view src/unexhp9k800.c @ 665:fdefd0186b75

[xemacs-hg @ 2001-09-20 06:28:42 by ben] The great integral types renaming. The purpose of this is to rationalize the names used for various integral types, so that they match their intended uses and follow consist conventions, and eliminate types that were not semantically different from each other. The conventions are: -- All integral types that measure quantities of anything are signed. Some people disagree vociferously with this, but their arguments are mostly theoretical, and are vastly outweighed by the practical headaches of mixing signed and unsigned values, and more importantly by the far increased likelihood of inadvertent bugs: Because of the broken "viral" nature of unsigned quantities in C (operations involving mixed signed/unsigned are done unsigned, when exactly the opposite is nearly always wanted), even a single error in declaring a quantity unsigned that should be signed, or even the even more subtle error of comparing signed and unsigned values and forgetting the necessary cast, can be catastrophic, as comparisons will yield wrong results. -Wsign-compare is turned on specifically to catch this, but this tends to result in a great number of warnings when mixing signed and unsigned, and the casts are annoying. More has been written on this elsewhere. -- All such quantity types just mentioned boil down to EMACS_INT, which is 32 bits on 32-bit machines and 64 bits on 64-bit machines. This is guaranteed to be the same size as Lisp objects of type `int', and (as far as I can tell) of size_t (unsigned!) and ssize_t. The only type below that is not an EMACS_INT is Hashcode, which is an unsigned value of the same size as EMACS_INT. -- Type names should be relatively short (no more than 10 characters or so), with the first letter capitalized and no underscores if they can at all be avoided. -- "count" == a zero-based measurement of some quantity. Includes sizes, offsets, and indexes. -- "bpos" == a one-based measurement of a position in a buffer. "Charbpos" and "Bytebpos" count text in the buffer, rather than bytes in memory; thus Bytebpos does not directly correspond to the memory representation. Use "Membpos" for this. -- "Char" refers to internal-format characters, not to the C type "char", which is really a byte. -- For the actual name changes, see the script below. I ran the following script to do the conversion. (NOTE: This script is idempotent. You can safely run it multiple times and it will not screw up previous results -- in fact, it will do nothing if nothing has changed. Thus, it can be run repeatedly as necessary to handle patches coming in from old workspaces, or old branches.) There are two tags, just before and just after the change: `pre-integral-type-rename' and `post-integral-type-rename'. When merging code from the main trunk into a branch, the best thing to do is first merge up to `pre-integral-type-rename', then apply the script and associated changes, then merge from `post-integral-type-change' to the present. (Alternatively, just do the merging in one operation; but you may then have a lot of conflicts needing to be resolved by hand.) Script `fixtypes.sh' follows: ----------------------------------- cut ------------------------------------ files="*.[ch] s/*.h m/*.h config.h.in ../configure.in Makefile.in.in ../lib-src/*.[ch] ../lwlib/*.[ch]" gr Memory_Count Bytecount $files gr Lstream_Data_Count Bytecount $files gr Element_Count Elemcount $files gr Hash_Code Hashcode $files gr extcount bytecount $files gr bufpos charbpos $files gr bytind bytebpos $files gr memind membpos $files gr bufbyte intbyte $files gr Extcount Bytecount $files gr Bufpos Charbpos $files gr Bytind Bytebpos $files gr Memind Membpos $files gr Bufbyte Intbyte $files gr EXTCOUNT BYTECOUNT $files gr BUFPOS CHARBPOS $files gr BYTIND BYTEBPOS $files gr MEMIND MEMBPOS $files gr BUFBYTE INTBYTE $files gr MEMORY_COUNT BYTECOUNT $files gr LSTREAM_DATA_COUNT BYTECOUNT $files gr ELEMENT_COUNT ELEMCOUNT $files gr HASH_CODE HASHCODE $files ----------------------------------- cut ------------------------------------ `fixtypes.sh' is a Bourne-shell script; it uses 'gr': ----------------------------------- cut ------------------------------------ #!