Toolchain/binutils-gdb
1
0
Fork 0
forked from Imagelibrary/binutils-gdb
Code Pull Requests Activity

aarch64; Add support for vector offset ranges

Browse Source 
Some SME2 instructions operate on a range of consecutive ZA vectors.
This is indicated by syntax such as:

   za[<Wv>, <imml>:<immh>]

Like with the earlier vgx2 and vgx4 support, we get better error
messages if the parser allows all ZA indices to have a range.
We can then reject invalid cases during constraint checking.
This commit is contained in:
Richard Sandiford
2023-03-30 11:09:11 +01:00
parent e2dc4040f3
commit 586c62819f
13 changed files with 151 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
include/opcode/aarch64.h
View File

@@ -1114,14 +1114,29 @@ const aarch64_cond* get_inverted_cond (const aarch64_cond *cond);
/* Information about a reference to part of ZA. */
struct aarch64_indexed_za
{
int regno; /* <ZAn> */
/* Which tile is being accessed. Unused (and 0) for an index into ZA. */
int regno;
struct
{
int regno; /* <Wv> */
int64_t imm; /* <imm> */
/* The 32-bit index register. */
int regno;
/* The first (or only) immediate offset. */
int64_t imm;
/* The last immediate offset minus the first immediate offset.
Unlike the range size, this is guaranteed not to overflow
when the end offset > the start offset. */
uint64_t countm1;
} index;
/* The vector group size, or 0 if none. */
unsigned group_size : 8;
unsigned v : 1; /* <HV> horizontal or vertical vector indicator. */
/* True if a tile access is vertical, false if it is horizontal.
Unused (and 0) for an index into ZA. */
unsigned v : 1;
};
/* Information about a list of registers. */