public inbox for binutils@sourceware.org
 help / color / mirror / Atom feed
* Gas regressions for cris-elf
@ 2006-04-06  4:01 Hans-Peter Nilsson
  2006-04-06  4:36 ` Alan Modra
  0 siblings, 1 reply; 3+ messages in thread
From: Hans-Peter Nilsson @ 2006-04-06  4:01 UTC (permalink / raw)
  To: amodra; +Cc: binutils

(My autotester has been off for a couple of days, or this would
have been noticed earlier.)

Could it be that your change of:

2006-04-04  Alan Modra  <amodra@bigpond.net.au>

	PR 997
	* frags.c (frag_offset_fixed_p): New function.
	* frags.h (frag_offset_fixed_p): Declare.
	* expr.c (expr): Use frag_offset_fixed_p when simplifying subtraction.
	(resolve_expression): Likewise.

is responsible for:

...
Running /h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/cris.exp ...
FAIL: gas/cris/rd-dw2-2
FAIL: False broken words
...

(The broken-dot-word handling shouldn't be a victim: the key is
the word "False"; there are no ".word a-b" expressions there.)

In gas.log (same source, same messages for both failures; the
difference is whether or not using --gdwarf2):

/h/hp/binutils/cvs_latest/src/gas/testsuite/lib/run ../as-new  --em=criself --gdwarf2 -o dump.o /h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s: Assembler messages:
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:37: Error: Immediate value not in 8 bit range: -327683
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:37: Error: Immediate value not in 16 bit range: -327683
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:39: Error: Immediate value not in 8 bit range: -327683
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:39: Error: Immediate value not in 16 bit range: -327683
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:42: Error: Immediate value not in 16 bit range: -327683
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:44: Error: Immediate value not in 16 bit range: -327683
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:47: Error: Immediate value not in 16 bit range: -262233
/h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:49: Error: Immediate value not in 16 bit range: -262233

Nothing really strange about that code IMHO; it's just
exercising frags a bit, which at one time conspired with some
broken-dot-word flaw.

brgds, H-P

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

* Re: Gas regressions for cris-elf
  2006-04-06  4:01 Gas regressions for cris-elf Hans-Peter Nilsson
@ 2006-04-06  4:36 ` Alan Modra
  2006-04-06  9:49   ` Alan Modra
  0 siblings, 1 reply; 3+ messages in thread
From: Alan Modra @ 2006-04-06  4:36 UTC (permalink / raw)
  To: Hans-Peter Nilsson; +Cc: binutils

On Thu, Apr 06, 2006 at 05:48:56AM +0200, Hans-Peter Nilsson wrote:
> /h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:37: Error: Immediate value not in 8 bit range: -327683

Yes, this is probably my patch.

 .data
 .globl back_ref_data_zero
back_ref_data_zero:
 .space 42,0
 .globl back_ref_data_fortytwo
back_ref_data_fortytwo:

 .text
 add.b back_ref_data_fortytwo-back_ref_data_zero,r5

back_ref_data_fortytwo-back_ref_data_zero ought to be evaluated
immediately, because even though the syms are in separate frags they
have a fixed relationship.  I'm probably calculating .space
incorrectly.  Will investigate.

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre

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

* Re: Gas regressions for cris-elf
  2006-04-06  4:36 ` Alan Modra
@ 2006-04-06  9:49   ` Alan Modra
  0 siblings, 0 replies; 3+ messages in thread
From: Alan Modra @ 2006-04-06  9:49 UTC (permalink / raw)
  To: Hans-Peter Nilsson, binutils

On Thu, Apr 06, 2006 at 01:31:46PM +0930, Alan Modra wrote:
> On Thu, Apr 06, 2006 at 05:48:56AM +0200, Hans-Peter Nilsson wrote:
> > /h/hp/binutils/cvs_latest/src/gas/testsuite/gas/cris/binop-segref.s:37: Error: Immediate value not in 8 bit range: -327683
> 
> Yes, this is probably my patch.

	* frags.c (frag_offset_fixed_p): Reinitialise offset before
	second scan.

Index: gas/frags.c
===================================================================
RCS file: /cvs/src/src/gas/frags.c,v
retrieving revision 1.19
diff -u -p -r1.19 frags.c
--- gas/frags.c	4 Apr 2006 08:04:57 -0000	1.19
+++ gas/frags.c	6 Apr 2006 04:31:44 -0000
@@ -419,6 +419,7 @@ frag_offset_fixed_p (fragS *frag1, fragS
     }
 
   /* Maybe frag1 is after frag2.  */
+  off = frag1->fr_address - frag2->fr_address;
   frag = frag2;
   while (frag->fr_type == rs_fill)
     {

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre

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

end of thread, other threads:[~2006-04-06  4:36 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2006-04-06  4:01 Gas regressions for cris-elf Hans-Peter Nilsson
2006-04-06  4:36 ` Alan Modra
2006-04-06  9:49   ` Alan Modra

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).