/bin/sh # Usage is like this: # gr FROM TO FILES ... # globally replace FROM with TO in FILES. FROM and TO are regular expressions. # backup files are stored in the `backup' directory. from="$1" to="$2" shift 2 echo ${1+"$@"} | xargs global-replace "s/$from/$to/g" ----------------------------------- cut ------------------------------------ `gr' in turn uses a Perl script to do its real work, `global-replace', which follows: ----------------------------------- cut ------------------------------------ : #-*- Perl -*- ### global-modify --- modify the contents of a file by a Perl expression ## Copyright (C) 1999 Martin Buchholz. ## Copyright (C) 2001 Ben Wing. ## Authors: Martin Buchholz <martin@xemacs.org>, Ben Wing <ben@xemacs.org> ## Maintainer: Ben Wing <ben@xemacs.org> ## Current Version: 1.0, May 5, 2001 # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2, or (at your option) # any later version. # # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # # You should have received a copy of the GNU General Public License # along with XEmacs; see the file COPYING. If not, write to the Free # Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA # 02111-1307, USA. eval 'exec perl -w -S $0 ${1+"$@"}' if 0; use strict; use FileHandle; use Carp; use Getopt::Long; use File::Basename; (my $myName = $0) =~ s@.*/@@; my $usage=" Usage: $myName [--help] [--backup-dir=DIR] [--line-mode] [--hunk-mode] PERLEXPR FILE ... Globally modify a file, either line by line or in one big hunk. Typical usage is like this: [with GNU print, GNU xargs: guaranteed to handle spaces, quotes, etc. in file names] find . -name '*.[ch]' -print0 | xargs -0 $0 's/\bCONST\b/const/g'\n [with non-GNU print, xargs] find . -name '*.[ch]' -print | xargs $0 's/\bCONST\b/const/g'\n The file is read in, either line by line (with --line-mode specified) or in one big hunk (with --hunk-mode specified; it's the default), and the Perl expression is then evalled with \$_ set to the line or hunk of text, including the terminating newline if there is one. It should destructively modify the value there, storing the changed result in \$_. Files in which any modifications are made are backed up to the directory specified using --backup-dir, or to `backup' by default. To disable this, use --backup-dir= with no argument. Hunk mode is the default because it is MUCH MUCH faster than line-by-line. Use line-by-line only when it matters, e.g. you want to do a replacement only once per line (the default without the `g' argument). Conversely, when using hunk mode, *ALWAYS* use `g'; otherwise, you will only make one replacement in the entire file! "; my %options = (); $Getopt::Long::ignorecase = 0; &GetOptions ( \%options, 'help', 'backup-dir=s', 'line-mode', 'hunk-mode', ); die $usage if $options{"help"} or @ARGV <= 1; my $code = shift; die $usage if grep (-d || ! -w, @ARGV); sub SafeOpen { open ((my $fh = new FileHandle), $_[0]); confess "Can't open $_[0]: $!" if ! defined $fh; return $fh; } sub SafeClose { close $_[0] or confess "Can't close $_[0]: $!"; } sub FileContents { my $fh = SafeOpen ("< $_[0]"); my $olddollarslash = $/; local $/ = undef; my $contents = <$fh>; $/ = $olddollarslash; return $contents; } sub WriteStringToFile { my $fh = SafeOpen ("> $_[0]"); binmode $fh; print $fh $_[1] or confess "$_[0]: $!\n"; SafeClose $fh; } foreach my $file (@ARGV) { my $changed_p = 0; my $new_contents = ""; if ($options{"line-mode"}) { my $fh = SafeOpen $file; while (<$fh>) { my $save_line = $_; eval $code; $changed_p = 1 if $save_line ne $_; $new_contents .