| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20240328114004.5793230a@gandalf.local.home>
Date: Thu, 28 Mar 2024 11:40:04 -0400
From: Steven Rostedt <rostedt@...dmis.org>
To: Yi-De Wu <yi-de.wu@...iatek.com>
Cc: Yingshiuan Pan <yingshiuan.pan@...iatek.com>, Ze-Yu Wang
<ze-yu.wang@...iatek.com>, Rob Herring <robh+dt@...nel.org>, Krzysztof
Kozlowski <krzysztof.kozlowski+dt@...aro.org>, Conor Dooley
<conor+dt@...nel.org>, Jonathan Corbet <corbet@....net>, Catalin Marinas
<catalin.marinas@....com>, Wihl Deacon <will@...nel.org>, "Masami
Hiramatsu" <mhiramat@...nel.org>, Mathieu Desnoyers
<mathieu.desnoyers@...icios.com>, Richard Cochran
<richardcochran@...il.com>, Matthias Brugger <matthias.bgg@...il.com>,
AngeloGioacchino Del Regno <angelogioacchino.delregno@...labora.com>,
<devicetree@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
<linux-doc@...r.kernel.org>, <linux-arm-kernel@...ts.infradead.org>,
<linux-trace-kernel@...r.kernel.org>, <netdev@...r.kernel.org>,
<linux-mediatek@...ts.infradead.org>, David Bradil <dbrazdil@...gle.com>,
Trilok Soni <quic_tsoni@...cinc.com>, Jade Shih <jades.shih@...iatek.com>,
Ivan Tseng <ivan.tseng@...iatek.com>, My Chuang <my.chuang@...iatek.com>,
Shawn Hsiao <shawn.hsiao@...iatek.com>, PeiLun Suei
<peilun.suei@...iatek.com>, Liju Chen <liju-clr.chen@...iatek.com>, "Willix
Yeh" <chi-shen.yeh@...iatek.com>, Kevenny Hsieh
<kevenny.hsieh@...iatek.com>
Subject: Re: [PATCH v8 19/20] virt: geniezone: Add tracing support for hyp
call and vcpu exit_reason
On Thu, 28 Dec 2023 18:51:46 +0800
Yi-De Wu <yi-de.wu@...iatek.com> wrote:
> Add tracepoints for hypervisor calls and VCPU exit reasons in GenieZone
> driver. It aids performance debugging by providing more information
> about hypervisor operations and VCPU behavior.
>
> Command Usage:
> echo geniezone:* >> /sys/kernel/tracing/set_event
> echo 1 > /sys/kernel/tracing/tracing_on
> echo 0 > /sys/kernel/tracing/tracing_on
> cat /sys/kernel/tracing/trace
>
> For example:
> crosvm_vcpu0-4838 [004] ..... 76053.536034: mtk_hypcall_enter: id=0xbb001005
> crosvm_vcpu0-4838 [004] ..... 76053.540039: mtk_hypcall_leave: id=0xbb001005 invalid=0
> crosvm_vcpu0-4838 [004] ..... 76053.540040: mtk_vcpu_exit: vcpu exit_reason=0x92920003
Cleaning out patchwork, I noticed this patch.
You can make the above more informative by having it output:
crosvm_vcpu0-4838 [004] ..... 76053.540040: mtk_vcpu_exit: vcpu exit_reason=IRQ
>
> This example tracks a hypervisor function call by an ID (`0xbb001005`)
> from initiation to termination, which is supported (invalid=0). A vCPU
> exit is triggered by an Interrupt Request (IRQ) (exit reason: 0x92920003).
