This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: gas testsuite approch
- From: Hans-Peter Nilsson <hp at bitrange dot com>
- To: "Svein E. Seldal" <Svein dot Seldal at solidas dot com>
- Cc: binutils <binutils at sources dot redhat dot com>
- Date: Sun, 15 Sep 2002 19:47:13 -0400 (EDT)
- Subject: Re: gas testsuite approch
On Mon, 16 Sep 2002, Svein E. Seldal wrote:
> Take the all/gas.exp test, the "x930509.s" sub-test for example. This
> test is rather simple and the objective is to calculate the difference
> of two (forward) addresses. On a byte-based target, like i386, the
> result of this subtraction will be 4.
>
> However, on a target like the tic4x, the CPU is unaware of bytes
> (sizeof(char)=sizeof(long)=1=32-bits), and will result in the value of 1
> which will make the test fail because the test expects 4.
>
> What will be the best approach to fix this problem?
You don't mention the "prior art" xfails in
gas/testsuite/gas/all/gas.exp so perhaps you haven't looked
there. It seems the sentence "C30 counts a four byte offset as
a difference of one" matches your case too. If you xfail that
test and add it but modified for your target to your
target-specific test-suite, I think nothing would be lost
test-wise. If you don't like that approach, I think you're
facing larger changes.
brgds, H-P