From patchwork Wed Jan 8 11:45:32 2025 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Vincent Donnefort X-Patchwork-Id: 13930655 Received: from mail-wm1-f73.google.com (mail-wm1-f73.google.com [209.85.128.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 7F3921FBEA5 for ; Wed, 8 Jan 2025 11:46:19 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.128.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1736336781; cv=none; b=QZxH0yOsf31crgMGVDH5bi38uFwEeLAIO2FF89OgOB46uOeK8PXNgMsGqJq0uTVWn3vxDwMHocM/3qaF1lQkDTgv6aVPN25wz+agA35D+147I7D2BfZZfLEgU/ttDYPmuQqZo2nk+m/2Okxo6kSVlcKUD/OOLwryRS55prRLzD8= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1736336781; c=relaxed/simple; bh=grcM2cIQ4yikLiM8Dd1jmPsJ/2R8HySvbTvwaZZ+ID0=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Cc:Content-Type; b=UcX0ID3UxAAhP541V8djddC0WhaI2LXzIy9efUBDvY381FqRk+Jy6kMTpR9hhn7sHm64oMXEuYNitZO029sIc/eLJLU/m629WNsnVa7K5w1o68lud/huz3R42gidJrx8hlhi+35SaUfc/vHzWmUn5+ILjV79VZemOX3im8iRKwA= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--vdonnefort.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=EtgqNOCL; arc=none smtp.client-ip=209.85.128.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--vdonnefort.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="EtgqNOCL" Received: by mail-wm1-f73.google.com with SMTP id 5b1f17b1804b1-4361b090d23so48862875e9.0 for ; Wed, 08 Jan 2025 03:46:19 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1736336778; x=1736941578; darn=vger.kernel.org; h=cc:to:from:subject:message-id:references:mime-version:in-reply-to :date:from:to:cc:subject:date:message-id:reply-to; bh=E/ljNWnFFLsO0qM4wfk7we1pD6ZL8txquRN7nzlRbfw=; b=EtgqNOCLg3gfPT8J8Zf+fnN2NCq7Zsm22FG08kBbc6yHSFpKfvDQ6LETx89i52NxRy xMUMvCx10FkvT2QboJf7tARIjIj8MAYk5Ji/af9NbPsXvhSI3x9PtTt0v99Xj4ZLb7+o AcXLSkWJFGpfOmd6gU1sx3UT+TJw4aqnHaVltBIiiGIxoaIrieVLP1gkbAxvpXkKquKR zRSkapA/vP+3Bg8diHRmHohNn2hnDcRK2ixiXqpnmqsxfbCr65zHjbg+gTYM5eVFWANO gtrdDJlSvS9jtrHU8CMsCsn90n5rNUshbE0SqtR3c4ISXtqF3IXMpnoHN/0kwOEvXK/K pp6g== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1736336778; x=1736941578; h=cc:to:from:subject:message-id:references:mime-version:in-reply-to :date:x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=E/ljNWnFFLsO0qM4wfk7we1pD6ZL8txquRN7nzlRbfw=; b=KRDLlJOC/qqZHGdFrhbBve4KDqD7GuCIyXygYIrLyiI8dyb0tcGBnuA3152EwdHAyh S7UwVbOMYysSYxgitd3ohDeDosb5AT0sSr2HVrF8ATuO34jkPuThEjsYaVpS1e6aU1GX 72o0vddeNCrXJP7vRZjD66wkMwlTHTZV5kzZyMpKCro4cmARSZXXNBdRIpxiwWFSId3w u7mdZuxYIzA9iNIdWuu+GK70rcEeEAFL9aj8wTOaCBNt0V0gbOU6ZngUfhZQtRbU3I8Q N1KPlqjfa89h1p54M3SX5aCjz8TgNIa0c2xj7OnCPJSviIxKbj0rd8n1YE4DP+Ovjjm7 RPYw== X-Forwarded-Encrypted: i=1; AJvYcCWJH/T15pNxZRxt9hKj9KKXze2YnMsZ+FgbABxc7Z6JwDKgXMiR2cPRJSgwg8P6k/2u5s3DHnx1ZBi8VdIQJhftYcA=@vger.kernel.org X-Gm-Message-State: AOJu0YyFRDQdhTNz9hsFICQ6GLyqK9dNouEH8XCXg8frtMVvm5jHe5Ys 