From patchwork Thu Sep 24 15:13:01 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "H.J. Lu" X-Patchwork-Id: 40488 Return-Path: X-Original-To: patchwork@sourceware.org Delivered-To: patchwork@sourceware.org Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id 89FC13959E59; Thu, 24 Sep 2020 15:13:41 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 89FC13959E59 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=sourceware.org; s=default; t=1600960421; bh=IRPRW9hCJtDszjfwnBc3XqPS4opLkb+jxnGndEdmp6c=; h=References:In-Reply-To:Date:Subject:To:List-Id:List-Unsubscribe: List-Archive:List-Post:List-Help:List-Subscribe:From:Reply-To:Cc: From; b=xlHj2WK2vTCiYSy0Xr6WG85YJYOTgSN64D1ARrYpWVvEKwV+zFcfI9zqr6k8kCoFN TLRFQit4cR1/Iuue64x7/VXbJ301kMK6KpQOdOKiwpkSHUKrsGCFWyRhlSZk3QKKq6 TG8cbwTwWH4n1/40amAc27H2SHpnxgMRxsELPrpY= X-Original-To: libc-alpha@sourceware.org Delivered-To: libc-alpha@sourceware.org Received: from mail-il1-x144.google.com (mail-il1-x144.google.com [IPv6:2607:f8b0:4864:20::144]) by sourceware.org (Postfix) with ESMTPS id 4BF81386F012 for ; Thu, 24 Sep 2020 15:13:38 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.3.2 sourceware.org 4BF81386F012 Received: by mail-il1-x144.google.com with SMTP id y9so3517897ilq.2 for ; Thu, 24 Sep 2020 08:13:38 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=IRPRW9hCJtDszjfwnBc3XqPS4opLkb+jxnGndEdmp6c=; b=HioVidWw41J5K4fTJ9pH2gXUA73yFQcWavgjDcEv2wsvHTKJCkyQGIouOASioz+rJ4 3il5mrHM/tRKhnRrbKl0BiGtFHVLmxn1X04grwrLJI4HA3/ScV7VZWTQSQDFCdJC4uaL ocVo2prxmjquU7FdPI0wzHa9NXQZFKg5outla/UbhwXSZhFne3SzBUOLSv1eAR5fUBfz kFTaubo8NAexScf5TEtznY8Du1Nsv5t58sFZ736COxw5bbC9nuZpdDzEjtOotIWheFXT Vi0A4j0qt24v2eaGDgL8sBHSx7PXqlDPx6rDHopVn85RSCnALvyM4xkbClk0CBV/Xr6x ys3Q== X-Gm-Message-State: AOAM533Wya6KU6QJVlD20n/1nhblJHNGYhdL2869QLuY99LUG8FmWODL b6N5UkDubV0xDskt4mEYRGCGvLghJyeLmIewinVQhXuW X-Google-Smtp-Source: ABdhPJy7+q7G0eJQBR5I0GAB9hrlGLoZQnNCZ0a2aUkyn2Vcq4QKHCudI85MYO/GkXhicpCkEzc1SCrHkCDy+R+Mhu8= X-Received: by 2002:a92:1589:: with SMTP id 9mr4441758ilv.292.1600960417731; Thu, 24 Sep 2020 08:13:37 -0700 (PDT) MIME-Version: 1.0 References: <20200612201056.228614-1-hjl.tools@gmail.com> <20200612201056.228614-3-hjl.tools@gmail.com> In-Reply-To: Date: Thu, 24 Sep 2020 08:13:01 -0700 Message-ID: Subject: V2 [PATCH] bench-strncmp.c: Add workloads on page boundary To: "Carlos O'Donell" X-Spam-Status: No, score=-8.0 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, FREEMAIL_FROM, GIT_PATCH_0, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.2 X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on server2.sourceware.org X-BeenThere: libc-alpha@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Libc-alpha mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-Patchwork-Original-From: "H.J. Lu via Libc-alpha" From: "H.J. Lu" Reply-To: "H.J. Lu" Cc: GNU C Library Errors-To: libc-alpha-bounces@sourceware.org Sender: "Libc-alpha" On Wed, Sep 23, 2020 at 5:46 PM Carlos O'Donell wrote: > > On 6/12/20 4:10 PM, H.J. Lu via Libc-alpha wrote: > > Add strncmp workloads on page boundary. > > This benchmark need some work. > > Please post v2. > > > --- > > benchtests/bench-strncmp.c | 117 +++++++++++++++++++++++++++++++++++++ > > 1 file changed, 117 insertions(+) > > > > diff --git a/benchtests/bench-strncmp.c b/benchtests/bench-strncmp.c > > index 95a59c9465..065c7e7789 100644 > > --- a/benchtests/bench-strncmp.c > > +++ b/benchtests/bench-strncmp.c > > @@ -27,6 +27,7 @@ > > > > #ifdef WIDE > > # define L(str) L##str > > +# define STRDUP wcsdup > > OK. > > > # define SIMPLE_STRNCMP simple_wcsncmp > > > > /* Wcsncmp uses signed semantics for comparison, not unsigned. > > @@ -48,6 +49,7 @@ simple_wcsncmp (const CHAR *s1, const CHAR *s2, size_t n) > > > > #else > > # define L(str) str > > +# define STRDUP strdup > > OK. > > > # define SIMPLE_STRNCMP simple_strncmp > > > > /* Strncmp uses unsigned semantics for comparison. */ > > @@ -190,6 +192,118 @@ do_test (json_ctx_t *json_ctx, size_t align1, size_t align2, size_t len, size_t > > json_element_object_end (json_ctx); > > } > > > > +static void > > +do_test_page_boundary_1 (json_ctx_t *json_ctx, CHAR *s1, CHAR *s2, > > Rename to "do_one_test_page_boundary" Done. > > + size_t align1, size_t align2, size_t len, > > + size_t n, int exp_result) > > +{ > > + json_element_object_begin (json_ctx); > > + json_attr_uint (json_ctx, "strlen", (double) len); > > + json_attr_uint (json_ctx, "len", (double) n); > > + json_attr_uint (json_ctx, "align1", (double) align1); > > + json_attr_uint (json_ctx, "align2", (double) align2); > > + json_array_begin (json_ctx, "timings"); > > + FOR_EACH_IMPL (impl, 0) > > + do_one_test (json_ctx, impl, s1, s2, n, exp_result); > > + json_array_end (json_ctx); > > + json_element_object_end (json_ctx); > > OK. > > > +} > > + > > > Add a comment: > > /* To trigger bug 25933 we need a size that is equal to the > vector length times 4. In the case of AVX2 for Intel we > need 32 * 4. We make this test generic and run it for all > architectures as additional boundary testing for such > related algorithms. */ Done. > > +static void > > +do_test_page_boundary (json_ctx_t *json_ctx) > > OK. > > > +{ > > + size_t size = 32 * 4; > > + size_t len; > > + CHAR *s1 = (CHAR *) (buf1 + (BUF1PAGES - 1) * page_size); > > + CHAR *s2 = (CHAR *) (buf2 + (BUF1PAGES - 1) * page_size); > > + int exp_result; > > + > > + memset (s1, 'a', page_size); > > + memset (s2, 'a', page_size); > > + > > + s1[(page_size / CHARBYTES) - 1] = (CHAR) 0; > > + > > Add comment: > > /* Iterate over a size that is just below where we expect > the bug to trigger up to the size we expect will trigger > the bug e.g. [99-128]. Likewise iterate the start of > two strings between 30 and 31 bytes away from the > boundary to simulate alignment changes. */ Done. > > + for (size_t s = 99; s <= size; s++) > > + for (size_t s1a = 31; s1a < 32; s1a++) > > Please make s1a iterate over [30,32) like s2a. Done. > > + for (size_t s2a = 30; s2a < 32; s2a++) > > + { > > + size_t align1 = (page_size / CHARBYTES - s) - s1a; > > + size_t align2 = (page_size / CHARBYTES - s) - s2a; > > + CHAR *s1p = s1 + align1; > > + CHAR *s2p = s2 + align2; > > + len = (page_size / CHARBYTES) - 1 - align1; > > + exp_result = SIMPLE_STRNCMP (s1p, s2p, s); > > + do_test_page_boundary_1 (json_ctx, s1p, s2p, align1, align2, > > + len, s, exp_result); > > + } > > +} > > + > > +static void > > +do_page_test (json_ctx_t *json_ctx, size_t offset1, size_t offset2, > > Rename to do_one_test_page() Done. > > + CHAR *s2) > > +{ > > + CHAR *s1; > > + int exp_result; > > + > > + if (offset1 * CHARBYTES >= page_size > > + || offset2 * CHARBYTES >= page_size) > > + return; > > OK. Nothing bigger than a page size because we only have one page to work within. > > > + > > + s1 = (CHAR *) buf1; > > + s1 += offset1; > > + s2 += offset2; > > + > > + size_t len = (page_size / CHARBYTES) - offset1; > > + > > + exp_result= *s1; > > + > > + json_element_object_begin (json_ctx); > > + json_attr_uint (json_ctx, "strlen", (double) len); > > + json_attr_uint (json_ctx, "len", (double) page_size); > > + json_attr_uint (json_ctx, "align1", (double) offset1); > > + json_attr_uint (json_ctx, "align2", (double) offset2); > > + json_array_begin (json_ctx, "timings"); > > + { > > + FOR_EACH_IMPL (impl, 0) > > + do_one_test (json_ctx, impl, s1, s2, page_size, -exp_result); > > + } > > + json_array_end (json_ctx); > > + json_element_object_end (json_ctx); > > + > > + json_element_object_begin (json_ctx); > > + json_attr_uint (json_ctx, "strlen", (double) len); > > + json_attr_uint (json_ctx, "len", (double) page_size); > > + json_attr_uint (json_ctx, "align1", (double) offset1); > > + json_attr_uint (json_ctx, "align2", (double) offset2); > > + json_array_begin (json_ctx, "timings"); > > + { > > + FOR_EACH_IMPL (impl, 0) > > + do_one_test (json_ctx, impl, s1, s2, page_size, exp_result); > > + } > > + json_array_end (json_ctx); > > + json_element_object_end (json_ctx); > > +} > > OK. > > > + > > +static void > > +do_test_page (json_ctx_t *json_ctx) > > +{ > > + size_t i; > > + CHAR *s1, *s2; > > + > > + s1 = (CHAR *) buf1; > > + for (i = 0; i < (page_size / CHARBYTES) - 1; i++) > > + s1[i] = 23; > > Add a comment for this magic fill number. Done. > > + s1[i] = 0; > > + > > + s2 = STRDUP (s1); > > + > > + for (i = 0; i < 64; ++i) > > Add a comment for these magic numbers. Done. > > + do_page_test (json_ctx, (3988 / CHARBYTES) + i, > > + (2636 / CHARBYTES), s2); > > Call do_one_test_page() Done. > > + > > + free (s2); > > +} > > + > > int > > test_main (void) > > { > > @@ -267,6 +381,9 @@ test_main (void) > > do_test_limit (&json_ctx, 0, 0, 15 - i, 16 - i, 255, -1); > > } > > > > + do_test_page_boundary (&json_ctx); > > + do_test_page (&json_ctx); > > OK. > > > + > > json_array_end (&json_ctx); > > json_attr_object_end (&json_ctx); > > json_attr_object_end (&json_ctx); > > > Here is the updated patch. OK for master? Thanks. Reviewed-by: Carlos O'Donell From 70919738604840c387e342aaea4e6b13f8a16043 Mon Sep 17 00:00:00 2001 From: "H.J. Lu" Date: Thu, 11 Jun 2020 08:52:42 -0700 Subject: [PATCH] bench-strncmp.c: Add workloads on page boundary Add strncmp workloads on page boundary. --- benchtests/bench-strncmp.c | 128 +++++++++++++++++++++++++++++++++++++ 1 file changed, 128 insertions(+) diff --git a/benchtests/bench-strncmp.c b/benchtests/bench-strncmp.c index 95a59c9465..b307f0c555 100644 --- a/benchtests/bench-strncmp.c +++ b/benchtests/bench-strncmp.c @@ -27,6 +27,7 @@ #ifdef WIDE # define L(str) L##str +# define STRDUP wcsdup # define SIMPLE_STRNCMP simple_wcsncmp /* Wcsncmp uses signed semantics for comparison, not unsigned. @@ -48,6 +49,7 @@ simple_wcsncmp (const CHAR *s1, const CHAR *s2, size_t n) #else # define L(str) str +# define STRDUP strdup # define SIMPLE_STRNCMP simple_strncmp /* Strncmp uses unsigned semantics for comparison. */ @@ -190,6 +192,129 @@ do_test (json_ctx_t *json_ctx, size_t align1, size_t align2, size_t len, size_t json_element_object_end (json_ctx); } +static void +do_one_test_page_boundary (json_ctx_t *json_ctx, CHAR *s1, CHAR *s2, + size_t align1, size_t align2, size_t len, + size_t n, int exp_result) +{ + json_element_object_begin (json_ctx); + json_attr_uint (json_ctx, "strlen", (double) len); + json_attr_uint (json_ctx, "len", (double) n); + json_attr_uint (json_ctx, "align1", (double) align1); + json_attr_uint (json_ctx, "align2", (double) align2); + json_array_begin (json_ctx, "timings"); + FOR_EACH_IMPL (impl, 0) + do_one_test (json_ctx, impl, s1, s2, n, exp_result); + json_array_end (json_ctx); + json_element_object_end (json_ctx); +} + +static void +do_test_page_boundary (json_ctx_t *json_ctx) +{ + /* To trigger bug 25933, we need a size that is equal to the vector + length times 4. In the case of AVX2 for Intel, we need 32 * 4. We + make this test generic and run it for all architectures as additional + boundary testing for such related algorithms. */ + size_t size = 32 * 4; + size_t len; + CHAR *s1 = (CHAR *) (buf1 + (BUF1PAGES - 1) * page_size); + CHAR *s2 = (CHAR *) (buf2 + (BUF1PAGES - 1) * page_size); + int exp_result; + + memset (s1, 'a', page_size); + memset (s2, 'a', page_size); + + s1[(page_size / CHARBYTES) - 1] = (CHAR) 0; + + /* Iterate over a size that is just below where we expect the bug to + trigger up to the size we expect will trigger the bug e.g. [99-128]. + Likewise iterate the start of two strings between 30 and 31 bytes + away from the boundary to simulate alignment changes. */ + for (size_t s = 99; s <= size; s++) + for (size_t s1a = 30; s1a < 32; s1a++) + for (size_t s2a = 30; s2a < 32; s2a++) + { + size_t align1 = (page_size / CHARBYTES - s) - s1a; + size_t align2 = (page_size / CHARBYTES - s) - s2a; + CHAR *s1p = s1 + align1; + CHAR *s2p = s2 + align2; + len = (page_size / CHARBYTES) - 1 - align1; + exp_result = SIMPLE_STRNCMP (s1p, s2p, s); + do_one_test_page_boundary (json_ctx, s1p, s2p, align1, align2, + len, s, exp_result); + } +} + +static void +do_one_test_page (json_ctx_t *json_ctx, size_t offset1, size_t offset2, + CHAR *s2) +{ + CHAR *s1; + int exp_result; + + if (offset1 * CHARBYTES >= page_size + || offset2 * CHARBYTES >= page_size) + return; + + s1 = (CHAR *) buf1; + s1 += offset1; + s2 += offset2; + + size_t len = (page_size / CHARBYTES) - offset1; + + exp_result= *s1; + + json_element_object_begin (json_ctx); + json_attr_uint (json_ctx, "strlen", (double) len); + json_attr_uint (json_ctx, "len", (double) page_size); + json_attr_uint (json_ctx, "align1", (double) offset1); + json_attr_uint (json_ctx, "align2", (double) offset2); + json_array_begin (json_ctx, "timings"); + { + FOR_EACH_IMPL (impl, 0) + do_one_test (json_ctx, impl, s1, s2, page_size, -exp_result); + } + json_array_end (json_ctx); + json_element_object_end (json_ctx); + + json_element_object_begin (json_ctx); + json_attr_uint (json_ctx, "strlen", (double) len); + json_attr_uint (json_ctx, "len", (double) page_size); + json_attr_uint (json_ctx, "align1", (double) offset1); + json_attr_uint (json_ctx, "align2", (double) offset2); + json_array_begin (json_ctx, "timings"); + { + FOR_EACH_IMPL (impl, 0) + do_one_test (json_ctx, impl, s1, s2, page_size, exp_result); + } + json_array_end (json_ctx); + json_element_object_end (json_ctx); +} + +static void +do_test_page (json_ctx_t *json_ctx) +{ + size_t i; + CHAR *s1, *s2; + + s1 = (CHAR *) buf1; + /* Fill buf1 with 23. */ + for (i = 0; i < (page_size / CHARBYTES) - 1; i++) + s1[i] = 23; + s1[i] = 0; + + /* Make a copy of buf1. */ + s2 = STRDUP (s1); + + /* Test should terminate within the page boundary. */ + for (i = 0; i < (108 / CHARBYTES); ++i) + do_one_test_page (json_ctx, ((page_size - 108) / CHARBYTES) + i, + ((page_size - 1460) / CHARBYTES), s2); + + free (s2); +} + int test_main (void) { @@ -267,6 +392,9 @@ test_main (void) do_test_limit (&json_ctx, 0, 0, 15 - i, 16 - i, 255, -1); } + do_test_page_boundary (&json_ctx); + do_test_page (&json_ctx); + json_array_end (&json_ctx); json_attr_object_end (&json_ctx); json_attr_object_end (&json_ctx); -- 2.26.2