Re: patch for fixing illegal instruction (wldrb, wldrh, wstrb, wstrh)

["Bridge Wu" <mingqiao dot wu at gmail dot com>]

On 9/23/06, Joseph S. Myers <joseph@codesourcery.com> wrote:
> I can't tell from this description whether your new test is running,
> whether it is FAILing as it should without the code patch applied, and
> whether it is PASSing as it should with the code patch applied.  There are
> pre-existing FAILs because there are pre-existing bugs in binutils; that's
> why you need to compare the before-and-after results of "make -k check".
> By results, I mean the contents of the .sum files created by "make -k
> check".
>
> If someone is to review your patch, you'll need to send a complete patch
> again that includes both the code changes and the addition of the .d and
> .s files.

I mean it is FAILing with the code patch applied. The diff between old
and new sum file is as below.

@@ -1,4 +1,5 @@
./ld/ld.sum:FAIL: Thumb shared library with ARM entry points
./ld/ld.sum:FAIL: Mixed ARM/Thumb shared library
./ld/ld.sum:FAIL: Using Thumb lib by another lib
+./gas/testsuite/gas.sum:FAIL: Intel(r) Wireless MMX(tm) technology
instructions version 1
./gas/testsuite/gas.sum:FAIL: Neon optional register operands


I saw only one expect file (arm.exp) existed in gas/testsuite/gas/arm
folder which also contained my testcase. From arm.exp shown as below,
the new test should be run according .d file. I suspect my .d file is
not written well. Or something else wrong?

if {[istarget *arm*-*-*] || [istarget *xscale*-*-*]} {
   run_dump_tests [lsort [glob -nocomplain $srcdir/$subdir/*.d]]
}

Once I run the test case successfully, I will submit an update patch
with source changes, .s and .d test file. Thanks.

--
best regards,
-Bridge