From patchwork Mon Nov 6 21:01:50 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mark Salyzyn X-Patchwork-Id: 10044487 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 8C7B0604D3 for ; Mon, 6 Nov 2017 21:03:21 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 796322A00D for ; Mon, 6 Nov 2017 21:03:21 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 780962A010; Mon, 6 Nov 2017 21:03:21 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-4.2 required=2.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID,RCVD_IN_DNSWL_MED autolearn=unavailable version=3.3.1 Received: from bombadil.infradead.org (bombadil.infradead.org [65.50.211.133]) (using TLSv1.2 with cipher AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.wl.linuxfoundation.org (Postfix) with ESMTPS id 1FCDB2A01C for ; Mon, 6 Nov 2017 21:02:53 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20170209; h=Sender: Content-Transfer-Encoding:Content-Type:MIME-Version:Cc:List-Subscribe: List-Help:List-Post:List-Archive:List-Unsubscribe:List-Id:Message-Id:Date: Subject:To:From:Reply-To:Content-ID:Content-Description:Resent-Date: Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:In-Reply-To: References:List-Owner; bh=8bRq3sH96mUzLpFNYqc1kMSyOsMpi1gXV3diqvr1CV4=; b=Bqz WmltyqJwjgc0BnA6LkhLVjTILoCJl/HL6EPKKQOWDwIgzk301mt6bEqwrffA6n8CeX13ibFCiQ0aX +FPGIGeZm8i/WRsd5eahZNYYuo8BXfwCn5TFLJHchEQIQGQJLrcgXTpw9MltKxr2bTe7lOpN8q1cZ iaB3c9EjxjXm+4y3Kb5EkjWhaNLqmaqVHuOqcRziAheZNll9O2/7DpirqU1Nj8cB9jjytCbuwk5yF E4noV8rNaR4hKXqlUgjoF9HGv7sz1geSathi3iwB2Esfwur3HYAMm+Z/1f7m/aAdtPLUscQ11La23 lZ74NDl5+0SGeylOJIRKQEj90Nccssw==; Received: from localhost ([127.0.0.1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.87 #1 (Red Hat Linux)) id 1eBoXd-000239-QJ; Mon, 06 Nov 2017 21:02:45 +0000 Received: from mail-it0-x243.google.com ([2607:f8b0:4001:c0b::243]) by bombadil.infradead.org with esmtps (Exim 4.87 #1 (Red Hat Linux)) id 1eBoXH-0001ZE-Lq for linux-arm-kernel@lists.infradead.org; Mon, 06 Nov 2017 21:02:42 +0000 Received: by mail-it0-x243.google.com with SMTP id y15so7060413ita.4 for ; Mon, 06 Nov 2017 13:02:02 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=android.com; s=20161025; h=from:to:cc:subject:date:message-id; bh=pn6USHw50KK30Mk8TOuZ80xXOcXOA1dRNTqzs+bhLEc=; b=qUBRC2HaQhJ7U81IvN6zLkHkSHvFYVGne3QNsyJjs1JlOG2UXpmNbYKdwbHmEZEMHn vTgXw0EKywHtoVEGS43mRBZCjKuHqeaSQo2SPwVuCWU8Vs0Ki6hkG2bqs+Ok5mET/DeU rko9GZ85XcxzwWNfe1tiTn5ZC/sAL97li2SK11xL//TtePFJrwaZ6mW9Sjkj4du+jvZT 6NYgTZr61unk55raZid+1KghVYQB3Z+9v7eIY1YWJq4hAi3zhEagFr+Oj4z6/TZT4OLt IL4F0o5a46x+kZUfuxRGlP9jyGqcXYY6RzVkNTdfRShk8w68bxImM971dUY4Fghxap2N 44KA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id; bh=pn6USHw50KK30Mk8TOuZ80xXOcXOA1dRNTqzs+bhLEc=; b=PKmkvaO8MEtB+fsdu0nnV9FGglndhVcK4Hj9v8bQdasz3rfqEH7U2iDFhtwQOxsPOx LjwhpSkYUI8OM4HuwnmywJktwUYERuANVHD2ALzVdSTQ2Qssr/vb+752zaNqeEAOiojf 