[pypy-commit] pypy jit-dynamic-getarrayitem: optimize {get, set}interiorfield_{raw, gc} for itemsizes that can be addressed using special x86 addressing

[alex_gaynor]

Author: Alex Gaynor <alex.gaynor at gmail.com>
Branch: jit-dynamic-getarrayitem
Changeset: r49379:5545a3c93d51
Date: 2011-11-13 15:34 -0500
http://bitbucket.org/pypy/pypy/changeset/5545a3c93d51/

Log:	optimize {get,set}interiorfield_{raw,gc} for itemsizes that can be
	addressed using special x86 addressing

diff --git a/pypy/jit/backend/x86/assembler.py b/pypy/jit/backend/x86/assembler.py
--- a/pypy/jit/backend/x86/assembler.py
+++ b/pypy/jit/backend/x86/assembler.py
@@ -8,8 +8,8 @@
 from pypy.rpython.lltypesystem.lloperation import llop
 from pypy.rpython.annlowlevel import llhelper
 from pypy.jit.backend.model import CompiledLoopToken
-from pypy.jit.backend.x86.regalloc import (RegAlloc, get_ebp_ofs,
-                                           _get_scale, gpr_reg_mgr_cls)
+from pypy.jit.backend.x86.regalloc import (RegAlloc, get_ebp_ofs, _get_scale,
+    gpr_reg_mgr_cls, _valid_addressing_size)
 
 from pypy.jit.backend.x86.arch import (FRAME_FIXED_SIZE, FORCE_INDEX_OFS, WORD,
                                        IS_X86_32, IS_X86_64)
@@ -1601,8 +1601,10 @@
         assert isinstance(itemsize_loc, ImmedLoc)
         if isinstance(index_loc, ImmedLoc):
             temp_loc = imm(index_loc.value * itemsize_loc.value)
+        elif _valid_addressing_size(itemsize_loc.value):
+            return AddressLoc(base_loc, index_loc, _get_scale(itemsize_loc.value), ofs_loc.value)
         else:
-            # XXX should not use IMUL in most cases
+            # XXX should not use IMUL in more cases, it can use a clever LEA
             assert isinstance(temp_loc, RegLoc)
             assert isinstance(index_loc, RegLoc)
             assert not temp_loc.is_xmm
diff --git a/pypy/jit/backend/x86/regalloc.py b/pypy/jit/backend/x86/regalloc.py
--- a/pypy/jit/backend/x86/regalloc.py
+++ b/pypy/jit/backend/x86/regalloc.py
@@ -1434,8 +1434,11 @@
     # i.e. the n'th word beyond the fixed frame size.
     return -WORD * (FRAME_FIXED_SIZE + position)
 
+def _valid_addressing_size(size):
+    return size == 1 or size == 2 or size == 4 or size == 8
+
 def _get_scale(size):
-    assert size == 1 or size == 2 or size == 4 or size == 8
+    assert _valid_addressing_size(size)
     if size < 4:
         return size - 1         # 1, 2 => 0, 1
     else: