This is the mail archive of the binutils@sourceware.org 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]
Other format: [Raw text]

h8300 and mn10200 wrong code


I'm not sure what our policy is with regards to xfails in the
testsuite, but I'm inclined to think we should only xfail tests where
the target doesn't support the feature being tested, or where some
target peculiarity makes the test fail but the target does in fact
correctly support the feature under test.

For the testcase below, which is testing a simple difference of two
forward referenced labels, both h8300 and mn10200 will silently
generate wrong code.  I don't care enough about either target to fix
them, but I also think we shouldn't hide this type of failure.

	* gas/all/gas.exp (do_930509a): Don't xfail h8300 and mn10200.

Index: gas/testsuite/gas/all/gas.exp
===================================================================
RCS file: /cvs/src/src/gas/testsuite/gas/all/gas.exp,v
retrieving revision 1.70
diff -u -p -r1.70 gas.exp
--- gas/testsuite/gas/all/gas.exp	6 Apr 2011 17:09:51 -0000	1.70
+++ gas/testsuite/gas/all/gas.exp	12 Apr 2011 00:08:01 -0000
@@ -243,7 +243,8 @@ if { ![istarget hppa*-*-*] &&
      ![istarget *c54x*-*-*] } then {
     # the vax fails because VMS can apparently actually handle this
     # case in relocs, so gas doesn't handle it itself.
-    setup_xfail "h8300*-*-elf*" "mn10200*-*-*" "mn10300*-*-*" "vax*-*-vms*"
+    # mn10300 emits two relocs to handle the difference of two symbols.
+    setup_xfail "mn10300*-*-*" "vax*-*-vms*"
     do_930509a
 }
 

-- 
Alan Modra
Australia Development Lab, IBM


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