4y9XpxWMFsyZ8Gw92G6RbqIJTojc5RZ94ah29zmGw9seP5aMoEKm7R+h5x92MuV1Y9n7 10FxVH0n7dgv0x+4hLQ2QJMMrcMf1dya2OBmYfnYM7yiJJUnVbXOIyYGYm7SvhNo8d/T 18goCiRXEluj3aLCVCNdb5Vw7C0A9TrK2WDiU2toQ2UzjTBmtVYQI0QG8fmCyEWQQd2l JCEw== X-Gm-Message-State: AJaThX6l98yqD13pAaBGSBdKrY6R3uAFPR+TWH2io/nzFNcVXZWDO9uO sfTPCRX1JBebEhH99ao6hJXdjA== X-Google-Smtp-Source: ABhQp+TDdUoprzSB6EAY9C+aOKUkYXHd/lX+MTlaWAm7Lz5x101wMnQevUY2gUG5iR92T+HLWHnBiw== X-Received: by 10.36.13.19 with SMTP id 19mr10735928itx.107.1510002122068; Mon, 06 Nov 2017 13:02:02 -0800 (PST) Received: from nebulus.mtv.corp.google.com ([100.98.120.17]) by smtp.gmail.com with ESMTPSA id 81sm4955678itl.20.2017.11.06.13.02.00 (version=TLS1_2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Mon, 06 Nov 2017 13:02:01 -0800 (PST) From: Mark Salyzyn To: linux-kernel@vger.kernel.org Subject: [PATCH v5 04/12] arm: vdso: do calculations outside reader loops Date: Mon, 6 Nov 2017 13:01:50 -0800 Message-Id: <20171106210157.94798-1-salyzyn@android.com> X-Mailer: git-send-email 2.15.0.403.gc27cc4dac6-goog X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20171106_130224_487300_353E6D30 X-CRM114-Status: GOOD ( 18.41 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Mark Rutland , Laura Abbott , Kees Cook , Ard Biesheuvel , Catalin Marinas , Kevin Brodsky , Will Deacon , Russell King , Andy Lutomirski , James Morse , Andrew Pinski , Dmitry Safonov , Andy Gross , John Stultz , Thomas Gleixner , Mark Salyzyn , linux-arm-kernel@lists.infradead.org MIME-Version: 1.0 Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+patchwork-linux-arm=patchwork.kernel.org@lists.infradead.org X-Virus-Scanned: ClamAV using ClamSMTP Take an effort to recode the arm64 vdso code from assembler to C previously submitted by Andrew Pinski , rework it for use in both arm and arm64, overlapping any optimizations for each architecture. But instead of landing it in arm64, land the result into lib/vdso and unify both implementations to simplify future maintenance. In variable timer reading loops, pick up just the values until all are synchronized, then outside of loop pick up cntvct and perform calculations to determine final offset, shifted and multiplied output value. This replaces get_ns with get_clock_shifted_nsec as cntvct reader. Signed-off-by: Mark Salyzyn Cc: James Morse Cc: Russell King Cc: Catalin Marinas Cc: Will Deacon Cc: Andy Lutomirski Cc: Dmitry Safonov Cc: John Stultz Cc: Mark Rutland Cc: Laura Abbott Cc: Kees Cook Cc: Ard Biesheuvel Cc: Andy Gross Cc: Kevin Brodsky Cc: Andrew Pinski Cc: Thomas Gleixner Cc: linux-kernel@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org v2: - split first CL into 5 of 7 pieces v4: - split into two, moving ARCH_CLOCK_FIXED_MASK to later - update commit message to reflect overall reasoning - adjust for dropping forced inline - replace typeof() with