From patchwork Sat Dec 17 06:57:26 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Richard Henderson X-Patchwork-Id: 18533 Received: (qmail 93276 invoked by alias); 17 Dec 2016 06:57:53 -0000 Mailing-List: contact libc-alpha-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-alpha-owner@sourceware.org Delivered-To: mailing list libc-alpha@sourceware.org Received: (qmail 92927 invoked by uid 89); 17 Dec 2016 06:57:51 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-1.6 required=5.0 tests=BAYES_00, FREEMAIL_ENVFROM_END_DIGIT, FREEMAIL_FROM, RCVD_IN_DNSWL_NONE, SPF_PASS autolearn=no version=3.3.2 spammy=MERGE, eos X-HELO: mail-pg0-f67.google.com X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:sender:from:to:subject:date:message-id :in-reply-to:references; bh=RSUMFkVJGuR6ztKBROgWhuYEuAmVGpyKKrjL0eIQ5lo=; b=AhvyW2aKanCfetlstp0NNNhHCFpM6T4wfG5cq2/aCEP+1jCU9JA6bk6raBfzfWaPxy tXy6oR804JpsVPFUP896ed1Iog0OPuore01dNz+IOVTzxEsBSihOjmQol1uxPq+g0X0k 3B1G/e6+0pzaiJPQQlZe8Hc8ewgQwMsXWy9tQS5RypQ8KcWuzOUtSpPtpivez/trWsGV BSFKZi8rsfqalGyCs8qokZ1KVYW6C/nDwJjHztd8aGost+D5OZkQlfAmew8JshSicXKT /bj1711A/PHsJL3olSmzaqn6naxWZh/EkY1g6MBHOPr6SkURuqIor2TrON5FipWSi+Yr d/hA== X-Gm-Message-State: AKaTC01ZviMnyQWwj8lyBJkrHar92CuY/Bxbgb0LHOALurKsMwr8dIu++R0IfhbOXWGCCA== X-Received: by 10.84.164.106 with SMTP id m39mr14866935plg.97.1481957865033; Fri, 16 Dec 2016 22:57:45 -0800 (PST) From: Richard Henderson To: libc-alpha@sourceware.org Subject: [PATCH 08/11] Improve generic strcmp Date: Fri, 16 Dec 2016 22:57:26 -0800 Message-Id: <20161217065729.28561-9-rth@twiddle.net> In-Reply-To: <20161217065729.28561-1-rth@twiddle.net> References: <20161217065729.28561-1-rth@twiddle.net> * string/strcmp.c: Rewrite using memcopy.h and haszero.h. --- string/strcmp.c | 100 +++++++++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 92 insertions(+), 8 deletions(-) diff --git a/string/strcmp.c b/string/strcmp.c index 4b16f99..2b32854 100644 --- a/string/strcmp.c +++ b/string/strcmp.c @@ -16,6 +16,11 @@ . */ #include +#include +#include +#include +#include +#include #undef strcmp @@ -29,19 +34,98 @@ int STRCMP (const char *p1, const char *p2) { - const unsigned char *s1 = (const unsigned char *) p1; - const unsigned char *s2 = (const unsigned char *) p2; + const unsigned long int *x1, *x2; + unsigned long int w1, w2; unsigned char c1, c2; + uintptr_t i, n, ofs; + int diff; - do + /* Handle the unaligned bytes of p1 first. */ + n = -(uintptr_t)p1 % sizeof(unsigned long int); + for (i = 0; i < n; ++i) { - c1 = (unsigned char) *s1++; - c2 = (unsigned char) *s2++; - if (c1 == '\0') - return c1 - c2; + c1 = *p1++; + c2 = *p2++; + diff = c1 - c2; + if (c1 == '\0' || diff) + return diff; } - while (c1 == c2); + /* P1 is now aligned to unsigned long. P2 may or may not be. */ + x1 = (const unsigned long int *)p1; + w1 = *x1++; + ofs = (uintptr_t)p2 % sizeof(unsigned long int); + if (ofs == 0) + { + x2 = (const unsigned long int *)p2; + w2 = *x2++; + /* Aligned loop. If a difference is found, exit to compare the + bytes. Else if a zero is found we have equal strings. */ + while (w1 == w2) + { + if (haszero (w1)) + return 0; + w1 = *x1++; + w2 = *x2++; + } + } + else + { + unsigned long int w2a, w2b; + uintptr_t sh_1, sh_2; + + x2 = (const unsigned long int *)(p2 - ofs); + w2a = *x2++; + sh_1 = ofs * CHAR_BIT; + sh_2 = sizeof(unsigned long int) * CHAR_BIT - sh_1; + + /* Align the first partial of P2, with 0xff for the rest of the + bytes so that we can also apply the haszero test to see if we + have already reached EOS. If we have, then we can simply fall + through to the final comparison. */ + w2 = MERGE (w2a, sh_1, -1UL, sh_2); + if (!haszero (w2)) + { + /* Unaligned loop. The invariant is that W2B, which is "ahead" + of W1, does not contain end-of-string. Therefore it is safe + (and necessary) to read another word from each while we do + not have a difference. */ + while (1) + { + w2b = *x2++; + w2 = MERGE (w2a, sh_1, w2b, sh_2); + if (w1 != w2) + goto final_cmp; + if (haszero (w2b)) + break; + w1 = *x1++; + w2a = w2b; + } + + /* Zero found in the second partial of P2. If we had EOS + in the aligned word, we have equality. */ + if (haszero (w1)) + return 0; + + /* Load the final word of P1 and align the final partial of P2. */ + w1 = *x1++; + w2 = MERGE (w2b, sh_1, 0, sh_2); + } + } + + final_cmp: + /* We have two aligned words of data. */ + for (i = 0; i < sizeof(unsigned long int) - 1; ++i) + { + c1 = extractbyte (w1, i); + c2 = extractbyte (w2, i); + diff = c1 - c2; + if (c1 == '\0' || diff) + return diff; + } + c1 = extractbyte (w1, i); + c2 = extractbyte (w2, i); return c1 - c2; } + libc_hidden_builtin_def (strcmp)