T0b1/wasmtime
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Cranelift: Redundant stack-slot-to-stack-slot copy removal. PR#773.

Browse Source 
This is also https://bugzilla.mozilla.org/show_bug.cgi?id=1552737.

Cranelift currently has a tendency to create redundant copies (self-copies) of
values from a stack slot back to the same stack slot.  This generates a
pointless load and store and an unnecessary register use.  The copies are
created by `visit_inst` in regalloc/reload.rs.  They appear to occur mostly,
but not exclusively, at loop heads.  It's unclear why this happens.

This patch adds a special case to `visit_inst` to find such copies.  They are
converted into a new instruction, `copy_nop`, which takes and produces the
same SSA names, so as not to break any of the SSA invariants, but which has a
zero-length encoding, hence removing the copy at emission time.

`copy_nop`s source and destination operands must be stack slots and of course
the *same* stack slot.  The verifier has been enhanced to check this, since
misuse of `copy_nop` will likely lead to hard-to-find incorrect-code bugs.

Attempts were made to write a standalone .clif test case.  But these failed
because it appears the .clif parser accepts but ignores location hints that
are stack slots.  So it's impossible to write, in clif, the exact form of
`copy` instruction that triggers the transformation.
This commit is contained in:
Julian Seward
2019-05-27 11:55:23 +02:00
committed by Benjamin Bouvier
parent 6935033c9e
commit 03368895fe
7 changed files with 116 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
cranelift/codegen/meta-python/isa/x86/encodings.py
View File

@@ -341,6 +341,13 @@ enc_x86_64(x86.pop.i64, r.popq, 0x58)
X86_64.enc(base.copy_special, *r.copysp.rex(0x89, w=1))
X86_32.enc(base.copy_special, *r.copysp(0x89))
# Stack-slot-to-the-same-stack-slot copy, which is guaranteed to turn
# into a no-op.
X86_64.enc(base.copy_nop.i64, r.stacknull, 0)
X86_64.enc(base.copy_nop.i32, r.stacknull, 0)
X86_64.enc(base.copy_nop.f64, r.stacknull, 0)
X86_64.enc(base.copy_nop.f32, r.stacknull, 0)
# Adjust SP down by a dynamic value (or up, with a negative operand).
X86_32.enc(base.adjust_sp_down.i32, *r.adjustsp(0x29))
X86_64.enc(base.adjust_sp_down.i64, *r.adjustsp.rex(0x29, w=1))