= $_; } } else { my $orig_contents = $_ = FileContents $file; eval $code; if ($_ ne $orig_contents) { $changed_p = 1; $new_contents = $_; } } if ($changed_p) { my $backdir = $options{"backup-dir"}; $backdir = "backup" if !defined ($backdir); if ($backdir) { my ($name, $path, $suffix) = fileparse ($file, ""); my $backfulldir = $path . $backdir; my $backfile = "$backfulldir/$name"; mkdir $backfulldir, 0755 unless -d $backfulldir; print "modifying $file (original saved in $backfile)\n"; rename $file, $backfile; } WriteStringToFile ($file, $new_contents); } } ----------------------------------- cut ------------------------------------ In addition to those programs, I needed to fix up a few other things, particularly relating to the duplicate definitions of types, now that some types merged with others. Specifically: 1. in lisp.h, removed duplicate declarations of Bytecount. The changed code should now look like this: (In each code snippet below, the first and last lines are the same as the original, as are all lines outside of those lines. That allows you to locate the section to be replaced, and replace the stuff in that section, verifying that there isn't anything new added that would need to be kept.) --------------------------------- snip ------------------------------------- /* Counts of bytes or chars */ typedef EMACS_INT Bytecount; typedef EMACS_INT Charcount; /* Counts of elements */ typedef EMACS_INT Elemcount; /* Hash codes */ typedef unsigned long Hashcode; /* ------------------------ dynamic arrays ------------------- */ --------------------------------- snip ------------------------------------- 2. in lstream.h, removed duplicate declaration of Bytecount. Rewrote the comment about this type. The changed code should now look like this: --------------------------------- snip ------------------------------------- #endif /* The have been some arguments over the what the type should be that specifies a count of bytes in a data block to be written out or read in, using Lstream_read(), Lstream_write(), and related functions. Originally it was long, which worked fine; Martin "corrected" these to size_t and ssize_t on the grounds that this is theoretically cleaner and is in keeping with the C standards. Unfortunately, this practice is horribly error-prone due to design flaws in the way that mixed signed/unsigned arithmetic happens. In fact, by doing this change, Martin introduced a subtle but fatal error that caused the operation of sending large mail messages to the SMTP server under Windows to fail. By putting all values back to be signed, avoiding any signed/unsigned mixing, the bug immediately went away. The type then in use was Lstream_Data_Count, so that it be reverted cleanly if a vote came to that. Now it is Bytecount. Some earlier comments about why the type must be signed: This MUST BE SIGNED, since it also is used in functions that return the number of bytes actually read to or written from in an operation, and these functions can return -1 to signal error. Note that the standard Unix read() and write() functions define the count going in as a size_t, which is UNSIGNED, and the count going out as an ssize_t, which is SIGNED. This is a horrible design flaw. Not only is it highly likely to lead to logic errors when a -1 gets interpreted as a large positive number, but operations are bound to fail in all sorts of horrible ways when a number in the upper-half of the size_t range is passed in -- this number is unrepresentable as an ssize_t, so code that checks to see how many bytes are actually written (which is mandatory if you are dealing with certain types of devices) will get completely screwed up. --ben */ typedef enum lstream_buffering --------------------------------- snip ------------------------------------- 3. in dumper.c, there are four places, all inside of switch() statements, where XD_BYTECOUNT appears twice as a case tag. In each case, the two case blocks contain identical code, and you should *REMOVE THE SECOND* and leave the first.
author ben
date Thu, 20 Sep 2001 06:31:11 +0000
parents abe6d1db359e
children 04bc9d2f42c7
line wrap: on
line source

