patch for unexmacosx.c: supporting non-prebound dynamic library

patch for unexmacosx.c: supporting non-prebound dynamic library

[Nozomu Ando]

[-- Attachment #1: Type: text/plain, Size: 351 bytes --]

Hello all,

On Mac OS X, emacs of CVS HEAD cannot unexec
with non-prebound dynamically linked library (e.g.  libXaw.dylib).

I have written some code for handle this situation.

With the following patch,
I have built Emacs/Carbon and Emacs/X11-dylib on Mac OS X
successfully, and the both seem to work fine.

I hope this helps.

Regards,
Nozomu Ando


[-- Attachment #2: patch-unexmacosx.txt --]
[-- Type: text/plain, Size: 5272 bytes --]

Index: mac/INSTALL
===================================================================
RCS file: /cvsroot/emacs/emacs/mac/INSTALL,v
retrieving revision 1.14
diff -u -r1.14 INSTALL
--- mac/INSTALL	4 Feb 2003 14:12:13 -0000	1.14
+++ mac/INSTALL	12 Oct 2003 13:23:01 -0000
@@ -58,16 +58,11 @@
 correctly.  You may want to create a symlink or alias to this path to
 quickly access both the terminal and GUI versions.
 
-If you are building Emacs to run on Mac OS X and X Window, you need to
-create a directory containing statically-linked X libraries.
+If you are building Emacs to run on Mac OS X and X Window,
+instead of typing `./configure' above, type
 
-  sudo mkdir /usr/X11R6/libstatic
-  cd /usr/X11R6/libstatic
-  sudo ln -s ../lib/lib*.a ../lib/X11 .
+  ./configure --without-carbon --with-x
 
-Instead of typing `./configure' above, type
-
-  ./configure --without-carbon --with-x --x-libraries=/usr/X11R6/libstatic
 
 To use colors in a terminal, put the following lines in the file
 ~/.termcap and log in again.
@@ -159,11 +154,6 @@
 	      	    $prefix/bin/emacs to reduce disk space.  Note, this
 		    option may removed in the future.
 
-If you are intending to build a binary distribution for X windows, you
-will probably want to follow the directions above to create static
-X11R6 libraries and run the make-package script like this
-
-./make-package --with-x -C,--x-libraries=/usr/X11R6/libstatic
 
 For usage of other options, use the --help option.
 
Index: src/unexmacosx.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/unexmacosx.c,v
retrieving revision 1.7
diff -u -r1.7 unexmacosx.c
--- src/unexmacosx.c	1 Sep 2003 15:45:57 -0000	1.7
+++ src/unexmacosx.c	12 Oct 2003 13:23:04 -0000
@@ -95,6 +95,10 @@
 #include <unistd.h>
 #include <mach/mach.h>
 #include <mach-o/loader.h>
+#include <mach-o/reloc.h>
+#if defined (__ppc__)
+#include <mach-o/ppc/reloc.h>
+#endif
 #include <objc/malloc.h>
 
 #define VERBOSE 1
@@ -158,6 +162,11 @@
 
 malloc_zone_t *emacs_zone;
 
+/* file offset of input file's data segment */
+off_t data_segment_old_fileoff;
+
+struct segment_command *data_segment_scp;
+
 /* Read n bytes from infd into memory starting at address dest.
    Return true if successful, false otherwise.  */
 static int
@@ -763,6 +772,65 @@
   curr_header_offset += lc->cmdsize;
 }
 