>
> /* VM exit reason */
> enum {
> GZVM_EXIT_UNKNOWN = 0x92920000,
> GZVM_EXIT_MMIO = 0x92920001,
> GZVM_EXIT_HYPERCALL = 0x92920002,
> GZVM_EXIT_IRQ = 0x92920003,
> GZVM_EXIT_EXCEPTION = 0x92920004,
> GZVM_EXIT_DEBUG = 0x92920005,
> GZVM_EXIT_FAIL_ENTRY = 0x92920006,
> GZVM_EXIT_INTERNAL_ERROR = 0x92920007,
> GZVM_EXIT_SYSTEM_EVENT = 0x92920008,
> GZVM_EXIT_SHUTDOWN = 0x92920009,
> GZVM_EXIT_GZ = 0x9292000a,
> };
>
> Signed-off-by: Liju-clr Chen <liju-clr.chen@...iatek.com>
> Signed-off-by: Yi-De Wu <yi-de.wu@...iatek.com>
> ---
> arch/arm64/geniezone/vm.c | 5 +++
> drivers/virt/geniezone/gzvm_vcpu.c | 3 ++
> include/trace/events/geniezone.h | 54 ++++++++++++++++++++++++++++++
> 3 files changed, 62 insertions(+)
> create mode 100644 include/trace/events/geniezone.h
>
> diff --git a/arch/arm64/geniezone/vm.c b/arch/arm64/geniezone/vm.c
> index a9d264bbb3b1..5667643251b5 100644
> --- a/arch/arm64/geniezone/vm.c
> +++ b/arch/arm64/geniezone/vm.c
> @@ -7,6 +7,8 @@
> #include <linux/err.h>
> #include <linux/uaccess.h>
>
> +#define CREATE_TRACE_POINTS
> +#include <trace/events/geniezone.h>
> #include <linux/gzvm.h>
> #include <linux/gzvm_drv.h>
> #include "gzvm_arch_common.h"
> @@ -33,7 +35,10 @@ int gzvm_hypcall_wrapper(unsigned long a0, unsigned long a1,
> unsigned long a6, unsigned long a7,
> struct arm_smccc_res *res)
> {
> + trace_mtk_hypcall_enter(a0);
> arm_smccc_hvc(a0, a1, a2, a3, a4, a5, a6, a7, res);
> + trace_mtk_hypcall_leave(a0, (res->a0 != ERR_NOT_SUPPORTED) ? 0 : 1);
> +
> return gzvm_err_to_errno(res->a0);
> }
>
> diff --git a/drivers/virt/geniezone/gzvm_vcpu.c b/drivers/virt/geniezone/gzvm_vcpu.c
> index 86c690749277..138ec064596b 100644
> --- a/drivers/virt/geniezone/gzvm_vcpu.c
> +++ b/drivers/virt/geniezone/gzvm_vcpu.c
> @@ -10,6 +10,8 @@
> #include <linux/mm.h>
> #include <linux/platform_device.h>
> #include <linux/slab.h>
> +
> +#include <trace/events/geniezone.h>
> #include <linux/gzvm_drv.h>
>
> /* maximum size needed for holding an integer */
> @@ -103,6 +105,7 @@ static long gzvm_vcpu_run(struct gzvm_vcpu *vcpu, void __user *argp)
>
> while (!need_userspace && !signal_pending(current)) {
> gzvm_arch_vcpu_run(vcpu, &exit_reason);
> + trace_mtk_vcpu_exit(exit_reason);
>
> switch (exit_reason) {
> case GZVM_EXIT_MMIO:
> diff --git a/include/trace/events/geniezone.h b/include/trace/events/geniezone.h
> new file mode 100644
> index 000000000000..1fa44f9c4b3c
> --- /dev/null
> +++ b/include/trace/events/geniezone.h
> @@ -0,0 +1,54 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2023 MediaTek Inc.
> + */
> +
> +#undef TRACE_SYSTEM
> +#define TRACE_SYSTEM geniezone
> +
> +#define _TRACE_GENIEZONE_H
> +
> +#include <linux/tracepoint.h>
#define GZVM_EXIT_REASONS \
EM(UNKNOWN) \
EM(MMIO) \
EM(HYPERCALL) \
EM(IRQ) \
EM(EXCEPTION) \
EM(DEBUG) \
EM(FAIL_ENTRY) \
EM(INTERNAL_ERROR) \
EM(SYSTEM_EVENT) \
EM(SHUTDOWN) \
EMe(GZ)
#undef EM
#undef EMe
#define EM(a) TRACE_DEFINE_ENUM(GZVM_EXIT_##a);
#define EMe(a) TRACE_DEFINE_ENUM(GZVM_EXIT_##a);
GZVM_EXIT_REASONS
#undef EM
#undef EMe
#define EM(a) { GZVM_EXIT_##a, #a },
#define EMe(a) { GZVM_EXIT_##a, #a }
> +
> +TRACE_EVENT(mtk_hypcall_enter,
> + TP_PROTO(unsigned long id),
> +
> + TP_ARGS(id),
> +
> + TP_STRUCT__entry(__field(unsigned long, id)),
> +
> + TP_fast_assign(__entry->id = id;),
> +
> + TP_printk("id=0x%lx", __entry->id)
> +);
> +
> +TRACE_EVENT(mtk_hypcall_leave,
> + TP_PROTO(unsigned long id, unsigned long invalid),
> +
> + TP_ARGS(id, invalid),
> +
> + TP_STRUCT__entry(__field(unsigned long, id)
> + __field(unsigned long, invalid)
> + ),
> +
> + TP_fast_assign(__entry->id = id;
> + __entry->invalid = invalid;
> + ),
> +
> + TP_printk("id=0x%lx invalid=%lu", __entry->id, __entry->invalid)
> +);
> +
> +TRACE_EVENT(mtk_vcpu_exit,
> + TP_PROTO(unsigned long exit_reason),
> +
> + TP_ARGS(exit_reason),
> +
> + TP_STRUCT__entry(__field(unsigned long, exit_reason)),
> +
> + TP_fast_assign(__entry->exit_reason = exit_reason;),
> +
> + TP_printk("vcpu exit_reason=0x%lx", __entry->exit_reason)
TP_printk("vcpu exit_reason=0x%lx",
__print_symbolic(__entry->exit_reason, GZVM_EXIT_REASONS))
And instead of having the cryptic enum values printed, you will have human
readable reasons.
-- Steve
> +);
> +
> +/* This part must be outside protection */
> +#include <trace/define_trace.h>
Powered by blists - more mailing lists