[PATCH] m68k: Require m68020up rather than m68000up for CHK.L instruction

[PATCH] m68k: Require m68020up rather than m68000up for CHK.L instruction

[Fredrik Noring]

The long variant of CHK is MC68020, MC68030, MC68040 only, according
to "Motorola M68000 family programmer's reference manual", M68000PM/AD
rev. 1, 1992, p. 4-69. It's available with the MC68060 as well according
to "M68060 User's Manual", 1994, p. 10-23.

diff --git a/opcodes/m68k-opc.c b/opcodes/m68k-opc.c
index db198941f07..37df5031966 100644
--- a/opcodes/m68k-opc.c
+++ b/opcodes/m68k-opc.c
@@ -235,7 +235,7 @@ const struct m68k_opcode m68k_opcodes[] =
 {"chk2w", 4, 	two(0001300,0004000),	two(0177700,07777), "!sR1", m68020up | cpu32 | fido_a },
 {"chk2l", 4, 	two(0002300,0004000),	two(0177700,07777), "!sR1", m68020up | cpu32 | fido_a },
 
-{"chkl", 2,	one(0040400),		one(0170700), ";lDd", m68000up },
+{"chkl", 2,	one(0040400),		one(0170700), ";lDd", m68020up },
 {"chkw", 2,	one(0040600),		one(0170700), ";wDd", m68000up },
 
 #define SCOPE_LINE (0x1 << 3)

Re: [PATCH] m68k: Require m68020up rather than m68000up for CHK.L instruction

[Nick Clifton]

Hi Fredrik,

> The long variant of CHK is MC68020, MC68030, MC68040 only, according
> to "Motorola M68000 family programmer's reference manual", M68000PM/AD
> rev. 1, 1992, p. 4-69. It's available with the MC68060 as well according
> to "M68060 User's Manual", 1994, p. 10-23.

Thanks for reporting this problem.  I have applied the patch you provided.

Cheers
   Nick

Re: [PATCH] m68k: Require m68020up rather than m68000up for CHK.L instruction

[Fredrik Noring]

Thanks Nick,

On Thu, Jan 07, 2021 at 02:45:57PM +0000, Nick Clifton wrote:
> Thanks for reporting this problem.  I have applied the patch you provided.

Is anyone interested in general improvements? I had some ideas for a
project, for example separating syntax from semantics, but I ended up
rewriting the disassembler essentially from scratch (restricted to the
plain 68000, admittedly, but still).

It could also be made reentrant, thread-safe, faster, and so on.

[ As it turns out, the m68k binutils disassembler and assembler are
approximate, in the sense that a reassembled disassembly will not
necessarily be exactly the same as the original. I had to fix that too. ]

Fredrik

Re: [PATCH] m68k: Require m68020up rather than m68000up for CHK.L instruction

[Nick Clifton]

Hi Fredrik,

> Is anyone interested in general improvements?

Yes and no. :-)

Yes - we are always interested in any improvements to the binutils
source code, no matter the architecture targetted.

But no in the sense that I do not know of anyone else who is using the
m68k architecture, so you might be on your own.  However please do not
let this stop you...

Cheers
   Nick

Re: [PATCH] m68k: Require m68020up rather than m68000up for CHK.L instruction

[Jeff Law]

On 1/7/21 9:39 AM, Nick Clifton via Binutils wrote:
> Hi Fredrik,
>
>> Is anyone interested in general improvements?
>
> Yes and no. :-)
>
> Yes - we are always interested in any improvements to the binutils
> source code, no matter the architecture targetted.
Yup.

>
> But no in the sense that I do not know of anyone else who is using the
> m68k architecture, so you might be on your own.  However please do not
> let this stop you...
The debian guys still build a distro for m68k and they have a small, but
vocal user base.  That user base even came up with ~$5k for some key
updates to the GCC m68k port to keep it from being removed (our old
friend Bernd S. did the work and took home the $5k).

It bootstraps weekly in the upstream GCC tester (using qemu
emulation).   So it's generally working.


jeff