unofficial mirror of bug-gnu-emacs@gnu.org 
 help / color / mirror / code / Atom feed
* Bug in Win32 version of emacs on x64 machines
@ 2007-07-17 18:29 Eric Youngdale
  2007-07-18 22:49 ` Jason Rumney
  0 siblings, 1 reply; 3+ messages in thread
From: Eric Youngdale @ 2007-07-17 18:29 UTC (permalink / raw)
  To: bug-gnu-emacs

If you are running the 32-bit version of emacs on a 64-bit machine with
a 64-bit version of windows, and the SHELL environment variable points
to a 64-bit executable, then M-x shell will cause emacs to crash.

The reason for this is that in w32proc.c, it starts checking the
executable to be loaded to see if it is a cygwin executable, but it
doesn't bother to check to see if it is a 64-bit executable or a 32-bit
executable and it assumes that the it can dereference structures as if
it were a 32-bit executable.  This causes emacs to crash.   The
quick-and-dirty fix is enclosed - basically restrict the test and see if
the executable type is x86 before doing the cygwin check.

This fix does work with VC6 - don't know about anything older, nor do I
know if anyone cares in this age.

Anyways, this fix seems to fix the problem I described above.  I am also
playing with an x64 port of emacs - I just got it running - need to test
and resolve a few other potential problems before I worry about
submitting diffs....

*** w32proc.c.~1~ Sat Jan 20 23:18:15 2007
--- w32proc.c Tue Jul 17 13:46:42 2007
***************
*** 63,68 ****
--- 63,72 ----
  #include "syssignal.h"
  #include "w32term.h"
  
+ #ifndef IMAGE_FILE_MACHINE_AMD64
+ #define IMAGE_FILE_MACHINE_AMD64             0x8664  // AMD64 (K8)
+ #endif
+ 
  #define RVA_TO_PTR(var,section,filedata) \
    ((void *)((section)->PointerToRawData
\
  	    + ((DWORD)(var) - (section)->VirtualAddress)
\
***************
*** 630,636 ****
  	 executables use the OS/2 1.x format. */
  
        IMAGE_DOS_HEADER * dos_header;
!       IMAGE_NT_HEADERS * nt_header;
  
        dos_header = (PIMAGE_DOS_HEADER) executable.file_base;
        if (dos_header->e_magic != IMAGE_DOS_SIGNATURE)
--- 634,641 ----
  	 executables use the OS/2 1.x format. */
  
        IMAGE_DOS_HEADER * dos_header;
!       IMAGE_NT_HEADERS32 * nt_header;
!       IMAGE_NT_HEADERS64 * nt_header64;
  
        dos_header = (PIMAGE_DOS_HEADER) executable.file_base;
        if (dos_header->e_magic != IMAGE_DOS_SIGNATURE)
***************
*** 637,642 ****
--- 642,648 ----
  	goto unwind;
  
        nt_header = (PIMAGE_NT_HEADERS) ((char *) dos_header +
dos_header->e_lfanew);
+       nt_header64 = (PIMAGE_NT_HEADERS64) ((char *) dos_header +
dos_header->e_lfanew);
  
