This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: [PATCH] Typo in tc-arm.c
- From: Richard Earnshaw <rearnsha at arm dot com>
- To: Mark Mitchell <mark at codesourcery dot com>
- Cc: Paul Brook <paul at codesourcery dot com>, Richard Earnshaw <rearnsha at arm dot com>, binutils at sources dot redhat dot com
- Date: Tue, 09 Dec 2003 13:49:27 +0000
- Subject: Re: [PATCH] Typo in tc-arm.c
- Organization: ARM Ltd.
- Reply-to: Richard dot Earnshaw at arm dot com
> On Mon, 2003-12-08 at 04:54, Paul Brook wrote:
> > Patch below fixes typo in do_umaal.
> >
> > Looks obviously correct, but I need someone else to apply.
>
> Paul --
>
> Two thoughts:
>
> (a) you should probably request CVS write-after-approval access to
> binutils,
>
I agree (FWIW).
> (b) since we clearly fail to have sufficiently many test cases for
> umaal, would you please add one?
>
Yep. The way I've been trying to add test cases is along the following
lines.
For each instruction there should be a test that checks the basic bit
pattern (by using r0 in every register position):
umaal r0, r0, r0, r0
A test for each register position, ideally with r15 (to check that the
correct bits are being set.
umaal r0, r0, r0, r15
umaal r0, r0, r15, r0
umaal r0, r15, r0, r0
umaal r15, r0, r0, r0
and a final test that checks the conditional operation (if any) and a
random set of registers.
umaaleq r1, r8, r7, r6
Now it is possible that some instructions have restrictions on what
registers can be used in which arguments (and don't allow duplicates). So
some of the rules have to be bent slightly. Registers r9, r11, and r13
are reasonable substitutes for r15 in most circumstances (top and bottom
bit set).
R.