/* Unexec for HP 9000 Series 800 machines.
   Bob Desinger <hpsemc!bd@hplabs.hp.com>

This file is part of XEmacs.

XEmacs is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2, or (at your option) any
later version.

XEmacs is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
for more details.

You should have received a copy of the GNU General Public License
along with XEmacs; see the file COPYING.  If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
Boston, MA 02111-1307, USA.  */

/* Synched up with: Not synched with FSF. */

/*

  Unexec creates a copy of the old a.out file, and replaces the old data
  area with the current data area.  When the new file is executed, the
  process will see the same data structures and data values that the
  original process had when unexec was called.

  Unlike other versions of unexec, this one copies symbol table and
  debug information to the new a.out file.  Thus, the new a.out file
  may be debugged with symbolic debuggers.

  If you fix any bugs in this, I'd like to incorporate your fixes.
  Send them to uunet!hpda!hpsemc!jmorris or jmorris%hpsemc@hplabs.HP.COM.

  CAVEATS:
  This routine saves the current value of all static and external
  variables.  This means that any data structure that needs to be
  initialized must be explicitly reset.  Variables will not have their
  expected default values.

  Unfortunately, the HP-UX signal handler has internal initialization
  flags which are not explicitly reset.  Thus, for signals to work in
  conjunction with this routine, the following code must executed when
  the new process starts up.

  void _sigreturn();
  ...
  sigsetreturn(_sigreturn);
*/


#include <config.h>
#include <stdlib.h>
#include <stdio.h>
#include <fcntl.h>
#include <errno.h>

#include <a.out.h>
#include "lisp.h"

/*
 * Minor modification to enable dumping with shared libraries added by
 * Dipankar Gupta (dg@hplb.hpl.hp.com). I studied Oliver Laumann's
 * more elaborate dynamic loading scheme in ELK while implementing
 * this, but don't use any of his machinery.
 *
 * Stores the BRK value at dump time, and uses the RUN_TIME_REMAP hook
 * to break back to the stored value when the dumped executable is restarted.
 *
 * CAVEATS (addenda):
 * 1. Text area of the shlibs are not stored. Thus, if a shared library is
 *    replaced between the time of dump and execution, all bets are off.
 *
 * 2. Assumes that the data and bss area are adjacent, which is true of the
 *    current VM implementation.
 *
 * 3. Any setup that defines HPUX_USE_SHLIBS *must* also define
 *    RUN_TIME_REMAP.
 */

#ifdef HPUX_USE_SHLIBS
#include <dl.h>			/* User-space dynamic loader entry points */
static void Save_Shared_Data (void);
static void Restore_Shared_Data (void);
#endif

void write_header(int file, struct header *hdr, struct som_exec_auxhdr *auxhdr);
void read_header (int file, struct header *hdr, struct som_exec_auxhdr *auxhdr);
void save_data_space (int file, struct header *hdr,
		      struct som_exec_auxhdr *auxhdr, int size);
void copy_rest (int old, int new);
void copy_file (int old, int new, int size);
void update_file_ptrs(int file, struct header *hdr,
		      struct som_exec_auxhdr *auxhdr,
		      unsigned int location, int offset);
int calculate_checksum(struct header *hdr);

/* Create a new a.out file, same as old but with current data space */
int
unexec (char *new_name,		/* name of the new a.out file to be created */
        char *old_name,		/* name of the old a.out file */
        uintptr_t new_end_of_text, /* ptr to new edata/etext; NOT USED YET */
        uintptr_t dummy1, uintptr_t dummy2) /* not used by emacs */
{
  int old, new;
  int old_size, new_size;
  struct header hdr;
  struct som_exec_auxhdr auxhdr;
  long i;

  /* For the greatest flexibility, should create a temporary file in
     the same directory as the new file.  When everything is complete,
     rename the temp file to the new name.
     This way, a program could update its own a.out file even while
     it is still executing.  If problems occur, everything is still
     intact.  NOT implemented.  */

  /* Open the input and output a.out files */
  old = open (old_name, O_RDONLY);
  if (old < 0)
    { perror(old_name); exit(1); }
  new = open (new_name, O_CREAT|O_RDWR|O_TRUNC, 0777);
  if (new < 0)
    { perror(new_name); exit(1); }