types vdso_wtm_clock_nsec_t, vdso_xtime_clock_sec and __kernel_time_t. v5: - drop including linux/time.h in favour of uapi/linux/time.h in vgettimeofday.c to limit architectural includes. - add linux/time.h to compiler.h for NSEC_PER_SEC definition. - replace last timespec_add_ns with __iter_div_u64_rem. --- arch/arm/include/asm/vdso_datapage.h | 18 ++++++-- arch/arm/vdso/compiler.h | 1 + arch/arm/vdso/vgettimeofday.c | 90 ++++++++++++++++++++++++------------ 3 files changed, 76 insertions(+), 33 deletions(-) diff --git a/arch/arm/include/asm/vdso_datapage.h b/arch/arm/include/asm/vdso_datapage.h index fa3e1856244d..8dd7303db4ec 100644 --- a/arch/arm/include/asm/vdso_datapage.h +++ b/arch/arm/include/asm/vdso_datapage.h @@ -24,6 +24,16 @@ #include +#ifndef _VDSO_WTM_CLOCK_SEC_T +#define _VDSO_WTM_CLOCK_SEC_T +typedef u32 vdso_wtm_clock_nsec_t; +#endif + +#ifndef _VDSO_XTIME_CLOCK_SEC_T +#define _VDSO_XTIME_CLOCK_SEC_T +typedef u32 vdso_xtime_clock_sec_t; +#endif + /* Try to be cache-friendly on systems that don't implement the * generic timer: fit the unconditionally updated fields in the first * 32 bytes. @@ -35,9 +45,11 @@ struct vdso_data { u32 xtime_coarse_sec; /* coarse time */ u32 xtime_coarse_nsec; - u32 wtm_clock_sec; /* wall to monotonic offset */ - u32 wtm_clock_nsec; - u32 xtime_clock_sec; /* CLOCK_REALTIME - seconds */ + /* wall to monotonic offset */ + u32 wtm_clock_sec; + vdso_wtm_clock_nsec_t wtm_clock_nsec; + /* CLOCK_REALTIME - seconds */ + vdso_xtime_clock_sec_t xtime_clock_sec; u32 cs_mono_mult; /* clocksource multiplier */ u64 cs_cycle_last; /* last cycle value */ diff --git a/arch/arm/vdso/compiler.h b/arch/arm/vdso/compiler.h index af24502797e8..3edddb705a1b 100644 --- a/arch/arm/vdso/compiler.h +++ b/arch/arm/vdso/compiler.h @@ -27,6 +27,7 @@ #include /* for cpu_relax() */ #include #include +#include /* for NSEC_PER_SEC */ #ifndef CONFIG_AEABI #error This code depends on AEABI system call conventions diff --git a/arch/arm/vdso/vgettimeofday.c b/arch/arm/vdso/vgettimeofday.c index 522094b147a2..59893fca03b3 100644 --- a/arch/arm/vdso/vgettimeofday.c +++ b/arch/arm/vdso/vgettimeofday.c @@ -24,7 +24,8 @@ #include #include /* for notrace */ -#include +#include /* for __iter_div_u64_rem() */ +#include /* for struct timespec */ #include "compiler.h" #include "datapage.h" @@ -79,6 +80,7 @@ static notrace int do_monotonic_coarse(const struct vdso_data *vd, { struct timespec tomono; u32 seq; + u64 nsec; do { seq = vdso_read_begin(vd); @@ -92,33 +94,41 @@ static notrace int do_monotonic_coarse(const struct vdso_data *vd, } while (vdso_read_retry(vd, seq)); ts->tv_sec += tomono.tv_sec; - timespec_add_ns(ts, tomono.tv_nsec); + /* open coding timespec_add_ns */ + ts->tv_sec += __iter_div_u64_rem(ts->tv_nsec + tomono.tv_nsec, + NSEC_PER_SEC, &nsec); + ts->tv_nsec = nsec; return 0; } #ifdef