Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Subject: Re: [PATCH V2 2/2] drivers/perf: hisi: add driver for HNS3 PMU
To:     John Garry <john.garry@huawei.com>, <will@kernel.org>,
        <mark.rutland@arm.com>
CC:     <linux-arm-kernel@lists.infradead.org>,
        <linux-kernel@vger.kernel.org>, <linuxarm@huawei.com>,
        <liuqi115@huawei.com>, <zhangshaokun@hisilicon.com>,
        <f.fangjian@huawei.com>, <lipeng321@huawei.com>,
        <shenjian15@huawei.com>, <moyufeng@huawei.com>
References: <20220228123955.30284-1-huangguangbin2@huawei.com>
 <20220228123955.30284-3-huangguangbin2@huawei.com>
 <6089561f-039b-43b5-7a6b-43b00ce370f2@huawei.com>
From:   "huangguangbin (A)" <huangguangbin2@huawei.com>
Message-ID: <5b77fceb-830f-2592-351e-4185fc8843f0@huawei.com>
Date:   Fri, 25 Mar 2022 20:08:36 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:68.0) Gecko/20100101
 Thunderbird/68.2.2
MIME-Version: 1.0
In-Reply-To: <6089561f-039b-43b5-7a6b-43b00ce370f2@huawei.com>
Content-Type: text/plain; charset="utf-8"; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk


On 2022/3/3 0:10, John Garry wrote:
> On 28/02/2022 12:39, Guangbin Huang wrote:
>> HNS3 PMU End Point device supports to collect performance statistics
>> of bandwidth, latency, packet rate, interrupt rate in HiSilicon SoC
>> NIC.
>>
>> NIC of each IO DIE has one PMU device for it. Driver registers each
>> PMU device to perf, and exports information of supported events,
>> filter mode of each event, identifier and so on via sysfs.
>>
>> Each PMU device has its own control, counter and interrupt registers,
>> and supports up to 8 events by hardware.
>>
> 
> this is looking before, but still comments...
> 
Ok，I will modify it.

>> Filter options contains:
>> config       - select event
>> port         - select physical port of nic
>> tc           - select tc(must be used with port)
>> func         - select PF/VF
>> queue        - select queue of PF/VF(must be used with func)
>> intr         - select interrupt number(must be used with func)
>> global       - select all functions of IO DIE
>>
>> Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
>> ---
>>   MAINTAINERS                       |    6 +
>>   drivers/perf/hisilicon/Kconfig    |   10 +
>>   drivers/perf/hisilicon/Makefile   |    1 +
>>   drivers/perf/hisilicon/hns3_pmu.c | 1632 +++++++++++++++++++++++++++++
>>   include/linux/cpuhotplug.h        |    1 +
>>   5 files changed, 1650 insertions(+)
>>   create mode 100644 drivers/perf/hisilicon/hns3_pmu.c
>>
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index 69a2935daf6c..34b87348503a 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -8691,6 +8691,12 @@ F:    Documentation/admin-guide/perf/hisi-pcie-pmu.rst
>>   F:    Documentation/admin-guide/perf/hisi-pmu.rst
>>   F:    drivers/perf/hisilicon
>> +HISILICON HNS3 PMU DRIVER
>> +M:    Guangbin Huang <huangguangbin2@huawei.com>
>> +S:    Supported
>> +F:    Documentation/admin-guide/perf/hns3-pmu.rst
>> +F:    drivers/perf/hisilicon/hns3_pmu.c
>> +
>>   HISILICON QM AND ZIP Controller DRIVER
>>   M:    Zhou Wang <wangzhou1@hisilicon.com>
>>   L:    linux-crypto@vger.kernel.org
>> diff --git a/drivers/perf/hisilicon/Kconfig b/drivers/perf/hisilicon/Kconfig
>> index 5546218b5598..fea5d884ccb5 100644
>> --- a/drivers/perf/hisilicon/Kconfig
>> +++ b/drivers/perf/hisilicon/Kconfig
>> @@ -14,3 +14,13 @@ config HISI_PCIE_PMU
>>         RCiEP devices.
>>         Adds the PCIe PMU into perf events system for monitoring latency,
>>         bandwidth etc.
>> +
>> +config HNS3_PMU
>> +    tristate "HNS3 PERF PMU"
>> +    depends on ARM64 || COMPILE_TEST
>> +    depends on PCI
>> +    help
>> +      Provide support for HNS3 performance monitoring unit (PMU) IEP
> 
> like PCIe PMU, is this RCiEP?