  /* Read the old headers */
  read_header(old, &hdr, &auxhdr);

#ifdef HPUX_USE_SHLIBS
  Save_Shared_Data(); /* Save break value (added: dg@hplb.hpl.hp.com) */
#endif
  /* Decide how large the new and old data areas are */
  old_size = auxhdr.exec_dsize;
  /* I suspect these two statements are separate
     to avoid a compiler bug in hpux version 8.  */
  i = (long) sbrk (0);
  new_size = i - auxhdr.exec_dmem;

  /* Copy the old file to the new, up to the data space */
  lseek(old, 0, 0);
  copy_file(old, new, auxhdr.exec_dfile);

  /* Skip the old data segment and write a new one */
  lseek(old, old_size, 1);
  save_data_space(new, &hdr, &auxhdr, new_size);

  /* Copy the rest of the file */
  copy_rest(old, new);

  /* Update file pointers since we probably changed size of data area */
  update_file_ptrs(new, &hdr, &auxhdr, auxhdr.exec_dfile, new_size-old_size);

  /* Save the modified header */
  write_header(new, &hdr, &auxhdr);

  /* Close the binary file */
  close (old);
  close (new);
  return 0;
}

/* Save current data space in the file, update header.  */

void
save_data_space (int file, struct header *hdr,
		 struct som_exec_auxhdr *auxhdr, int size)
{
  /* Write the entire data space out to the file */
  if (write(file, (void *)auxhdr->exec_dmem, size) != size)
    { perror("Can't save new data space"); exit(1); }

  /* Update the header to reflect the new data size */
  auxhdr->exec_dsize = size;
  auxhdr->exec_bsize = 0;
}

/* Update the values of file pointers when something is inserted.  */

void
update_file_ptrs(int file, struct header *hdr,
		 struct som_exec_auxhdr *auxhdr,
		 unsigned int location, int offset)
{
  struct subspace_dictionary_record subspace;
  int i;

  /* Increase the overall size of the module */
  hdr->som_length += offset;

  /* Update the various file pointers in the header */
#define update(ptr) if (ptr > location) ptr = ptr + offset
  update(hdr->aux_header_location);
  update(hdr->space_strings_location);
  update(hdr->init_array_location);
  update(hdr->compiler_location);
  update(hdr->symbol_location);
  update(hdr->fixup_request_location);
  update(hdr->symbol_strings_location);
  update(hdr->unloadable_sp_location);
  update(auxhdr->exec_tfile);
  update(auxhdr->exec_dfile);

  /* Do for each subspace dictionary entry */
  lseek(file, hdr->subspace_location, 0);
  for (i = 0; i < hdr->subspace_total; i++)
    {
      if (read(file, &subspace, sizeof(subspace)) != sizeof(subspace))
	{ perror("Can't read subspace record"); exit(1); }

      /* If subspace has a file location, update it */
      if (subspace.initialization_length > 0
	  && subspace.file_loc_init_value > location)
	{
	  subspace.file_loc_init_value += offset;
	  lseek(file, -sizeof(subspace), 1);
	  if (write(file, &subspace, sizeof(subspace)) != sizeof(subspace))
	    { perror("Can't update subspace record"); exit(1); }
	}
    }

  /* Do for each initialization pointer record */
  /* (I don't think it applies to executable files, only relocatables) */
#undef update
}

/* Read in the header records from an a.out file.  */

void
read_header(int file, struct header *hdr, struct som_exec_auxhdr *auxhdr)
{

  /* Read the header in */
  lseek(file, 0, 0);
  if (read(file, hdr, sizeof(*hdr)) != sizeof(*hdr))
    { perror("Couldn't read header from a.out file"); exit(1); }

  if (hdr->a_magic != EXEC_MAGIC && hdr->a_magic != SHARE_MAGIC
      &&  hdr->a_magic != DEMAND_MAGIC)
    {
      fprintf(stderr, "a.out file doesn't have legal magic number\n");
      exit(1);
    }

  lseek(file, hdr->aux_header_location, 0);
  if (read(file, auxhdr, sizeof(*auxhdr)) != sizeof(*auxhdr))
    {
      perror("Couldn't read auxiliary header from a.out file");
      exit(1);
    }
}

