketchy at best. I have logs. + They just aren't ChangeLogs. + |
