ARM64: Optimize pair of "str wzr, [reg]" to "str xzr"

[kunalspathak]

str     wzr, [x1,#8]
str     wzr, [x1,#12]

can be combined into str xzr if the stores are happening in subsequent memory.

str     xzr, [x1,#8]

I collected no. of such str pairs in framework libraries and found approx. 450 pairs in 353 methods.

Details:

str_str_wzr_to_str_xzr.txt

category:cq
theme:optimization
skill-level:intermediate
cost:small
impact:medium

[Dotnet-GitSync-Bot]

I couldn't figure out the best area label to add to this issue. Please help me learn by adding exactly one area label.

[BruceForstall]

In addition, a sequence of 4 str wzr could be collapsed down to a single stp xzr, xzr by further optimizing using #35134.

[BruceForstall]

In the attached file, there are some interesting example that are more complicated than simple adjacency peephole optimization. E.g.,

; Assembly listing for method System.Data.ExpressionParser:StartScan():this
[0024]        B9003C1F          str     wzr, [x0,#60]
[0025]        B900301F          str     wzr, [x0,#48]
[0026]        B900341F          str     wzr, [x0,#52]
[0027]        B900401F          str     wzr, [x0,#64]

Could be:

str     xzr, [x0,#48]
str     xzr, [x0,#60]

but it require recognizing that the non-contiguous offsets 60 and 64 can be combined.