Yes. I will add it on next version.

> 
>> +      devices.
>> +      Adds the HNS3 PMU into perf events system for monitoring latency,
>> +      bandwidth etc.
>> diff --git a/drivers/perf/hisilicon/Makefile b/drivers/perf/hisilicon/Makefile
>> index 506ed39e3266..13297ec2798f 100644
>> --- a/drivers/perf/hisilicon/Makefile
>> +++ b/drivers/perf/hisilicon/Makefile
>> @@ -4,3 +4,4 @@ obj-$(CONFIG_HISI_PMU) += hisi_uncore_pmu.o hisi_uncore_l3c_pmu.o \
>>                 hisi_uncore_pa_pmu.o
>>   obj-$(CONFIG_HISI_PCIE_PMU) += hisi_pcie_pmu.o
>> +obj-$(CONFIG_HNS3_PMU) += hns3_pmu.o
>> diff --git a/drivers/perf/hisilicon/hns3_pmu.c b/drivers/perf/hisilicon/hns3_pmu.c
>> new file mode 100644
>> index 000000000000..733b36f0191b
>> --- /dev/null
>> +++ b/drivers/perf/hisilicon/hns3_pmu.c
>> @@ -0,0 +1,1632 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>> +/*
>> + * This driver adds support for HNS3 PMU iEP device. Related perf events are
>> + * bandwidth, latency, packet rate, interrupt rate etc.
>> + *
>> + * Copyright (C) 2022 HiSilicon Limited
>> + */
>> +#include <linux/bitfield.h>
>> +#include <linux/bitmap.h>
>> +#include <linux/bug.h>
>> +#include <linux/cpuhotplug.h>
>> +#include <linux/cpumask.h>
>> +#include <linux/delay.h>
>> +#include <linux/device.h>
>> +#include <linux/err.h>
>> +#include <linux/interrupt.h>
>> +#include <linux/iopoll.h>
>> +#include <linux/irq.h>
>> +#include <linux/kernel.h>
>> +#include <linux/list.h>
>> +#include <linux/module.h>
>> +#include <linux/pci.h>
>> +#include <linux/perf_event.h>
>> +#include <linux/smp.h>
>> +
>> +/* registers offset address */
>> +#define HNS3_PMU_REG_GLOBAL_CTRL        0x0000
>> +#define HNS3_PMU_REG_CLOCK_FREQ            0x0020
>> +#define HNS3_PMU_REG_BDF            0x0FE0
> 
> nit: generally people prefer lowercase for hex numbers
> 
Ok.

>> +#define HNS3_PMU_REG_VERSION            0x0FE4
>> +#define HNS3_PMU_REG_DEVICE_ID            0x0FE8
>> +
>> +#define HNS3_PMU_REG_EVENT_OFFSET        0x1000
>> +#define HNS3_PMU_REG_EVENT_SIZE            0x1000
>> +#define HNS3_PMU_REG_EVENT_CTRL_LOW        0x00
>> +#define HNS3_PMU_REG_EVENT_CTRL_HIGH        0x04
>> +#define HNS3_PMU_REG_EVENT_INTR_STATUS        0x08
>> +#define HNS3_PMU_REG_EVENT_INTR_MASK        0x0C
>> +#define HNS3_PMU_REG_EVENT_COUNTER        0x10
>> +#define HNS3_PMU_REG_EVENT_EXT_COUNTER        0x18
>> +#define HNS3_PMU_REG_EVENT_QID_CTRL        0x28
>> +#define HNS3_PMU_REG_EVENT_QID_PARA        0x2C
>> +
>> +#define HNS3_PMU_FILTER_SUPPORT_GLOBAL        BIT(0)
>> +#define HNS3_PMU_FILTER_SUPPORT_PORT        BIT(1)
>> +#define HNS3_PMU_FILTER_SUPPORT_PORT_TC        BIT(2)
>> +#define HNS3_PMU_FILTER_SUPPORT_FUNC        BIT(3)
>> +#define HNS3_PMU_FILTER_SUPPORT_FUNC_QUEUE    BIT(4)
>> +#define HNS3_PMU_FILTER_SUPPORT_FUNC_INTR    BIT(5)
>> +
>> +#define HNS3_PMU_FILTER_ALL_TC            0xF