+/* Fix up relocation entries. */
+static void
+unrelocate (const char *name, off_t reloff, int nrel)
+{
+  int i, unreloc_count;
+  struct relocation_info reloc_info;
+  struct scattered_relocation_info *sc_reloc_info
+    = (struct scattered_relocation_info *) &reloc_info;
+
+  for (unreloc_count = 0, i = 0; i < nrel; i++)
+    {
+      if (lseek (infd, reloff, L_SET) != reloff)
+	unexec_error ("unrelocate: %s:%d cannot seek to reloc_info", name, i);
+      if (!unexec_read (&reloc_info, sizeof (reloc_info)))
+	unexec_error ("unrelocate: %s:%d cannot read reloc_info", name, i);
+      reloff += sizeof (reloc_info);
+
+      if (sc_reloc_info->r_scattered == 0)
+	switch (reloc_info.r_type)
+	  {
+	  case GENERIC_RELOC_VANILLA:
+	    if (reloc_info.r_address >= data_segment_scp->vmaddr
+		&& reloc_info.r_address < (data_segment_scp->vmaddr
+					   + data_segment_scp->vmsize))
+	      {
+		off_t src_off = data_segment_old_fileoff
+		  + reloc_info.r_address - data_segment_scp->vmaddr;
+		off_t dst_off = data_segment_scp->fileoff
+		  + reloc_info.r_address - data_segment_scp->vmaddr;
+
+		if (!unexec_copy (dst_off, src_off, 1 << reloc_info.r_length))
+		  unexec_error ("unrelocate: %s:%d cannot copy original value",
+				name, i);
+		unreloc_count++;
+	      }
+	    break;
+	  default:
+	    unexec_error ("unrelocate: %s:%d cannot handle type = %d",
+			  name, i, reloc_info.r_type);
+	  }
+      else
+	switch (sc_reloc_info->r_type)
+	  {
+#if defined (__ppc__)
+	  case PPC_RELOC_PB_LA_PTR:
+	    /* nothing to do for prebound lazy pointer */
+	    break;
+#endif
+	  default:
+	    unexec_error ("unrelocate: %s:%d cannot handle scattered type = %d",
+			  name, i, sc_reloc_info->r_type);
+	  }
+    }
+
+  if (nrel > 0)
+    printf ("Fixed up %d/%d %s relocation entries in data segment.\n",
+	    unreloc_count, nrel, name);
+}
+
 /* Copy a LC_DYSYMTAB load command from the input file to the output
    file, adjusting the file offset fields.  */
 static void
@@ -770,10 +838,8 @@
 {
   struct dysymtab_command *dstp = (struct dysymtab_command *) lc;
 
-  /* If Mach-O executable is not prebound, relocation entries need
-     fixing up.  This is not supported currently.  */
-  if (!(mh.flags & MH_PREBOUND) && (dstp->nextrel != 0 || dstp->nlocrel != 0))
-    unexec_error ("cannot handle LC_DYSYMTAB with relocation entries");
+  unrelocate ("local", dstp->locreloff, dstp->nlocrel);
+  unrelocate ("external", dstp->extreloff, dstp->nextrel);
 
   if (dstp->nextrel > 0) {
     dstp->extreloff += delta;
@@ -845,6 +911,11 @@
 	  struct segment_command *scp = (struct segment_command *) lca[i];
 	  if (strncmp (scp->segname, SEG_DATA, 16) == 0)
 	    {
+	      /* save data segment file offset and segment_command for
+		 unrelocate */
+	      data_segment_old_fileoff = scp->fileoff;
+	      data_segment_scp = scp;
+
 	      copy_data_segment (lca[i]);
 	    }
 	  else

[-- Attachment #3: Type: text/plain, Size: 141 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/emacs-devel

Re: patch for unexmacosx.c: supporting non-prebound dynamic library

[Andrew Choi]

Nozomu Ando <nand@mac.com> writes:

> Hello all,
>
> On Mac OS X, emacs of CVS HEAD cannot unexec
> with non-prebound dynamically linked library (e.g.  libXaw.dylib).
>
> I have written some code for handle this situation.
>
> With the following patch,
> I have built Emacs/Carbon and Emacs/X11-dylib on Mac OS X
> successfully, and the both seem to work fine.
>
> I hope this helps.
>
> Regards,
> Nozomu Ando

Hi Nozumu,

Thank you again for your contribution.  I have just checked your patch
into CVS.

Andrew.

Do we need legal papers [was: patch for unexmacosx.c: supporting non-prebound dynamic library]

[Kim F. Storm]

Andrew Choi <akochoi@shaw.ca> writes:

> Nozomu Ando <nand@mac.com> writes:
> 
> > Hello all,
> >
> > On Mac OS X, emacs of CVS HEAD cannot unexec
> > with non-prebound dynamically linked library (e.g.  libXaw.dylib).
> >
> > I have written some code for handle this situation.
> >
> > With the following patch,
> > I have built Emacs/Carbon and Emacs/X11-dylib on Mac OS X
> > successfully, and the both seem to work fine.
> >
> > I hope this helps.
> >
> > Regards,
> > Nozomu Ando
> 
> Hi Nozumu,
> 
> Thank you again for your contribution.  I have just checked your patch
> into CVS.
> 
> Andrew.

Hi Andrew,

I looked at the patch, and it seems to be too big (it adds 60+ new
source lines) to install without legal papers, and I cannot see that
Nozumu Ando has signed papers.

I think you should check with RMS if we need papers for this.

Kim

Re: Do we need legal papers [was: patch for unexmacosx.c: supporting non-prebound dynamic library]

[Andrew Choi]

storm@cua.dk (Kim F. Storm) writes:

> Hi Andrew,
>
> I looked at the patch, and it seems to be too big (it adds 60+ new
> source lines) to install without legal papers, and I cannot see that
> Nozumu Ando has signed papers.
>
> I think you should check with RMS if we need papers for this.
>
> Kim

We did already.  RMS handled this himself.

Re: Do we need legal papers [was: patch for unexmacosx.c: supporting non-prebound dynamic library]

[Richard Stallman]

    > I think you should check with RMS if we need papers for this.
    >
    > Kim

    We did already.  RMS handled this himself.

I only handled the first step: requesting papers from Nozomu Ando.
There is no indication that we have received and filed his signed
papers yet.  Until they are received and filed, we cannot install his changes.

Did anything specific lead you to the conclusion that we had finished
handling his papers?

Re: Do we need legal papers

[Andrew Choi]

Richard Stallman <rms@gnu.org> writes:

>     > I think you should check with RMS if we need papers for this.
>     >
>     > Kim
>
>     We did already.  RMS handled this himself.
>
> I only handled the first step: requesting papers from Nozomu Ando.
> There is no indication that we have received and filed his signed
> papers yet.  Until they are received and filed, we cannot install his
> changes.
>
> Did anything specific lead you to the conclusion that we had finished
> handling his papers?

This is too much.  We got the following message.

-----
Date: Wed, 15 Oct 2003 22:47:10 +0900
From: Nozomu Ando <nand@mac.com>
Subject: Re: Your Emacs changes
To: rms@gnu.org
Cc: Nozomu Ando <nand@mac.com>, akochoi@shaw.ca

On Oct 14, 2003, at 3:02 PM, Richard Stallman wrote:

> Your Emacs changes have become big enough that we should get legal
> papers for them.  Are you willing to do that?
>
> If so, please email the following information to fsf-records@gnu.org,
> and we will send you the assignment form for your past and future
> changes.
>
> Please use your full legal name (in ASCII characters) as the subject
> line of the message.

Thank you for your considering my patch.
I sent the mail to fsf-records@gnu.org.

NOTE: My legal name (name on my passport) is "Yasuhito Komura".
("Nozomu Ando" is my pen name or such).
I'm sorry if this makes any confusing.

Regards,
Nozomu Ando

Re: Do we need legal papers

[Richard Stallman]

    > Did anything specific lead you to the conclusion that we had finished
    > handling his papers?

    This is too much.  We got the following message.

That message says that he began the process of filling out legal
papers.  He sent email to fsf-records, who presumably sent him papers
by snail mail.

Before we install the code, we have to wait for the signed papers to
be received and recorded.  (We may also need papers from his employer;
I do not know, but fsf-records would have told him.)

Have you read the file maintain.texi which explains these procedures?

Re: Do we need legal papers

[Andrew Choi]

Richard Stallman <rms@gnu.org> writes:

>     > Did anything specific lead you to the conclusion that we had finished
>     > handling his papers?
>
>     This is too much.  We got the following message.
>
> That message says that he began the process of filling out legal
> papers.  He sent email to fsf-records, who presumably sent him papers
> by snail mail.
>
> Before we install the code, we have to wait for the signed papers to
> be received and recorded.  (We may also need papers from his employer;
> I do not know, but fsf-records would have told him.)
>
> Have you read the file maintain.texi which explains these procedures?

Hi Richard,

Perhaps it is best if I'm relieved from being the Mac maintainer.  I
program for the love of programming and not for any other reasons.
These responsibilities that seem to be expected of a maintainer do not
suit me.  For the record I was never asked and I have never promised to
be a maintainer.  One day my name shows up in the MAINTAINER file and
that was it!  Please consider this my resignation as Mac maintainer of
Emacs as of today.