0eu00qLuEjFinTD3X05wvci7fTA+tBUhbPmTA4NazXHi7ORpoQGLV1qbVB6VPWPn2zRWWOr0ynW V+itDQo7b8GmzDW/YzQ== X-Google-Smtp-Source: AGHT+IHuYaoZkvQka603lwunTyfYYS1GNixlL63EeZDX8dZuJ7nKrAaTL9PyKuw1CaVQQyEwJdzqG3uc5Wv5Ofuj X-Received: from wmqa17.prod.google.com ([2002:a05:600c:3491:b0:434:fa72:f1bf]) (user=vdonnefort job=prod-delivery.src-stubby-dispatcher) by 2002:a05:600c:4f06:b0:436:6160:5b81 with SMTP id 5b1f17b1804b1-436e26b98d1mr21080765e9.14.1736336778060; Wed, 08 Jan 2025 03:46:18 -0800 (PST) Date: Wed, 8 Jan 2025 11:45:32 +0000 In-Reply-To: <20250108114536.627715-1-vdonnefort@google.com> Precedence: bulk X-Mailing-List: linux-trace-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250108114536.627715-1-vdonnefort@google.com> X-Mailer: git-send-email 2.47.1.613.gc27f4b7a9f-goog Message-ID: <20250108114536.627715-9-vdonnefort@google.com> Subject: [PATCH v2 08/12] KVM: arm64: Add clock for hyp tracefs From: Vincent Donnefort To: rostedt@goodmis.org, mhiramat@kernel.org, mathieu.desnoyers@efficios.com, linux-trace-kernel@vger.kernel.org, maz@kernel.org, oliver.upton@linux.dev, joey.gouly@arm.com, suzuki.poulose@arm.com, yuzenghui@huawei.com Cc: kvmarm@lists.linux.dev, linux-arm-kernel@lists.infradead.org, jstultz@google.com, qperret@google.com, will@kernel.org, kernel-team@android.com, linux-kernel@vger.kernel.org, Vincent Donnefort , Thomas Gleixner , Stephen Boyd , "Christopher S. Hall" , Richard Cochran Configure the hypervisor tracing clock before starting tracing. For tracing purpose, the boot clock is interesting as it doesn't stop on suspend. However, it is corrected on a regular basis, which implies we need to re-evaluate it every once in a while. Cc: John Stultz Cc: Thomas Gleixner Cc: Stephen Boyd Cc: Christopher S. Hall Cc: Richard Cochran Signed-off-by: Vincent Donnefort diff --git a/arch/arm64/include/asm/kvm_asm.h b/arch/arm64/include/asm/kvm_asm.h index 0cf171eacd9b..81bae121a4fe 100644 --- a/arch/arm64/include/asm/kvm_asm.h +++ b/arch/arm64/include/asm/kvm_asm.h @@ -79,6 +79,7 @@ enum __kvm_host_smccc_func { __KVM_HOST_SMCCC_FUNC___pkvm_init_vm, __KVM_HOST_SMCCC_FUNC___pkvm_init_vcpu, __KVM_HOST_SMCCC_FUNC___pkvm_teardown_vm, + __KVM_HOST_SMCCC_FUNC___pkvm_update_clock_tracing, __KVM_HOST_SMCCC_FUNC___pkvm_load_tracing, __KVM_HOST_SMCCC_FUNC___pkvm_teardown_tracing, __KVM_HOST_SMCCC_FUNC___pkvm_enable_tracing, diff --git a/arch/arm64/kvm/hyp/include/nvhe/trace.h b/arch/arm64/kvm/hyp/include/nvhe/trace.h index a7c0c73af0e5..df17683a3b12 100644 --- a/arch/arm64/kvm/hyp/include/nvhe/trace.h +++ b/arch/arm64/kvm/hyp/include/nvhe/trace.h @@ -16,6 +16,7 @@ struct hyp_buffer_page { void *tracing_reserve_entry(unsigned long length); void tracing_commit_entry(void); +void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc); int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size); void __pkvm_teardown_tracing(void); int __pkvm_enable_tracing(bool enable); @@ -24,6 +25,8 @@ int __pkvm_swap_reader_tracing(unsigned int cpu); static inline void *tracing_reserve_entry(unsigned long length) { return NULL; } static inline void tracing_commit_entry(void) { } +static inline +void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc) { } static inline int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size) { return -ENODEV; } static inline void __pkvm_teardown_tracing(void) { } static inline int __pkvm_enable_tracing(bool enable) { return -ENODEV; } diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c index 35d3228632e7..89a76af342c8 100644 --- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c +++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c @@ -379,6 +379,18 @@ static void handle___pkvm_teardown_vm(struct kvm_cpu_context *host_ctxt) cpu_reg(host_ctxt, 1) = __pkvm_teardown_vm(handle); } +static void handle___pkvm_update_clock_tracing(struct kvm_cpu_context *host_ctxt) +{ + DECLARE_REG(u32, mult, host_ctxt, 1); + DECLARE_REG(u32, shift, host_ctxt, 2); + DECLARE_REG(u64, epoch_ns, host_ctxt, 3); + DECLARE_REG(u64, epoch_cyc, host_ctxt, 4); + + __pkvm_update_clock_tracing(mult, shift, epoch_ns, epoch_cyc); + + cpu_reg(host_ctxt, 1) = 0; +} + static void handle___pkvm_load_tracing(struct kvm_cpu_context *host_ctxt) { DECLARE_REG(unsigned long, desc_hva, host_ctxt, 1); @@ -439,6 +451,7 @@ static const hcall_t host_hcall[] = { HANDLE_FUNC(__pkvm_init_vm), HANDLE_FUNC(__pkvm_init_vcpu), HANDLE_FUNC(__pkvm_teardown_vm), + HANDLE_FUNC(__pkvm_update_clock_tracing), HANDLE_FUNC(__pkvm_load_tracing), HANDLE_FUNC(__pkvm_teardown_tracing), HANDLE_FUNC(__pkvm_enable_tracing), diff --git a/arch/arm64/kvm/hyp/nvhe/trace.c b/arch/arm64/kvm/hyp/nvhe/trace.c index debb3ee7dd3a..022fe2e24f82 100644 --- a/arch/arm64/kvm/hyp/nvhe/trace.c +++ b/arch/arm64/kvm/hyp/nvhe/trace.c @@ -468,6 +468,21 @@ static void rb_teardown_bpage_backing(void) hyp_buffer_pages_backing.size = 0; } +void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc) +{ + int cpu; + + /* After this loop, all CPUs are observing the new bank... */ + for (cpu = 0; cpu < hyp_nr_cpus; cpu++) { + struct hyp_rb_per_cpu *cpu_buffer = per_cpu_ptr(&trace_rb, cpu); + + while (atomic_read(&cpu_buffer->status) == HYP_RB_WRITING); + } + + /* ...we can now override the old one and swap. */ + trace_clock_update(mult, shift, epoch_ns, epoch_cyc); +} + int __pkvm_swap_reader_tracing(unsigned int cpu) { struct hyp_rb_per_cpu *cpu_buffer; diff --git a/arch/arm64/kvm/hyp_trace.c b/arch/arm64/kvm/hyp_trace.c index be5bc602abc4..78193dbd8a33 100644 --- a/arch/arm64/kvm/hyp_trace.c +++ b/arch/arm64/kvm/hyp_trace.c @@ -16,10 +16,33 @@ #define RB_POLL_MS 100 +/* Same 