/* Write out the header records into an a.out file.  */
void
write_header(int file, struct header *hdr, struct som_exec_auxhdr *auxhdr)
{
  /* Update the checksum */
  hdr->checksum = calculate_checksum(hdr);

  /* Write the header back into the a.out file */
  lseek(file, 0, 0);
  if (write(file, hdr, sizeof(*hdr)) != sizeof(*hdr))
    { perror("Couldn't write header to a.out file"); exit(1); }
  lseek(file, hdr->aux_header_location, 0);
  if (write(file, auxhdr, sizeof(*auxhdr)) != sizeof(*auxhdr))
    { perror("Couldn't write auxiliary header to a.out file"); exit(1); }
}

/* Calculate the checksum of a SOM header record. */
int
calculate_checksum(struct header *hdr)
{
  int checksum, i, *ptr;

  checksum = 0;  ptr = (int *) hdr;

  for (i=0; i<sizeof(*hdr)/sizeof(int)-1; i++)
    checksum ^= ptr[i];

  return(checksum);
}

/* Copy size bytes from the old file to the new one.  */
void
copy_file (int old, int new, int size)
{
  int len;
  int buffer[8192];  /* word aligned will be faster */

  for (; size > 0; size -= len)
    {
      len = size < sizeof (buffer) ? size : sizeof (buffer);
      if (read (old, buffer, len) != len)
	{
	  perror ("Read failure on a.out file");
	  exit (1);
	}
      if (write (new, buffer, len) != len)
	{
	  perror ("Write failure in a.out file");
	  exit (1);
	}
    }
}

/* Copy the rest of the file, up to EOF.  */
void
copy_rest (int old, int new)
{
  int buffer[4096];
  int len;

  /* Copy bytes until end of file or error */
  while ( (len = read(old, buffer, sizeof(buffer))) > 0)
    if (write(new, buffer, len) != len) break;

  if (len != 0)
    { perror("Unable to copy the rest of the file"); exit(1); }
}

#ifdef	DEBUG
display_header(struct header *hdr, struct som_exec_auxhdr *auxhdr)
{
  /* Display the header information (debug) */
  printf("\n\nFILE HEADER\n");
  printf("magic number %d \n", hdr->a_magic);
  printf("text loc %.8x   size %d \n", auxhdr->exec_tmem, auxhdr->exec_tsize);
  printf("data loc %.8x   size %d \n", auxhdr->exec_dmem, auxhdr->exec_dsize);
  printf("entry     %x \n",   auxhdr->exec_entry);
  printf("Bss  segment size %u\n", auxhdr->exec_bsize);
  printf("\n");
  printf("data file loc %d    size %d\n",
	 auxhdr->exec_dfile, auxhdr->exec_dsize);
  printf("som_length %d\n", hdr->som_length);
  printf("unloadable sploc %d    size %d\n",
	 hdr->unloadable_sp_location, hdr->unloadable_sp_size);
}
#endif /* DEBUG */

#ifdef HPUX_USE_SHLIBS
/* Added machinery for shared libs... see comments at the beginning of this file. */

void *Brk_On_Dump = 0;		/* Brk value to restore... stored as a global */

static void
Save_Shared_Data (void)
{
  Brk_On_Dump = sbrk (0);
}

static void
Restore_Shared_Data (void)
{
  brk (Brk_On_Dump);
}

/* run_time_remap is the magic called by startup code in the dumped executable
   if RUN_TIME_REMAP is set. */
int
run_time_remap (char *dummy)
{
  Restore_Shared_Data ();
  return 0;
}
#endif /* HPUX_USE_SHLIBS */