[v4] | Patchew

[PATCH v4 41/48] tcg: Canonicalize alignment flags in MemOp

Richard Henderson posted 48 patches 4 years, 2 months ago

Diff against v1 v1 v1 v1 v1 v1 v2 v3 v4 v4 v7 v10 v12
Download mbox

Maintainers: Yoshinori Sato <ysato@users.sourceforge.jp>, Aurelien Jarno <aurelien@aurel32.net>, Aleksandar Rikalo <aleksandar.rikalo@syrmia.com>, Richard Henderson <richard.henderson@linaro.org>, Alistair Francis <Alistair.Francis@wdc.com>, Cornelia Huck <cohuck@redhat.com>, Laurent Vivier <laurent@vivier.eu>, Paolo Bonzini <pbonzini@redhat.com>, Jiaxun Yang <jiaxun.yang@flygoat.com>, "Philippe Mathieu-Daudé" <f4bug@amsat.org>, Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>, Stefan Weil <sw@weilnetz.de>, Greg Kurz <groug@kaod.org>, Eduardo Habkost <ehabkost@redhat.com>, David Gibson <david@gibson.dropbear.id.au>, Marcel Apfelbaum <marcel.apfelbaum@gmail.com>, "Edgar E. Iglesias" <edgar.iglesias@gmail.com>, Riku Voipio <riku.voipio@iki.fi>, "Alex Bennée" <alex.bennee@linaro.org>, Artyom Tarasenko <atar4qemu@gmail.com>, Palmer Dabbelt <palmer@dabbelt.com>, Taylor Simpson <tsimpson@quicinc.com>, Peter Maydell <peter.maydell@linaro.org>, David Hildenbrand <david@redhat.com>, Thomas Huth <thuth@redhat.com>

There is a newer version of this series

Expand all Fold all

[PATCH v4 41/48] tcg: Canonicalize alignment flags in MemOp

Posted by Richard Henderson 4 years, 2 months ago

Having observed e.g. al8+leq in dumps, canonicalize to al+leq.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/tcg-op.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index b1cfd36f29..61b492d89f 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -2765,7 +2765,12 @@ void tcg_gen_lookup_and_goto_ptr(void)
 static inline MemOp tcg_canonicalize_memop(MemOp op, bool is64, bool st)
 {
     /* Trigger the asserts within as early as possible.  */
-    (void)get_alignment_bits(op);
+    unsigned a_bits = get_alignment_bits(op);
+
+    /* Prefer MO_ALIGN+MO_XX over MO_ALIGN_XX+MO_XX */
+    if (a_bits == (op & MO_SIZE)) {
+        op = (op & ~MO_AMASK) | MO_ALIGN;
+    }
 
     switch (op & MO_SIZE) {
     case MO_8:
-- 
2.25.1