Andrew.

Re: Do we need legal papers

[David Kastrup]

Andrew Choi <akochoi@shaw.ca> writes:

> Richard Stallman <rms@gnu.org> writes:
> 
> > Before we install the code, we have to wait for the signed papers
> > to be received and recorded.  (We may also need papers from his
> > employer; I do not know, but fsf-records would have told him.)
> >
> > Have you read the file maintain.texi which explains these
> > procedures?
> 
> Perhaps it is best if I'm relieved from being the Mac maintainer.  I
> program for the love of programming and not for any other reasons.

The love of programming also implies that one wishes well for one's
outcome.  You are aware that at one point in Emacs' history large
parts had to be rewritten from scratch because somebody suddenly
decided that parts he contributed in a spirit of "love of programming"
should no longer get distributed under the GPL?

You are aware that just currently the company SCO is seeking for
billions of dollars of damage from users and contributors to Linux
because they claim that some contributions might not have been
properly legal?  You know that the resulting legal battle will drag on
for years and serve to seriously hamper Linux acceptance, because it
will be used for spreading fear about Linux use while the legal
struggles go on, even if one thinks that the outcome is preordained?

It's not even amiss to be careful when you never experienced
problems.  Unfortunately, there have been too many precedents.

I agree that the paperwork is a nuisance.  Yet I don't see that the
price to pay for it is not warranted.

> For the record I was never asked and I have never promised to be a
> maintainer.  One day my name shows up in the MAINTAINER file and
> that was it!

Of course, this is hardly how things should work.> 

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Do we need legal papers

[Kim F. Storm]

Andrew Choi <akochoi@shaw.ca> writes:

> 
> Perhaps it is best if I'm relieved from being the Mac maintainer.  I
> program for the love of programming and not for any other reasons.

Sure, we all love programming -- and that's why it is so important that
we all do whatever we can to protect the freedom to do so.  

For FSF/GNU projects like Emacs, the enforcement of the GPL and the
assignment of copyright to the FSF through legal papers is --
unfortunately -- the only way to do this.  

With Richard's non-compromising attitude on the subject of legal
papers, I believe that GNU emacs (as distributed by the FSF) cannot be
the target of similar FUD compains.

So Andrew, please view Richard's "invervention" as a defence for emacs
freedom, and reconsider you decision.



> These responsibilities that seem to be expected of a maintainer do not
> suit me.  

They don't "suit" any of us ... unfortunately, they are necessary to
preserve the freedom we have.

>           For the record I was never asked and I have never promised to
> be a maintainer.  One day my name shows up in the MAINTAINER file and
> that was it!  

I don't know who did that without consulting you -- but whoever did
it, most likely did it to acknowledge and in appreciation of your
contributions -- not to burden you with it.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Do we need legal papers

[Dave Carlton]

[-- Attachment #1.1: Type: text/plain, Size: 742 bytes --]

Andrew, I too would ask that you reconsider. I am a recent convert to 
EMACS and as you can see I only use a Macintosh and OSX. I would gladly 
take your place or help you out but I am just a neophyte when it comes 
to C and Lisp (I work in Forth and Open Firmware :). While I like the X 
version the problems of cut and paste (among others) between the OSX 
and X versions makes it too cumbersome for me to use.

So please please stay on board, I was so much looking forward to 21.4!

On Nov 04, 2003, at 16:19, Kim F. Storm wrote:

> Andrew Choi <akochoi@shaw.ca> writes:
>
>>
>> Perhaps it is best if I'm relieved from being the Mac maintainer.  I
>> program for the love of programming and not for any other reasons.
---
davec@apple.com


[-- Attachment #1.2: Type: text/enriched, Size: 829 bytes --]

Andrew, I too would ask that you reconsider. I am a recent convert to
EMACS and as you can see I only use a Macintosh and OSX. I would
gladly take your place or help you out but I am just a neophyte when
it comes to C and Lisp (I work in Forth and Open Firmware :). While I
like the X version the problems of cut and paste (among others)
between the OSX and X versions makes it too cumbersome for me to use. 


So please please stay on board, I was so much looking forward to 21.4!