>> +#define HNS3_PMU_FILTER_ALL_QUEUE        0xFFFF
>> +
>> +#define HNS3_PMU_CTRL_SUBEVENT_S        4
>> +#define HNS3_PMU_CTRL_FILTER_MODE_S        24
>> +
>> +#define HNS3_PMU_GLOBAL_START            BIT(0)
>> +
>> +#define HNS3_PMU_EVENT_STATUS_RESET        BIT(11)
>> +#define HNS3_PMU_EVENT_EN            BIT(12)
>> +#define HNS3_PMU_EVENT_OVERFLOW_RESTART        BIT(15)
>> +
>> +#define HNS3_PMU_QID_PARA_FUNC_S        0
>> +#define HNS3_PMU_QID_PARA_QUEUE_S        16
>> +
>> +#define HNS3_PMU_QID_CTRL_REQ_ENABLE        BIT(0)
>> +#define HNS3_PMU_QID_CTRL_DONE            BIT(1)
>> +#define HNS3_PMU_QID_CTRL_MISS            BIT(2)
>> +
>> +#define HNS3_PMU_INTR_MASK_OVERFLOW        BIT(1)
>> +
>> +#define HNS3_PMU_MAX_HW_EVENTS            8
>> +
>> +/*
>> + * Each hardware event contains two registers(counter and ext_counter) for
> 
> /s/registers(counter/registers (counter/
> 
Ok.

>> + * bandwidth, packet rate, latency and interrupt rate. These two registers will
>> + * be triggered to run at the same when a hardware event is enabled.
>> + *
>> + * Performance of each hardware event is calculated by: counter / ext_counter.
>> + *
>> + * As process of performance data is not recommended put in driver, we expose
>> + * ext_counter as a separate event for userspace and use bit 16 to indicate it.
>> + * For example, event 0x00001 and 0x10001 are actually one event for hardware
>> + * because bit 0-15 are same. If the bit 16 of one event is 0 means to get
>> + * counter, otherwise means to get ext_counter.
>> + */
>> +/* bandwidth events */
>> +#define HNS3_PMU_EVT_BW_SSU_EGU_BYTE_NUM        0x00001
>> +#define HNS3_PMU_EVT_BW_SSU_EGU_TIME            0x10001
>> +#define HNS3_PMU_EVT_BW_SSU_RPU_BYTE_NUM        0x00002
>> +#define HNS3_PMU_EVT_BW_SSU_RPU_TIME            0x10002
>> +#define HNS3_PMU_EVT_BW_SSU_ROCE_BYTE_NUM        0x00003
>> +#define HNS3_PMU_EVT_BW_SSU_ROCE_TIME            0x10003
>> +#define HNS3_PMU_EVT_BW_ROCE_SSU_BYTE_NUM        0x00004
>> +#define HNS3_PMU_EVT_BW_ROCE_SSU_TIME            0x10004
>> +#define HNS3_PMU_EVT_BW_TPU_SSU_BYTE_NUM        0x00005
>> +#define HNS3_PMU_EVT_BW_TPU_SSU_TIME            0x10005
>> +#define HNS3_PMU_EVT_BW_RPU_RCBRX_BYTE_NUM        0x00006
>> +#define HNS3_PMU_EVT_BW_RPU_RCBRX_TIME            0x10006
>> +#define HNS3_PMU_EVT_BW_RCBTX_TXSCH_BYTE_NUM        0x00008
>> +#define HNS3_PMU_EVT_BW_RCBTX_TXSCH_TIME        0x10008
>> +#define HNS3_PMU_EVT_BW_WR_FBD_BYTE_NUM            0x00009
>> +#define HNS3_PMU_EVT_BW_WR_FBD_TIME            0x10009
>> +#define HNS3_PMU_EVT_BW_WR_EBD_BYTE_NUM            0x0000A
>> +#define HNS3_PMU_EVT_BW_WR_EBD_TIME            0x1000A
>> +#define HNS3_PMU_EVT_BW_RD_FBD_BYTE_NUM            0x0000B
>> +#define HNS3_PMU_EVT_BW_RD_FBD_TIME            0x1000B
>> +#define HNS3_PMU_EVT_BW_RD_EBD_BYTE_NUM            0x0000C
>> +#define HNS3_PMU_EVT_BW_RD_EBD_TIME            0x1000C
