From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtpng3.m.smailru.net (smtpng3.m.smailru.net [94.100.177.149]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dev.tarantool.org (Postfix) with ESMTPS id 239E645C306 for ; Tue, 15 Dec 2020 17:23:02 +0300 (MSK) From: Sergey Nikiforov Date: Tue, 15 Dec 2020 17:22:46 +0300 Message-Id: In-Reply-To: References: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Subject: [Tarantool-patches] [PATCH v2 2/2] base64: Improve decoder performance List-Id: Tarantool development patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: tarantool-patches@dev.tarantool.org Cc: Vladislav Shpilevoy Unnecessary checks were removed from internal loops. Benchmark shows that performance is now ~1.15 times higher (release build, Intel Core I7-9700K, only one thread). --- Branch: https://github.com/tarantool/tarantool/tree/void234/gh-3069-fix-base64-memory-overrun-v2 third_party/base64.c | 36 +++++++++++++++++++++++++----------- 1 file changed, 25 insertions(+), 11 deletions(-) diff --git a/third_party/base64.c b/third_party/base64.c index 3350a98ff..f4fbbf477 100644 --- a/third_party/base64.c +++ b/third_party/base64.c @@ -257,7 +257,7 @@ base64_decode_block(const char *in_base64, int in_len, { case step_a: do { - if (in_pos == in_end || out_pos >= out_end) + if (in_pos >= in_end) { state->step = step_a; state->result = curr_byte; @@ -268,7 +268,7 @@ base64_decode_block(const char *in_base64, int in_len, curr_byte = (fragment & 0x03f) << 2; case step_b: do { - if (in_pos == in_end || out_pos >= out_end) + if (in_pos >= in_end) { state->step = step_b; state->result = curr_byte; @@ -277,13 +277,17 @@ base64_decode_block(const char *in_base64, int in_len, fragment = base64_decode_value(*in_pos++); } while (fragment < 0); curr_byte |= (fragment & 0x030) >> 4; - *out_pos++ = curr_byte; + *out_pos = curr_byte; curr_byte = (fragment & 0x00f) << 4; - if (out_pos < out_end) - *out_pos = curr_byte; + if (++out_pos >= out_end) + { + state->step = step_c; + state->result = curr_byte; + return out_pos - out_bin; + } case step_c: do { - if (in_pos == in_end || out_pos >= out_end) + if (in_pos >= in_end) { state->step = step_c; state->result = curr_byte; @@ -292,13 +296,17 @@ base64_decode_block(const char *in_base64, int in_len, fragment = base64_decode_value(*in_pos++); } while (fragment < 0); curr_byte |= (fragment & 0x03c) >> 2; - *out_pos++ = curr_byte; + *out_pos = curr_byte; curr_byte = (fragment & 0x003) << 6; - if (out_pos < out_end) - *out_pos = curr_byte; + if (++out_pos >= out_end) + { + state->step = step_d; + state->result = curr_byte; + return out_pos - out_bin; + } case step_d: do { - if (in_pos == in_end || out_pos >= out_end) + if (in_pos >= in_end) { state->step = step_d; state->result = curr_byte; @@ -307,7 +315,13 @@ base64_decode_block(const char *in_base64, int in_len, fragment = base64_decode_value(*in_pos++); } while (fragment < 0); curr_byte |= (fragment & 0x03f); - *out_pos++ = curr_byte; + *out_pos = curr_byte; + if (++out_pos >= out_end) + { + state->step = step_a; + state->result = curr_byte; + return out_pos - out_bin; + } } } /* control should not reach here */ -- 2.25.1