This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

[patch] fix MIPS gas 'lineno' testsuite entry.


The patch below adjusts the MIPS gas 'lineno' testsuite entry so that
it can work on ECOFF as well as ELF.

The comments are, in my opinion, adequately explanatory.  (The
underlying cause is the same as with the 'sync' test that I just
fixed.)

As with the 'sync' patch, this was tested on mips-elf and mips-ecoff
with the current sources, and I verified that this test case failed
for mips-ecoff in just about every situation i could easily test:

	cvs co -r binu_ss_19990502 binutils
	cvs co -r binu_ss_19990602 binutils
	cvs co -r binu_ss_19990721 binutils
	2.10 release (first release with the tests)
	2.10.1 release


As noted previously, I don't have a working assignment on file.  This
patch may be unacceptable at this time for that reason ("too large").
If somebody could tell me if they think it's within the
size/complexity limit and/or otherwise OK it...  If it's not OK now
for paperwork reasons, I'll keep it my queue and resubmit it when the
paperwork's ironed out.



cgd
===================================================================
Apply changelog in gas/testsuite.

2001-03-26  Chris Demetriou  <cgd@broadcom.com>

	* gas/mips/lineno.d: Allow for the possibility that
	the local labels won't be in the objdump output.  Match
	zeros in addresses using '+' for better compatibility with
	different targets.

Index: gas/mips/lineno.d
===================================================================
RCS file: /cvs/src/src/gas/testsuite/gas/mips/lineno.d,v
retrieving revision 1.1.1.1
diff -c -r1.1.1.1 lineno.d
*** lineno.d	1999/05/03 07:28:51	1.1.1.1
--- lineno.d	2001/03/24 08:08:15
***************
*** 51,58 ****
  .*lineno.s:32
  .*58:.*b.*
  .*5c:.*nop
! 
! 0000000000000060 <\$L1>:
  .*lineno.s:34
  .*60:.*move.*
  .*lineno.s:35
--- 51,61 ----
  .*lineno.s:32
  .*58:.*b.*
  .*5c:.*nop
! # Objdump puts an '$L1' symbol here for ELF, but not for ECOFF.  For
! # ELF, $L1 is in the .text section so objdump picks it for this location.
! # For ECOFF, it's in the *DEBUG* section, so objdump prefers the .text
! # symbol over $L1.
! #...
  .*lineno.s:34
  .*60:.*move.*
  .*lineno.s:35
***************
*** 65,71 ****
  .*70:.*jr.*
  .*74:.*nop
  
! 0000000000000078 <g>:
  g\(\):
  .*lineno.s:47
  .*78:.*addiu.*
--- 68,74 ----
  .*70:.*jr.*
  .*74:.*nop
  
! 0+0078 <g>:
  g\(\):
  .*lineno.s:47
  .*78:.*addiu.*
***************
*** 84,91 ****
  .*lineno.s:54
  .*94:.*b.*
  .*98:.*nop
! 
! 000000000000009c <\$L2>:
  .*lineno.s:56
  .*9c:.*move.*
  .*lineno.s:57
--- 87,94 ----
  .*lineno.s:54
  .*94:.*b.*
  .*98:.*nop
! # Objdump puts an '$L2' symbol here for ELF, but not for ECOFF.
! #...
  .*lineno.s:56
  .*9c:.*move.*
  .*lineno.s:57


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]