>> +#define HNS3_PMU_EVT_BW_RD_PAY_M0_BYTE_NUM        0x0000D
>> +#define HNS3_PMU_EVT_BW_RD_PAY_M0_TIME            0x1000D
>> +#define HNS3_PMU_EVT_BW_RD_PAY_M1_BYTE_NUM        0x0000E
>> +#define HNS3_PMU_EVT_BW_RD_PAY_M1_TIME            0x1000E
>> +#define HNS3_PMU_EVT_BW_WR_PAY_M0_BYTE_NUM        0x0000F
>> +#define HNS3_PMU_EVT_BW_WR_PAY_M0_TIME            0x1000F
>> +#define HNS3_PMU_EVT_BW_WR_PAY_M1_BYTE_NUM        0x00010
>> +#define HNS3_PMU_EVT_BW_WR_PAY_M1_TIME            0x10010
>> +
>> +/* packet rate events */
>> +#define HNS3_PMU_EVT_PPS_IGU_SSU_PACKET_NUM        0x00100
>> +#define HNS3_PMU_EVT_PPS_IGU_SSU_TIME            0x10100
>> +#define HNS3_PMU_EVT_PPS_SSU_EGU_PACKET_NUM        0x00101
>> +#define HNS3_PMU_EVT_PPS_SSU_EGU_TIME            0x10101
>> +#define HNS3_PMU_EVT_PPS_SSU_RPU_PACKET_NUM        0x00102
>> +#define HNS3_PMU_EVT_PPS_SSU_RPU_TIME            0x10102
>> +#define HNS3_PMU_EVT_PPS_SSU_ROCE_PACKET_NUM        0x00103
>> +#define HNS3_PMU_EVT_PPS_SSU_ROCE_TIME            0x10103
>> +#define HNS3_PMU_EVT_PPS_ROCE_SSU_PACKET_NUM        0x00104
>> +#define HNS3_PMU_EVT_PPS_ROCE_SSU_TIME            0x10104
>> +#define HNS3_PMU_EVT_PPS_TPU_SSU_PACKET_NUM        0x00105
>> +#define HNS3_PMU_EVT_PPS_TPU_SSU_TIME            0x10105
>> +#define HNS3_PMU_EVT_PPS_RPU_RCBRX_PACKET_NUM        0x00106
>> +#define HNS3_PMU_EVT_PPS_RPU_RCBRX_TIME            0x10106
>> +#define HNS3_PMU_EVT_PPS_RCBTX_TPU_PACKET_NUM        0x00107
>> +#define HNS3_PMU_EVT_PPS_RCBTX_TPU_TIME            0x10107
>> +#define HNS3_PMU_EVT_PPS_RCBTX_TXSCH_PACKET_NUM        0x00108
>> +#define HNS3_PMU_EVT_PPS_RCBTX_TXSCH_TIME        0x10108
>> +#define HNS3_PMU_EVT_PPS_WR_FBD_PACKET_NUM        0x00109
>> +#define HNS3_PMU_EVT_PPS_WR_FBD_TIME            0x10109
>> +#define HNS3_PMU_EVT_PPS_WR_EBD_PACKET_NUM        0x0010A
>> +#define HNS3_PMU_EVT_PPS_WR_EBD_TIME            0x1010A
>> +#define HNS3_PMU_EVT_PPS_RD_FBD_PACKET_NUM        0x0010B
>> +#define HNS3_PMU_EVT_PPS_RD_FBD_TIME            0x1010B
>> +#define HNS3_PMU_EVT_PPS_RD_EBD_PACKET_NUM        0x0010C
>> +#define HNS3_PMU_EVT_PPS_RD_EBD_TIME            0x1010C
>> +#define HNS3_PMU_EVT_PPS_RD_PAY_M0_PACKET_NUM        0x0010D
>> +#define HNS3_PMU_EVT_PPS_RD_PAY_M0_TIME            0x1010D
>> +#define HNS3_PMU_EVT_PPS_RD_PAY_M1_PACKET_NUM        0x0010E
>> +#define HNS3_PMU_EVT_PPS_RD_PAY_M1_TIME            0x1010E
>> +#define HNS3_PMU_EVT_PPS_WR_PAY_M0_PACKET_NUM        0x0010F
>> +#define HNS3_PMU_EVT_PPS_WR_PAY_M0_TIME            0x1010F
>> +#define HNS3_PMU_EVT_PPS_WR_PAY_M1_PACKET_NUM        0x00110
>> +#define HNS3_PMU_EVT_PPS_WR_PAY_M1_TIME            0x10110
>> +#define HNS3_PMU_EVT_PPS_NICROH_TX_PRE_PACKET_NUM    0x00111
>> +#define HNS3_PMU_EVT_PPS_NICROH_TX_PRE_TIME        0x10111
>> +#define HNS3_PMU_EVT_PPS_NICROH_RX_PRE_PACKET_NUM    0x00112
>> +#define HNS3_PMU_EVT_PPS_NICROH_RX_PRE_TIME        0x10112
