gas/NOTES

   1 -*- text -*-
   2
   3 PORTING:
   4
   5 Sorry, no description of the interfaces is written up yet.  Look at existing
   6 back ends and work from there.
   7
   8 New hosts: If your host system has a strange header file setup, create a
   9 config/ho-foo.h file for it and include the appropriate header files or
  10 definitions there.  If your host has a broken compiler, or some broken macros
  11 in header files, create a host-specific file and repair the damage there.
  12 (See, for example, ho-rs6000.h.  The "assert" macro on that system doesn't work
  13 right, and a flag is set to rewrite an expression in tc-m68k.c that the native
  14 compiler mis-compiles.)
  15
  16 New target formats: Look at the BFD_ASSEMBLER code.  The a.out code might be a
  17 fair example.  There are no "good" examples yet, unfortunately, nor any good
  18 documentation of the changes.
  19
  20 New target processors: Check first to see if the BFD_ASSEMBLER interface is
  21 supported by the file format code you need to use.
  22
  23 New environments: ???
  24
  25 DOCUMENTATION:
  26
  27 The internals of gas need documenting.
  28
  29 The documentation should also contain a "Trouble" section similar to gcc's
  30 manual: real bugs, common problems, incompatibilities, etc.
  31
  32 Anyone want to offer to maintain a man page?
  33
  34 BFD CONVERSION:
  35
  36 The "#ifdef BFD_ASSEMBLER" code is on its way in; the "#ifndef BFD_ASSEMBLER"
  37 code is on its way out.  The new code uses BFD data structures, and calls BFD
  38 for anything that needs to be written to the output file.  The old code did all
  39 the writing itself, or in a couple of cases, used BFD as a slightly higher
  40 level than stdio (i.e., bfd_seek, bfd_write -- these are not the preferred
  41 interface).
  42
  43 Because of this, some of this code is messy.  Lots of ifdef's, and the
  44 non-BFD_ASSEMBLER version often has multiple conditional tests inside it for
  45 various processors or formats.  As the various targets get converted over,
  46 these will gradually go away.
  47
  48 As of the moment I'm editing this file, only the "sun4" and "decstation-bsd"
  49 targets can really use the BFD code.  Other back ends still need merging or
  50 touching up.
  51
  52 TO DO:
  53
  54 Remove DONTDEF code, commented-out code.
  55
  56 Eliminate, as much as possible, anything not in config that is conditionalized
  57 on a CPU, format, or environment.
  58
  59 Merge COFF support into one version, supporting all the pseudo-ops used in
  60 either versions now, but using BFD for high-level operations.  (See second
  61 following item.)  Currently there are two versions (plus the new BFD code),
  62 which support different features, and are used on different targets.
  63
  64 Convert remaining a.out/b.out targets to using the BFD_ASSEMBLER code by
  65 default.
  66
  67 Finish conversion to using BFD for all object file writing.  (This is the
  68 BFD_ASSEMBLER code, not BFD or BFD_HEADERS.)  VMS might be the tough one here,
  69 since there's no BFD support for it at all yet.  Eliminate the old code.  Some
  70 of this can be done target by target, so doing a target where the CPU or
  71 format already supports BFD_ASSEMBLER mode may be easiest.
  72
  73 Fix lots of uses of empty strings to use null pointers.  Will improve
  74 efficiency, and should make code clearer too.
  75
  76 Clean up comments; lots of 'em are one previous maintainer griping about
  77 another previous maintainer, unrelated to the code.  (And with no names,
  78 they're not so fun to read. :-)
  79
  80 For sparc: "call 0" becomes "jmpl %g0,%l7", and similarly for absolute
  81 addresses in -4096...4095.  (Solaris assembler does this.  No
  82 relocation required, no absolute symbol needed.)  For addresses
  83 outside the range, for COFF, keep generating an absolute symbol to use
  84 for relocs.
  85
  86 Get Steve to document H8/500 stuff (and others).
  87
  88 Improve test suite.  Incorporate more reported net bugs, and non-confidential
  89 Cygnus customer bugs, and anything else.
  90
  91 Add support for i386/i486 16-bit mode, so operating system initialization code
  92 doesn't require a separate assembler nor lots of `.byte' directives.
  93
  94 See if it's more maintainable (and not too much of a performance loss) to use
  95 a yacc grammar for parsing input.  The lexer will have to be flexible, and the
  96 grammar will have to contain any construct used on any platform, but it may be
  97 easier to maintain, instead of having code in most of the back ends.
  98
  99 PIC support.
 100
 101 Torbjorn Granlund <tege@cygnus.com> writes, regarding alpha .align:
 102
 103    Please make sure the .align directive works as in digital's assembler.
 104    They fill the space with a sequence of "bis $31,$31,$31;ldq_u $31,0($30)"
 105    since these two instructions can dual-issue.  Since .align is ued a lot by
 106    gcc, it is an important optimization.
 107
 108 Torbjorn Granlund <tege@cygnus.com> writes, regarding i386/i486/pentium:
 109
 110    In a new publication from Intel, "Optimization for Intel's 32 bit
 111    Processors", they recommended code alignment on a 16 byte boundary if that
 112    requires less than 8 bytes of fill instructions.  The Pentium is not
 113    affected by such alignment, the 386 wants alignment on a 4 byte boundary.
 114    It is the 486 that is most helped by large alignment.
 115
 116    Recommended nop instructions:
 117    1 byte:  90                     xchg %eax,%eax
 118    2 bytes: 8b c0                  movl %eax,%eax
 119    3 bytes: 8d 76 00               leal 0(%esi),%esi
 120    4 bytes: 8d 74 26 00            leal 0(%esi),%esi
 121    5 bytes: 8b c0 8d 76 00         movl %eax,%eax; leal 0(%esi),%esi
 122    6 bytes: 8d b6 00 00 00 00      leal 0(%esi),%esi
 123    7 bytes: 8d b4 26 00 00 00 00   leal 0(%esi),%esi
 124
 125    Note that `leal 0(%esi),%esi' has a few different encodings...
 126
 127    There are faster instructions for certain lengths, that are not true nops.
 128    If you can determine that a register and the condition code is dead (by
 129    scanning forwards for a register that is written before it is read, and
 130    similar for cc) you can use a `incl reg' for a 3 times faster 1 cycle nop...
 131
 132 (From old "NOTES" file to-do list, not really reviewed:)
 133
 134 fix relocation types for i860, perhaps by adding a ref pointer to fixS?
 135
 136 remove the ifdef's from fx_callj tests?
 137
 138 space tighten sparc alignment?
 139
 140 md_ => tc_
 141
 142 share b.out with a.out.