Master broken by commit b76389f

Master broken by commit b76389f

[Eli Zaretskii]

In a 32-bit MinGW build --with-wide-int, I cannot compile today's master:

    CC       dispnew.o
  In file included from lisp.h:35:0,
		   from dispnew.c:27:
  ../lib/verify.h:207:21: error: static assertion failed: "verify (alignof (union vectorlike_header) % GCALIGNMENT == 0)"
   # define _GL_VERIFY _Static_assert
		       ^
  ../lib/verify.h:252:20: note: in expansion of macro '_GL_VERIFY'
   # define verify(R) _GL_VERIFY (R, "verify (" #R ")")
		      ^
  lisp.h:901:1: note: in expansion of macro 'verify'
   verify (alignof (union vectorlike_header) % GCALIGNMENT == 0);
   ^
  ../lib/verify.h:207:21: error: static assertion failed: "verify (alignof (struct Lisp_String) % GCALIGNMENT == 0)"
   # define _GL_VERIFY _Static_assert
		       ^
  ../lib/verify.h:252:20: note: in expansion of macro '_GL_VERIFY'
   # define verify(R) _GL_VERIFY (R, "verify (" #R ")")
		      ^
  lisp.h:1384:1: note: in expansion of macro 'verify'
   verify (alignof (struct Lisp_String) % GCALIGNMENT == 0);
   ^
  Makefile:382: recipe for target `dispnew.o' failed

and many similar problems in other C source files.

Re: Master broken by commit b76389f

[Eli Zaretskii]

> Date: Fri, 08 Jun 2018 12:49:12 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: emacs-devel@gnu.org
> 
> In a 32-bit MinGW build --with-wide-int, I cannot compile today's master:

In case it matters, this is with MinGW GCC 5.3.0.

Re: Master broken by commit b76389f

[Paul Eggert]

[-- Attachment #1: Type: text/plain, Size: 347 bytes --]

Thanks for checking that; I fixed it (at least on Fedora i686 --with-wide-int) 
by installing the attached patch into master. I'll check the same platform on 
the Bug#31750 proposed patches before installing them, as these patches are 
doing related (but more-significant) changes to stack and heap allocation in the 
Emacs Lisp interpreter.

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Port-alignment-verification-to-x86-with-wide-int.patch --]
[-- Type: text/x-patch; name="0001-Port-alignment-verification-to-x86-with-wide-int.patch", Size: 1892 bytes --]

From db353b8649cdae54146308e4c875e53d02b0aaee Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Fri, 8 Jun 2018 08:08:03 -0700
Subject: [PATCH] Port alignment verification to x86 --with-wide-int

Problem reported by Eli Zaretskii in:
https://lists.gnu.org/r/emacs-devel/2018-06/msg00238.html
* src/lisp.h (struct Lisp_Symbol, union vectorlike_header)
(struct Lisp_Cons, struct Lisp_String):
Do not check alignment if !USE_LSB_TAG, as alignment is
needed only if we are tagging the low-order bits.
---
 src/lisp.h | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/src/lisp.h b/src/lisp.h
index 10012b2..293cf27 100644
--- a/src/lisp.h
+++ b/src/lisp.h
@@ -785,7 +785,7 @@ struct Lisp_Symbol
     GCALIGNED_UNION
   } u;
 };
-verify (alignof (struct Lisp_Symbol) % GCALIGNMENT == 0);
+verify (!USE_LSB_TAG || alignof (struct Lisp_Symbol) % GCALIGNMENT == 0);
 
 /* Declare a Lisp-callable function.  The MAXARGS parameter has the same
    meaning as in the DEFUN macro, and is used to construct a prototype.  */
@@ -898,7 +898,7 @@ union vectorlike_header
     ptrdiff_t size;
     GCALIGNED_UNION
   };
-verify (alignof (union vectorlike_header) % GCALIGNMENT == 0);
+verify (!USE_LSB_TAG || alignof (union vectorlike_header) % GCALIGNMENT == 0);
 
 INLINE bool
 (SYMBOLP) (Lisp_Object x)
@@ -1259,7 +1259,7 @@ struct Lisp_Cons
     GCALIGNED_UNION
   } u;
 };
-verify (alignof (struct Lisp_Cons) % GCALIGNMENT == 0);
+verify (!USE_LSB_TAG || alignof (struct Lisp_Cons) % GCALIGNMENT == 0);
 
 INLINE bool
 (NILP) (Lisp_Object x)
@@ -1381,7 +1381,7 @@ struct Lisp_String
     GCALIGNED_UNION
   } u;
 };
-verify (alignof (struct Lisp_String) % GCALIGNMENT == 0);
+verify (!USE_LSB_TAG || alignof (struct Lisp_String) % GCALIGNMENT == 0);
 
 INLINE bool
 STRINGP (Lisp_Object x)
-- 
2.7.4

Re: Master broken by commit b76389f

[Eli Zaretskii]

> Cc: emacs-devel@gnu.org
> From: Paul Eggert <eggert@cs.ucla.edu>
> Date: Fri, 8 Jun 2018 08:23:19 -0700
> 
> Thanks for checking that; I fixed it (at least on Fedora i686 --with-wide-int) 
> by installing the attached patch into master. I'll check the same platform on 
> the Bug#31750 proposed patches before installing them, as these patches are 
> doing related (but more-significant) changes to stack and heap allocation in the 
> Emacs Lisp interpreter.

Thanks, it builds cleanly here now.