>> +
>> +/* latency events */
>> +#define HNS3_PMU_EVT_DLY_TX_PUSH_TIME            0x00202
>> +#define HNS3_PMU_EVT_DLY_TX_PUSH_PACKET_NUM        0x10202
>> +#define HNS3_PMU_EVT_DLY_TX_TIME            0x00204
>> +#define HNS3_PMU_EVT_DLY_TX_PACKET_NUM            0x10204
>> +#define HNS3_PMU_EVT_DLY_SSU_TX_NIC_TIME        0x00206
>> +#define HNS3_PMU_EVT_DLY_SSU_TX_NIC_PACKET_NUM        0x10206
>> +#define HNS3_PMU_EVT_DLY_SSU_TX_ROCE_TIME        0x00207
>> +#define HNS3_PMU_EVT_DLY_SSU_TX_ROCE_PACKET_NUM        0x10207
>> +#define HNS3_PMU_EVT_DLY_SSU_RX_NIC_TIME        0x00208
>> +#define HNS3_PMU_EVT_DLY_SSU_RX_NIC_PACKET_NUM        0x10208
>> +#define HNS3_PMU_EVT_DLY_SSU_RX_ROCE_TIME        0x00209
>> +#define HNS3_PMU_EVT_DLY_SSU_RX_ROCE_PACKET_NUM        0x10209
>> +#define HNS3_PMU_EVT_DLY_RPU_TIME            0x0020E
>> +#define HNS3_PMU_EVT_DLY_RPU_PACKET_NUM            0x1020E
>> +#define HNS3_PMU_EVT_DLY_TPU_TIME            0x0020F
>> +#define HNS3_PMU_EVT_DLY_TPU_PACKET_NUM            0x1020F
>> +#define HNS3_PMU_EVT_DLY_RPE_TIME            0x00210
>> +#define HNS3_PMU_EVT_DLY_RPE_PACKET_NUM            0x10210
>> +#define HNS3_PMU_EVT_DLY_TPE_TIME            0x00211
>> +#define HNS3_PMU_EVT_DLY_TPE_PACKET_NUM            0x10211
>> +#define HNS3_PMU_EVT_DLY_TPE_PUSH_TIME            0x00212
>> +#define HNS3_PMU_EVT_DLY_TPE_PUSH_PACKET_NUM        0x10212
>> +#define HNS3_PMU_EVT_DLY_WR_FBD_TIME            0x00213
>> +#define HNS3_PMU_EVT_DLY_WR_FBD_PACKET_NUM        0x10213
>> +#define HNS3_PMU_EVT_DLY_WR_EBD_TIME            0x00214
>> +#define HNS3_PMU_EVT_DLY_WR_EBD_PACKET_NUM        0x10214
>> +#define HNS3_PMU_EVT_DLY_RD_FBD_TIME            0x00215
>> +#define HNS3_PMU_EVT_DLY_RD_FBD_PACKET_NUM        0x10215
>> +#define HNS3_PMU_EVT_DLY_RD_EBD_TIME            0x00216
>> +#define HNS3_PMU_EVT_DLY_RD_EBD_PACKET_NUM        0x10216
>> +#define HNS3_PMU_EVT_DLY_RD_PAY_M0_TIME            0x00217
>> +#define HNS3_PMU_EVT_DLY_RD_PAY_M0_PACKET_NUM        0x10217
>> +#define HNS3_PMU_EVT_DLY_RD_PAY_M1_TIME            0x00218
>> +#define HNS3_PMU_EVT_DLY_RD_PAY_M1_PACKET_NUM        0x10218
>> +#define HNS3_PMU_EVT_DLY_WR_PAY_M0_TIME            0x00219
>> +#define HNS3_PMU_EVT_DLY_WR_PAY_M0_PACKET_NUM        0x10219
>> +#define HNS3_PMU_EVT_DLY_WR_PAY_M1_TIME            0x0021A
>> +#define HNS3_PMU_EVT_DLY_WR_PAY_M1_PACKET_NUM        0x1021A
>> +#define HNS3_PMU_EVT_DLY_MSIX_WRITE_TIME        0x0021C
>> +#define HNS3_PMU_EVT_DLY_MSIX_WRITE_PACKET_NUM        0x1021C
>> +
>> +/* interrupt rate events */
>> +#define HNS3_PMU_EVT_PPS_MSIX_NIC_INTR_NUM        0x00300
>> +#define HNS3_PMU_EVT_PPS_MSIX_NIC_TIME            0x10300
>> +
>> +/* filter mode supported by each bandwidth event */
>> +#define HNS3_PMU_FILTER_BW_SSU_EGU        0x07
>> +#define HNS3_PMU_FILTER_BW_SSU_RPU        0x1F
>> +#define HNS3_PMU_FILTER_BW_SSU_ROCE        0x0F