        if ((char *) nt_header > (char *) dos_header + executable.size)
  	{
***************
*** 650,678 ****
    	}
        else if (nt_header->Signature == IMAGE_NT_SIGNATURE)
    	{
! 	  /* Look for cygwin.dll in DLL import list. */
! 	  IMAGE_DATA_DIRECTORY import_dir =
!
nt_header->OptionalHeader.DataDirectory[IMAGE_DIRECTORY_ENTRY_IMPORT];
! 	  IMAGE_IMPORT_DESCRIPTOR * imports;
! 	  IMAGE_SECTION_HEADER * section;
! 
! 	  section = rva_to_section (import_dir.VirtualAddress,
nt_header);
! 	  imports = RVA_TO_PTR (import_dir.VirtualAddress, section,
executable);
! 
! 	  for ( ; imports->Name; imports++)
!   	    {
! 	      char * dllname = RVA_TO_PTR (imports->Name, section,
executable);
! 
! 	      /* The exact name of the cygwin dll has changed with
! 	         various releases, but hopefully this will be reasonably
! 	         future proof.  */
! 	      if (strncmp (dllname, "cygwin", 6) == 0)
  		{
! 		  *is_cygnus_app = TRUE;
! 		  break;
  		}
!   	    }
! 
  	  /* Check whether app is marked as a console or windowed (aka
               GUI) app.  Accept Posix and OS2 subsytem apps as console
               apps.  */
--- 656,694 ----
    	}
        else if (nt_header->Signature == IMAGE_NT_SIGNATURE)
    	{
! 	  /* cygwin is 32-bit only right now.  The code to check a
64-bit binary
! 	   is slightly different. */
! 	  if( nt_header->FileHeader.Machine == IMAGE_FILE_MACHINE_I386 )
! 	    {
! 	      /* Look for cygwin.dll in DLL import list. */
! 	      IMAGE_DATA_DIRECTORY import_dir =
!
nt_header->OptionalHeader.DataDirectory[IMAGE_DIRECTORY_ENTRY_IMPORT];
! 	      IMAGE_IMPORT_DESCRIPTOR * imports;
! 	      IMAGE_SECTION_HEADER * section;
! 	      
! 	      section = rva_to_section (import_dir.VirtualAddress,
nt_header);
! 	      imports = RVA_TO_PTR (import_dir.VirtualAddress, section,
executable);
! 	      
! 	      for ( ; imports->Name; imports++)
  		{
! 		  char * dllname = RVA_TO_PTR (imports->Name, section,
executable);
! 		  
! 		  /* The exact name of the cygwin dll has changed with
! 		     various releases, but hopefully this will be
reasonably
! 		     future proof.  */
! 		  if (strncmp (dllname, "cygwin", 6) == 0)
! 		    {
! 		      *is_cygnus_app = TRUE;
! 		      break;
! 		    }
  		}
! 	    }
! 	  else if( nt_header64->FileHeader.Machine ==
IMAGE_FILE_MACHINE_AMD64 )
! 	    {
! 	      /* Ultimately if 64-bit cygwin ever becomes a reality, a
different
! 		 set of checks need to be added here. For now, the stub
is present
! 		 to allow for that possibility. */
! 	    }
  	  /* Check whether app is marked as a console or windowed (aka
               GUI) app.  Accept Posix and OS2 subsytem apps as console
               apps.  */

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: Bug in Win32 version of emacs on x64 machines
  2007-07-17 18:29 Bug in Win32 version of emacs on x64 machines Eric Youngdale
@ 2007-07-18 22:49 ` Jason Rumney
  2007-07-19 21:20   ` Bug in Lose32 " Richard Stallman
  0 siblings, 1 reply; 3+ messages in thread
From: Jason Rumney @ 2007-07-18 22:49 UTC (permalink / raw)
  To: Eric Youngdale

Eric Youngdale wrote:
> If you are running the 32-bit version of emacs on a 64-bit machine with
> a 64-bit version of windows, and the SHELL environment variable points
> to a 64-bit executable, then M-x shell will cause emacs to crash.
>   
Thanks for the report and the patch.

I checked in a slightly different patch, as I noticed that the same
structure dereferencing problem exists with the is_gui check, though in
that case it won't cause a crash, just get it wrong.

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: Bug in Lose32 version of emacs on x64 machines
  2007-07-18 22:49 ` Jason Rumney
@ 2007-07-19 21:20   ` Richard Stallman
  0 siblings, 0 replies; 3+ messages in thread
From: Richard Stallman @ 2007-07-19 21:20 UTC (permalink / raw)
  To: Jason Rumney, Eric.Youngdale; +Cc: bug-gnu-emacs

Thanks for fixing this problem.  Meanwhile, I have fixed a bug in the
message subject.  In hacker terminology, calling something a "win" is
a form of praise.  We don't want to praise Microsoft Windows here.

Using Microsoft Windows is a problem -- you really need to defenestrate
your computers.  Are you working towards that?

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2007-07-19 21:20 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-07-17 18:29 Bug in Win32 version of emacs on x64 machines Eric Youngdale
2007-07-18 22:49 ` Jason Rumney
2007-07-19 21:20   ` Bug in Lose32 " Richard Stallman

Code repositories for project(s) associated with this public inbox

	https://git.savannah.gnu.org/cgit/emacs.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).