[zeynallow]

Profiling json_validate() on string-heavy input showed ~63% of the time in php_json_scan(): the first string pass (STR_P1) matches every byte as its own re2c token. json_validate() taking almost as long as json_decode() confirms the scan, not value construction, dominates.

On the encoding side, json_encode($v, JSON_UNESCAPED_UNICODE) of non-ASCII text was ~3x slower than of ASCII text: for every non-ASCII character php_json_escape_string() flushes the pending run, decodes the character with php_next_utf8_char() and appends it on its own.

Changes:

• Scanner: before (re)entering STR_P1, skip the run of printable ASCII bytes other than " and \, 16 bytes at a time with SSE2 (baseline on x86_64) and bytewise otherwise. The re2c rules are unchanged; they are just entered less often.
• Encoder: skip runs of bytes that are always copied verbatim (the same set the existing charmap lets through), and with JSON_UNESCAPED_UNICODE keep well-formed UTF-8 sequences in the pending run instead of copying them one by one. Validation matches php_next_utf8_char() (no overlongs, surrogates or code points above U+10FFFF), and U+2028/U+2029 are only kept when JSON_UNESCAPED_LINE_TERMINATORS is set. Everything else falls through to the existing code, so output and errors are unchanged.

Results

Release build (-O2), x86_64, median of 5 interleaved runs, 20 000-element payloads:

payload	operation	before	after	speedup
ASCII text (11 MB)	json_decode	27.07 ms	10.39 ms	2.60x
	json_validate	24.94 ms	8.62 ms	2.90x
	json_encode	18.91 ms	11.96 ms	1.58x
UTF-8 text (11 MB)	json_decode	29.47 ms	19.33 ms	1.52x
	json_encode	67.08 ms	59.20 ms	1.13x
	json_encode + JSON_UNESCAPED_UNICODE	62.20 ms	30.36 ms	2.05x
strings with escapes (4 MB)	json_decode	17.07 ms	13.86 ms	1.23x
	json_encode	15.34 ms	10.62 ms	1.44x
short mixed values (3.7 MB)	json_decode	18.87 ms	17.92 ms	1.05x
numbers only (1.7 MB)	json_decode	16.70 ms	16.84 ms	0.99x (noise)

Without SSE2 (scalar fallback only, e.g. aarch64) the gains are smaller: ASCII decode 1.78x, UTF-8 encode with JSON_UNESCAPED_UNICODE 1.71x, but decoding escape-heavy strings is ~5% slower. A NEON variant of the skip loop could follow if wanted.

Testing

• New ext/json/tests/string_scan_boundaries.phpt puts each special byte (control characters, invalid UTF-8, quotes, escapes, U+2028, ...) at every position of strings of length 0–48, across 16-byte chunk boundaries, for decode, validate and encode with several flag sets. It passes before and after this change, and caught an off-by-one in an earlier version of the SIMD mask.
• Differential fuzzing against an unpatched binary: 1 000 000 json_decode() calls (4 flag sets) plus json_validate() on random inputs, and 1 440 000 json_encode() calls (12 flag sets), all identical including errors. Also with the SIMD paths disabled.
• ext/json tests pass on release and debug builds, with and without the SIMD paths; no valgrind errors.