master: arm64/character-string-to-utf8: jump to ascii-loop from a better place

The branch "master" has been updated in SBCL:
       via  bd76fc4481386d6f4f295f7968e6c459e62fce99 (commit)
      from  0eb8b556ac414e9c3e4f5efaec69022af69537fc (commit)

- Log -----------------------------------------------------------------
commit bd76fc4481386d6f4f295f7968e6c459e62fce99
Author: Stas Boukarev <[email protected]>
Date:   Sat Aug 22 17:19:52 2026 +0300

    arm64/character-string-to-utf8: jump to ascii-loop from a better place
---
 src/code/arm64-simd.lisp | 27 ++++++++++++++-------------
 tests/utf-8.impure.lisp  |  3 ++-
 2 files changed, 16 insertions(+), 14 deletions(-)

diff --git a/src/code/arm64-simd.lisp b/src/code/arm64-simd.lisp
index 2ef9424f3..b86a50584 100644
--- a/src/code/arm64-simd.lisp
+++ b/src/code/arm64-simd.lisp
@@ -1721,9 +1721,11 @@
                    (inst ld1 (list bytes bytes2) (@ string) :4s)
                    ;; Stop if anything is 3-4 bytes in utf8
                    (inst orr temp bytes bytes2 :4s)
-                   1-2-LOOP-START
                    (inst umaxv temp temp :4s)
                    (inst umov tmp temp 0 :s)
+                   (inst cmp tmp 127)
+                   (inst b :ls 32-ASCII-START)
+                   1-2-LOOP-START
                    (inst cmp tmp #x800)
                    (inst b :ge FULL-START)
 
@@ -1746,8 +1748,6 @@
                    (inst umov tmp ascii 0 :h)
 
                    (inst ubfiz tmp2 tmp 4 8) ;; shift the low 8 bits left
-                   (inst cmp tmp2 #xFF0)
-                   (inst b :eq 32-ASCII-START)
 
                    ;; Construct
                    ;; (logior
@@ -1780,18 +1780,19 @@
                    32-ASCII-LOOP
                    (inst ld1 (list bytes bytes2) (@ string) :4s)
                    (inst orr temp bytes bytes2 :4s)
-                   (check-ascii temp temp2 1-2-LOOP-START :4s)
-
-                   ;; Narrow to 16 bits
-                   (inst uzp1 bytes bytes bytes2 :8h)
-
-                   ;; Track newlines
-                   (inst cmeq temp bytes newlines :8h)
-                   (inst cmeq f1 temp 0 :4s)
-                   (inst bif last-newlines indexes f1 :16b)
-                   (inst add indexes indexes increment :4s)
+                   (check-ascii temp temp2 1-2-LOOP-START :4s :gpr tmp)
 
                    32-ASCII-START
+
+                   ;; Narrow to 16 bits
+                   (inst uzp1 bytes bytes bytes2 :8h)
+
+                   ;; Track newlines
+                   (inst cmeq temp bytes newlines :8h)
+                   (inst cmeq f1 temp 0 :4s)
+                   (inst bif last-newlines indexes f1 :16b)
+                   (inst add indexes indexes increment :4s)
+
                    ;; Narrow to 8 bytes
                    (inst xtn bytes bytes :8b)
                    (inst str bytes (@ byte-array 8 :post-index) :d)
diff --git a/tests/utf-8.impure.lisp b/tests/utf-8.impure.lisp
index 0f1023f67..c55941dce 100644
--- a/tests/utf-8.impure.lisp
+++ b/tests/utf-8.impure.lisp
@@ -71,7 +71,8 @@
                         0
                         (random 4))
         while (< i (length string))
-        do (loop repeat (1+ (random (- (length string) i)))
+        do (loop repeat (1+ (random (min (- (length string) i)
+                                         80)))
                  for char = (case width
                               (0 (random 128))
                               (1 (+ 128 (random (- 2048 128))))

-----------------------------------------------------------------------


hooks/post-receive
-- 
SBCL