10min used by clocksource when width is more than 32-bits */ +#define CLOCK_MAX_CONVERSION_S 600 +/* + * Time to give for the clock init. Long enough to get a good mult/shift + * estimation. Short enough to not delay the tracing start too much. + */ +#define CLOCK_INIT_MS 100 +/* + * Time between clock checks. Must be small enough to catch clock deviation when + * it is still tiny. + */ +#define CLOCK_UPDATE_MS 500 + #define TRACEFS_DIR "hypervisor" #define TRACEFS_MODE_WRITE 0640 #define TRACEFS_MODE_READ 0440 +struct hyp_trace_clock { + u64 cycles; + u64 cyc_overflow64; + u64 boot; + u32 mult; + u32 shift; + struct delayed_work work; + struct completion ready; +}; + static struct hyp_trace_buffer { struct hyp_trace_desc *desc; struct ring_buffer_remote remote; @@ -28,6 +51,7 @@ static struct hyp_trace_buffer { bool tracing_on; int nr_readers; struct mutex lock; + struct hyp_trace_clock clock; } hyp_trace_buffer = { .lock = __MUTEX_INITIALIZER(hyp_trace_buffer.lock), }; @@ -74,6 +98,109 @@ bpage_backing_free(struct hyp_buffer_pages_backing *bpage_backing) free_pages_exact((void *)bpage_backing->start, bpage_backing->size); } +static void __hyp_clock_work(struct work_struct *work) +{ + struct delayed_work *dwork = to_delayed_work(work); + struct hyp_trace_buffer *hyp_buffer; + struct hyp_trace_clock *hyp_clock; + struct system_time_snapshot snap; + u64 rate, delta_cycles; + u64 boot, delta_boot; + + hyp_clock = container_of(dwork, struct hyp_trace_clock, work); + hyp_buffer = container_of(hyp_clock, struct hyp_trace_buffer, clock); + + ktime_get_snapshot(&snap); + boot = ktime_to_ns(snap.boot); + + delta_boot = boot - hyp_clock->boot; + delta_cycles = snap.cycles - hyp_clock->cycles; + + /* Compare hyp clock with the kernel boot clock */ + if (hyp_clock->mult) { + u64 err, cur = delta_cycles; + + if (WARN_ON_ONCE(cur >= hyp_clock->cyc_overflow64)) { + __uint128_t tmp = (__uint128_t)cur * hyp_clock->mult; + + cur = tmp >> hyp_clock->shift; + } else { + cur *= hyp_clock->mult; + cur >>= hyp_clock->shift; + } + cur += hyp_clock->boot; + + err = abs_diff(cur, boot); + /* No deviation, only update epoch if necessary */ + if (!err) { + if (delta_cycles >= (hyp_clock->cyc_overflow64 >> 1)) + goto fast_forward; + + goto resched; + } + + /* Warn if the error is above tracing precision (1us) */ + if (hyp_buffer->tracing_on && err > NSEC_PER_USEC) + pr_warn_ratelimited("hyp trace clock off by %lluus\n", + err / NSEC_PER_USEC); + } + + if (delta_boot > U32_MAX) { + do_div(delta_boot, NSEC_PER_SEC); + rate = delta_cycles; + } else { + rate = delta_cycles * NSEC_PER_SEC; + } + rate = div64_u64(rate, delta_boot); + + clocks_calc_mult_shift(&hyp_clock->mult, &hyp_clock->shift, + rate, NSEC_PER_SEC, CLOCK_MAX_CONVERSION_S); + + /* Add a comfortable 50% margin */ + hyp_clock->cyc_overflow64 = (U64_MAX / hyp_clock->mult) >> 1; + +fast_forward: + hyp_clock->cycles = snap.cycles; + hyp_clock->boot = boot; + kvm_call_hyp_nvhe(__pkvm_update_clock_tracing, hyp_clock->mult, + hyp_clock->shift, hyp_clock->boot, hyp_clock->cycles); + complete(&hyp_clock->ready); + +resched: + schedule_delayed_work(&hyp_clock->work, + msecs_to_jiffies(CLOCK_UPDATE_MS)); +} + +static void hyp_clock_start(struct hyp_trace_buffer *hyp_buffer) +{ + struct hyp_trace_clock *hyp_clock = &hyp_buffer->clock; + struct system_time_snapshot snap; + + ktime_get_snapshot(&snap); + + hyp_clock->boot = ktime_to_ns(snap.boot); + hyp_clock->cycles = snap.cycles; + hyp_clock->mult = 0; + + init_completion(&hyp_clock->ready); + INIT_DELAYED_WORK(&hyp_clock->work, __hyp_clock_work); + schedule_delayed_work(&hyp_clock->work, msecs_to_jiffies(CLOCK_INIT_MS)); +} + +static void hyp_clock_stop(struct hyp_trace_buffer *hyp_buffer) +{ + struct hyp_trace_clock *hyp_clock = &hyp_buffer->clock; + + cancel_delayed_work_sync(&hyp_clock->work); +} + +static void hyp_clock_wait(struct hyp_trace_buffer *hyp_buffer) +{ + struct hyp_trace_clock *hyp_clock = &hyp_buffer->clock; + + wait_for_completion(&hyp_clock->ready); +} + static int __get_reader_page(int cpu) { return kvm_call_hyp_nvhe(__pkvm_swap_reader_tracing, cpu); @@ -294,10 +421,14 @@ static int hyp_trace_start(void) if (hyp_buffer->tracing_on) goto out; + hyp_clock_start(hyp_buffer); + ret = hyp_trace_buffer_load(hyp_buffer, hyp_trace_buffer_size); if (ret) goto out; + hyp_clock_wait(hyp_buffer); + ret = kvm_call_hyp_nvhe(__pkvm_enable_tracing, true); if (ret) { hyp_trace_buffer_teardown(hyp_buffer); @@ -307,6 +438,9 @@ static int hyp_trace_start(void) hyp_buffer->tracing_on = true; out: + if (!hyp_buffer->tracing_on) + hyp_clock_stop(hyp_buffer); + mutex_unlock(&hyp_buffer->lock); return ret; @@ -326,6 +460,7 @@ static void hyp_trace_stop(void) if (!ret) { ring_buffer_poll_remote(hyp_buffer->trace_buffer, RING_BUFFER_ALL_CPUS); + hyp_clock_stop(hyp_buffer); hyp_buffer->tracing_on = false; hyp_trace_buffer_teardown(hyp_buffer); } @@ -614,6 +749,14 @@ static const struct file_operations hyp_trace_pipe_fops = { .release = hyp_trace_pipe_release, }; +static int hyp_trace_clock_show(struct seq_file *m, void *v) +{ + seq_puts(m, "[boot]\n"); + + return 0; +} +DEFINE_SHOW_ATTRIBUTE(hyp_trace_clock); + int hyp_trace_init_tracefs(void) { struct dentry *root, *per_cpu_root; @@ -638,6 +781,9 @@ int hyp_trace_init_tracefs(void) tracefs_create_file("trace_pipe", TRACEFS_MODE_WRITE, root, (void *)RING_BUFFER_ALL_CPUS, &hyp_trace_pipe_fops); + tracefs_create_file("trace_clock", TRACEFS_MODE_READ, root, NULL, + &hyp_trace_clock_fops); + per_cpu_root = tracefs_create_dir("per_cpu", root); if (!per_cpu_root) { pr_err("Failed to create tracefs folder "TRACEFS_DIR"/per_cpu/\n");