On Nov 04, 2003, at 16:19, Kim F. Storm wrote:


<excerpt>Andrew Choi <<akochoi@shaw.ca> writes:


<excerpt>

Perhaps it is best if I'm relieved from being the Mac maintainer.  I

program for the love of programming and not for any other reasons.

</excerpt></excerpt><fontfamily><param>Lucida Grande</param>---

davec@apple.com

</fontfamily>


[-- Attachment #2: Type: text/plain, Size: 141 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/emacs-devel

Re: Do we need legal papers

[Joseph C. Slater]

On Nov 5, 2003, at 12:34 PM, Dave Carlton wrote:

> Andrew, I too would ask that you reconsider. I am a recent convert to 
> EMACS and as you can see I only use a Macintosh and OSX. I would 
> gladly  <snip>
> On Nov 04, 2003, at 16:19, Kim F. Storm wrote:
>
>> Andrew Choi <akochoi@shaw.ca> writes:
>>
>>>
>>> Perhaps it is best if I'm relieved from being the Mac maintainer.  I
>>> program for the love of programming and not for any other reasons.

Pretty please. With sugar on it!

\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
(+1) 937-775-5085
http://www.cs.wright.edu/~jslater

Re: Do we need legal papers

[Richard Stallman]

I appreciate the work you have done so far.  If the explanations that
others have posted convince you to continue, would you please tell me?

Otherwise, I guess we have no one maintaining Emacs on the Mac for the
moment.  If someone else is interested in picking up the
maintainership, please let me know.

I am pretty sure I asked you at some point if you would please work on
further MacOS issues.  When someone recorded your name in MAINTAINERS,
it was probably a reflection of the fact he saw you doing the job.

Re: Do we need legal papers

[Andrew Choi]

Thank you for asking.  But of course I have decided quite definitely to 
quit before I wrote to resign.  I hope your next maintainer will serve 
the purpose of your project much better than I can.

On 5-Nov-03, at 4:32 PM, Richard Stallman wrote:

> I appreciate the work you have done so far.  If the explanations that
> others have posted convince you to continue, would you please tell me?
>
> Otherwise, I guess we have no one maintaining Emacs on the Mac for the
> moment.  If someone else is interested in picking up the
> maintainership, please let me know.
>
> I am pretty sure I asked you at some point if you would please work on
> further MacOS issues.  When someone recorded your name in MAINTAINERS,
> it was probably a reflection of the fact he saw you doing the job.

Re: Do we need legal papers

[Sébastien Kirche]

Le jeudi, 6 nov 2003, à 17:24 Europe/Paris, Andrew Choi a écrit :

> Thank you for asking.  But of course I have decided quite definitely 
> to quit before I wrote to resign.  I hope your next maintainer will 
> serve the purpose of your project much better than I can.

Anyway, if you have planned to stop being official maintainer, may we 
(i mean : other emacs osx users/developpers) count on you as kind of 
"consultant" ?

I am quite sure that there is many people that currently use or will 
try osx native emacs, but i don't remember to have see many regular osx 
developpers on emacs-devel, other than you and Kim Storm.
I am actually a developper but only recently on mac platform. I think i 
may one day contribute to Emacs development, and there is surely other 
developpers that will, but we need confirmed guys on osx like you...

Sébastien Kirche

-- 
E pluribus UNIX

Re: Do we need legal papers

[Kim F. Storm]

Sébastien Kirche <sebastien.kirche@sage.com> writes:

>                but i don't remember to have see many regular
> osx developpers on emacs-devel, other than you and Kim Storm.

Just for the record, I only run emacs under X on GNU/Linux systems,
and my main interest is developing emacs for use on free platforms.  

I do make mac and windows related changes from time to time, but only
if such changes are mandated by similar changes in the X support.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Do we need legal papers [was: patch for unexmacosx.c: supporting non-prebound dynamic library]

[Richard Stallman]

    I looked at the patch, and it seems to be too big (it adds 60+ new
    source lines) to install without legal papers, and I cannot see that
    Nozumu Ando has signed papers.

    I think you should check with RMS if we need papers for this.

If it is 60 lines long, we definitely need legal papers for it.

Andrew, you had better delete it now, while we wait to get papers for
it.  Would you please ack when it is deleted?

When we get Nozomu's papers, I will inform people.