>> +#define HNS3_PMU_FILTER_BW_ROCE_SSU        0x0F
>> +#define HNS3_PMU_FILTER_BW_TPU_SSU        0x1F
>> +#define HNS3_PMU_FILTER_BW_RPU_RCBRX        0x11
>> +#define HNS3_PMU_FILTER_BW_RCBTX_TXSCH        0x11
>> +#define HNS3_PMU_FILTER_BW_WR_FBD        0x1B
>> +#define HNS3_PMU_FILTER_BW_WR_EBD        0x11
>> +#define HNS3_PMU_FILTER_BW_RD_FBD        0x01
>> +#define HNS3_PMU_FILTER_BW_RD_EBD        0x1B
>> +#define HNS3_PMU_FILTER_BW_RD_PAY_M0        0x01
>> +#define HNS3_PMU_FILTER_BW_RD_PAY_M1        0x01
>> +#define HNS3_PMU_FILTER_BW_WR_PAY_M0        0x01
>> +#define HNS3_PMU_FILTER_BW_WR_PAY_M1        0x01
>> +
>> +/* filter mode supported by each packet rate event */
>> +#define HNS3_PMU_FILTER_PPS_IGU_SSU        0x07
>> +#define HNS3_PMU_FILTER_PPS_SSU_EGU        0x07
>> +#define HNS3_PMU_FILTER_PPS_SSU_RPU        0x1F
>> +#define HNS3_PMU_FILTER_PPS_SSU_ROCE        0x0F
>> +#define HNS3_PMU_FILTER_PPS_ROCE_SSU        0x0F
>> +#define HNS3_PMU_FILTER_PPS_TPU_SSU        0x1F
>> +#define HNS3_PMU_FILTER_PPS_RPU_RCBRX        0x11
>> +#define HNS3_PMU_FILTER_PPS_RCBTX_TPU        0x1F
>> +#define HNS3_PMU_FILTER_PPS_RCBTX_TXSCH        0x11
>> +#define HNS3_PMU_FILTER_PPS_WR_FBD        0x1B
>> +#define HNS3_PMU_FILTER_PPS_WR_EBD        0x11
>> +#define HNS3_PMU_FILTER_PPS_RD_FBD        0x01
>> +#define HNS3_PMU_FILTER_PPS_RD_EBD        0x1B
>> +#define HNS3_PMU_FILTER_PPS_RD_PAY_M0        0x01
>> +#define HNS3_PMU_FILTER_PPS_RD_PAY_M1        0x01
>> +#define HNS3_PMU_FILTER_PPS_WR_PAY_M0        0x01
>> +#define HNS3_PMU_FILTER_PPS_WR_PAY_M1        0x01
>> +#define HNS3_PMU_FILTER_PPS_NICROH_TX_PRE    0x01
>> +#define HNS3_PMU_FILTER_PPS_NICROH_RX_PRE    0x01
>> +
>> +/* filter mode supported by each latency event */
>> +#define HNS3_PMU_FILTER_DLY_TX_PUSH        0x01
>> +#define HNS3_PMU_FILTER_DLY_TX            0x01
>> +#define HNS3_PMU_FILTER_DLY_SSU_TX_NIC        0x07
>> +#define HNS3_PMU_FILTER_DLY_SSU_TX_ROCE        0x07
>> +#define HNS3_PMU_FILTER_DLY_SSU_RX_NIC        0x07
>> +#define HNS3_PMU_FILTER_DLY_SSU_RX_ROCE        0x07
>> +#define HNS3_PMU_FILTER_DLY_RPU            0x11
>> +#define HNS3_PMU_FILTER_DLY_TPU            0x1F
>> +#define HNS3_PMU_FILTER_DLY_RPE            0x01
>> +#define HNS3_PMU_FILTER_DLY_TPE            0x0B
>> +#define HNS3_PMU_FILTER_DLY_TPE_PUSH        0x1B
>> +#define HNS3_PMU_FILTER_DLY_WR_FBD        0x1B
>> +#define HNS3_PMU_FILTER_DLY_WR_EBD        0x11
>> +#define HNS3_PMU_FILTER_DLY_RD_FBD        0x01
>> +#define HNS3_PMU_FILTER_DLY_RD_EBD        0x1B
>> +#define HNS3_PMU_FILTER_DLY_RD_PAY_M0        0x01
>> +#define HNS3_PMU_FILTER_DLY_RD_PAY_M1        0x01
>> +#define HNS3_PMU_FILTER_DLY_WR_PAY_M0        0x01
>> +#define HNS3_PMU_FILTER_DLY_WR_PAY_M1        0x01
>> +#define HNS3_PMU_FILTER_DLY_MSIX_WRITE        0x01
>> +
> 
> Can we describe all these events just in userspace now?
