|
| From: | Richard Henderson |
| Subject: | Re: [PATCH v4 22/31] target/ppc: Implement PNOP |
| Date: | Thu, 13 May 2021 05:37:22 -0500 |
| User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.8.1 |
On 5/12/21 1:54 PM, matheus.ferst@eldorado.org.br wrote:
+### Prefixed No-operation Instruction
+
+&PNOP invalid_suffix:bool
+@PNOP 000001 11 0000-- 000000000000000000 \
+ ................................ &PNOP
+
+{
+ ## Invalid suffixes: Branch instruction
+ # bc[l][a]
+ PNOP ................................ \
+ 010000-------------------------- @PNOP invalid_suffix=1
For other cpus it has often turned out to be helpful to have a trans_INVALID or UNDEF or RESERVED or suchlike to use for cases like this. That way you don't need a special argument set, nor to set a flag as you do for each of these.
Also, the invalid suffixes themselves do not overlap, so you can get a slightly better decode via nested [], like so:
{
[
INVALID ...
INVALID ...
...
]
NOP ...
}
r~
| [Prev in Thread] | Current Thread | [Next in Thread] |