CONFIG_ARM_ARCH_TIMER -static notrace u64 get_ns(const struct vdso_data *vd) +/* + * Returns the clock delta, in nanoseconds left-shifted by the clock + * shift. + */ +static notrace u64 get_clock_shifted_nsec(const u64 cycle_last, + const u32 mult, + const u64 mask) { - u64 cycle_delta; - u64 cycle_now; - u64 nsec; - - cycle_now = arch_vdso_read_counter(); + u64 res; - cycle_delta = (cycle_now - vd->cs_cycle_last) & vd->cs_mask; + /* Read the virtual counter. */ + res = arch_vdso_read_counter(); - nsec = (cycle_delta * vd->cs_mono_mult) + vd->xtime_clock_snsec; - nsec >>= vd->cs_shift; + res = res - cycle_last; - return nsec; + res &= mask; + return res * mult; } static notrace int do_realtime(const struct vdso_data *vd, struct timespec *ts) { - u64 nsecs; - u32 seq; + u32 seq, mult, shift; + u64 nsec, cycle_last; + u64 mask; + vdso_xtime_clock_sec_t sec; do { seq = vdso_read_begin(vd); @@ -126,22 +136,33 @@ static notrace int do_realtime(const struct vdso_data *vd, struct timespec *ts) if (vd->use_syscall) return -1; - ts->tv_sec = vd->xtime_clock_sec; - nsecs = get_ns(vd); + cycle_last = vd->cs_cycle_last; - } while (vdso_read_retry(vd, seq)); + mult = vd->cs_mono_mult; + shift = vd->cs_shift; + mask = vd->cs_mask; + + sec = vd->xtime_clock_sec; + nsec = vd->xtime_clock_snsec; + + } while (unlikely(vdso_read_retry(vd, seq))); - ts->tv_nsec = 0; - timespec_add_ns(ts, nsecs); + nsec += get_clock_shifted_nsec(cycle_last, mult, mask); + nsec >>= shift; + /* open coding timespec_add_ns to save a ts->tv_nsec = 0 */ + ts->tv_sec = sec + __iter_div_u64_rem(nsec, NSEC_PER_SEC, &nsec); + ts->tv_nsec = nsec; return 0; } static notrace int do_monotonic(const struct vdso_data *vd, struct timespec *ts) { - struct timespec tomono; - u64 nsecs; - u32 seq; + u32 seq, mult, shift; + u64 nsec, cycle_last; + u64 mask; + vdso_wtm_clock_nsec_t wtm_nsec; + __kernel_time_t sec; do { seq = vdso_read_begin(vd); @@ -149,17 +170,26 @@ static notrace int do_monotonic(const struct vdso_data *vd, struct timespec *ts) if (vd->use_syscall) return -1; - ts->tv_sec = vd->xtime_clock_sec; - nsecs = get_ns(vd); + cycle_last = vd->cs_cycle_last; - tomono.tv_sec = vd->wtm_clock_sec; - tomono.tv_nsec = vd->wtm_clock_nsec; + mult = vd->cs_mono_mult; + shift = vd->cs_shift; + mask = vd->cs_mask; - } while (vdso_read_retry(vd, seq)); + sec = vd->xtime_clock_sec; + nsec = vd->xtime_clock_snsec; - ts->tv_sec += tomono.tv_sec; - ts->tv_nsec = 0; - timespec_add_ns(ts, nsecs + tomono.tv_nsec); + sec += vd->wtm_clock_sec; + wtm_nsec = vd->wtm_clock_nsec; + + } while (unlikely(vdso_read_retry(vd, seq))); + + nsec += get_clock_shifted_nsec(cycle_last, mult, mask); + nsec >>= shift; + nsec += wtm_nsec; + /* open coding timespec_add_ns to save a ts->tv_nsec = 0 */ + ts->tv_sec = sec + __iter_div_u64_rem(nsec, NSEC_PER_SEC, &nsec); + ts->tv_nsec = nsec; return 0; }