Add Arm64 encoding for IF_SVE_BL_1A #97223
Conversation
ghost
added
community-contribution
|
Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch Issue DetailsThese instructions use a pattern to specify which elements to count from a predicate. This is encoded as 5 bits in the instruction. I've added a new entry in New clr test output (identical to capstone):
|
|
@kunalspathak @dotnet/arm64-contrib |
a74nh
force-pushed
the
IF_SVE_BL_1A_github
branch
from
17504dc to
1563f8c
Compare
TIHan
left a comment
TIHan
left a comment
There was a problem hiding this comment.
This looks good imo. I see why you have to introduce the new entry because you have to flow that info to the part of the actual encoding.
I'd wait for Kunal's review too before merging.
kunalspathak
left a comment
kunalspathak
left a comment
There was a problem hiding this comment.
A minor comment. Otherwise, LGTM
| regNumber _idReg4 : REGNUM_BITS; | ||
| }; | ||
|
|
||
| insSvePattern _idSvePattern; |
There was a problem hiding this comment.
this looks ok to me, but wondering why not just have something like this after the end of struct definition.
#ifdef TARGET_ARM64
insSvePattern _idSvePattern;
#endifThere was a problem hiding this comment.
Happy to switch to that.
My reasoning was when you look at the whole file we have an #ifdef for each target. Many of them just have the same thing (_idReg3 and _idReg4). The nesting of the Arm32/Arm64 is a little messy to read and doesn't fit the rest of the file style. Splitting it made it simpler.
3 participants
These instructions use a pattern to specify which elements to count from a predicate. This is encoded as 5 bits in the instruction. I've added a new entry in
instrdescunion (to prevent it increasing the size).Imm encoding is a 4bit number representing 1 to 16, so 1 needs subtracting when encoding.
New clr test output (identical to capstone):