gprof/TODO

   1 Sun Feb  5 16:27:32 1995
   2
   3 - documentation
   4 - optimize bfd_find_nearest_line_num() (or replace by different interface)
   5 - add support for prof file format so that prof files can be displayed
   6   at the line-level (this is useful for the uprofile tool under DEC's
   7   OSF/1)
   8
   9 + cleanup _bfd_ecoff_find_nearest_line_num() fixes & description
  10 + ensure "cc -pg" produces good files under OSF/1 v3.0
  11 + make sure gprof works together with OSF/1 v3.0's profiling libraries
  12 + implement symtab_parse(); modify sym_lookup() to consider addr_high
  13 + change gprof.c to collect lists, then invoke symtab_parse() for
  14   each list
  15 + Questions:
  16         o is -c (--static-call-graph) useful at all?  i can't see
  17           how; if it were deleted, gprof would be completely machine
  18           independent => yup, it is
  19         o are (long) option names appropriate?
  20         o -k (--exclude-arc) cannot be implemented with getopt();
  21           is new syntax (-k from/to) acceptable?  If not, how to
  22           fix it?
  23         o in the FSF output, the call-graph index now prints
  24           the filename of static functions in parentheses; e.g.,
  25           static function foo() that is defined in file bar.c
  26           would be printed as:
  27
  28                         [4] foo (bar.c)
  29
  30           is this acceptable?  should it be done only optionally?
  31         o symbols with addresses that map back to a different
  32           name are suppressed (happens with labels, for example);
  33           is this acceptable?  should it be done only optionally?
  34 + generalize to allow arbitrary histograms (not just time histograms)
  35 + basic-block information currently replaces all symbols created from
  36   the core because of an ugly ordering conflict---for now, the current
  37   solution works, but something cleaner is desirable ==> cleaned up,
  38   but it's slower now
  39 + convert to very new file format (back to trivial format, that is :)
  40 + replace "dummy.h" for Alpha (if there is any use to it)
  41 + add support for execution time profiling at a basic-block level
  42 + fix filename-off-by-one bug for Alpha (see ~/tmp/d.[ch])---no longer
  43   relevant
  44 + "-pg -a" doesn't work as expected because mcleanup() will overwrite
  45   the file generated by __bb_exit_func() (or vice versa)
  46 + first basic-block of fac() seems to get credited to last basic-block
  47   of previous function => bug in basic_blocks.c
  48 + flat profile should provide automatic scaling for per-call times because
  49   otherwise they'll always be zero on a fast machine with tons of small
  50   functions
  51 + make "-a" imply to retain line number info (without actually generating
  52   the debugging information (unless -g is specified)---no, this is a
  53   bad idea, because it is not clear what level of debugging info should
  54   be requested (e.g., -g vs. -g3); leaving it up to the user seems best
  55 + add long options support (or at least use getopt instead of ad-hoc
  56   implementation)
  57 + split into files according to abstract objects that are manipulated
  58 + replace sccsid by rcsid & add "end of ..." to every .c file
  59 + use DBG() everywhere
  60 + fix spacing (" ," -> "," etc.)
  61 + use DEFUNs everywhere
  62 + make compile cleanly with -Wall
  63 + "gcc -pg -O2" doesn't work on tecc.c unless -fno-omit-frame-pointer is
  64   specified; find out why
  65 + make things portable (prototypes, const, etc.)
  66 + if NEW_GMON_OUT is not defined, have a flag that will allow to
  67   read new gmon.out style files.  The idea being that everyone
  68   will use the new format for basic-block style profiling but
  69   the old format for regular gpprofiling