Re: gas bug

[Alan Modra <amodra at bigpond dot net dot au>]

On Thu, Oct 04, 2001 at 08:30:17PM +0200, Jakob Eriksson wrote:
> 
> Found a bug in GAS!

Yep.

> This assembler was configured for a target of `m68k-coff'.

non-BFD assembler..

> jen@tazzy:/tmp/Accelerator$ m68k-coff-as --register-prefix-optional -al
> -m68332 -mno-68881 -I ./inc ./src/acc_csw.src -o acc_csw.obj --mri
> Segmentation fault

This will fix your particular problem.

	* subseg.c (subseg_text_p): Return 0 for absolute section.

And this patch will catch similar problems.

	* read.c (do_align): If in absolute section, warn about and
	ignore non-zero fill pattern.

gas shouldn't be trying to create frags in the absolute section.

-- 
Alan Modra

Index: gas/subsegs.c
===================================================================
RCS file: /cvs/src/src/gas/subsegs.c,v
retrieving revision 1.9
diff -u -p -r1.9 subsegs.c
--- subsegs.c	2001/08/01 01:44:25	1.9
+++ subsegs.c	2001/10/05 10:22:22
@@ -587,7 +587,7 @@ subseg_text_p (sec)
 #else /* ! BFD_ASSEMBLER */
   const char * const *p;
 
-  if (sec == data_section || sec == bss_section)
+  if (sec == data_section || sec == bss_section || sec == absolute_section)
     return 0;
 
   for (p = nontext_section_names; *p != NULL; ++p)
Index: gas/read.c
===================================================================
RCS file: /cvs/src/src/gas/read.c,v
retrieving revision 1.47
diff -u -p -r1.47 read.c
--- read.c	2001/09/19 05:33:19	1.47
+++ read.c	2001/10/05 10:22:22
@@ -1162,6 +1162,19 @@ do_align (n, fill, len, max)
      int len;
      int max;
 {
+  if (now_seg == absolute_section)
+    {
+      if (fill != NULL)
+	while (len-- > 0)
+	  if (*fill++ != '\0')
+	    {
+	      as_warn (_("ignoring fill value in absolute section"));
+	      break;
+	    }
+      fill = NULL;
+      len = 0;
+    }
+
 #ifdef md_do_align
   md_do_align (n, fill, len, max, just_record_alignment);
 #endif