> 
I don't think so, driver needs to check whether filter mode is right
of each event, so driver should record and show what filter modes are
supported of each event.

>> +/* filter mode supported by each interrupt rate event */
>> +#define HNS3_PMU_FILTER_INTR_MSIX_NIC        0x01
>> +
>> +enum hns3_pmu_hw_filter_mode {
>> +    HNS3_PMU_HW_FILTER_GLOBAL,
>> +    HNS3_PMU_HW_FILTER_PORT,
>> +    HNS3_PMU_HW_FILTER_PORT_TC,
>> +    HNS3_PMU_HW_FILTER_FUNC,
>> +    HNS3_PMU_HW_FILTER_FUNC_QUEUE,
>> +    HNS3_PMU_HW_FILTER_FUNC_INTR,
>> +};
>> +
>> +struct hns3_pmu_event_attr {
>> +    u32 event;
>> +    u16 filter_support;
>> +};
>> +
>> +struct hns3_pmu {
>> +    struct perf_event *hw_events[HNS3_PMU_MAX_HW_EVENTS];
>> +    struct hlist_node node;
>> +    struct pci_dev *pdev;
>> +    struct pmu pmu;
>> +    void __iomem *base;
>> +    int irq;
>> +    int on_cpu;
>> +    u32 identifier;
>> +    u32 hw_clk_freq; /* hardware clock frequency of PMU */
>> +    /* maximum and minimun bdf allowed by PMU */
>> +    u16 bdf_min;
>> +    u16 bdf_max;
>> +};
>> +
>> +#define to_hns3_pmu(p)  (container_of((p), struct hns3_pmu, pmu))
>> +#define attr_to_dattr(a) (container_of((a), struct device_attribute, attr))
>> +#define dattr_to_eattr(d) (container_of((d), struct dev_ext_attribute, attr))
>> +
>> +#define GET_PCI_DEVFN(bdf)  ((bdf) & 0xFF)
>> +
>> +#define FILTER_CONDITION_PORT(port) ((1 << (port)) & 0xFF)
>> +#define FILTER_CONDITION_PORT_TC(port, tc) (((port) << 3) | ((tc) & 0x07))
>> +#define FILTER_CONDITION_FUNC_INTR(func, intr) (((intr) << 8) | (func))
>> +
>> +#define BYTES_TO_BITS(bytes)        ((bytes) * 8)
>> +
>> +#define HNS3_PMU_FILTER_ATTR(_name, _config, _start, _end)            \
>> +    static inline u64 hns3_get_##_name(struct perf_event *event)  \
>> +    {                                                             \
>> +        return FIELD_GET(GENMASK_ULL(_end, _start),           \
>> +                 event->attr._config);                \
>> +    }
>> +
>> +HNS3_PMU_FILTER_ATTR(event, config, 0, 16);
>> +HNS3_PMU_FILTER_ATTR(subevent, config, 0, 7);
>> +HNS3_PMU_FILTER_ATTR(event_type, config, 8, 15);
>> +HNS3_PMU_FILTER_ATTR(ext_counter_used, config, 16, 16);
>> +HNS3_PMU_FILTER_ATTR(real_event, config, 0, 15);
>> +HNS3_PMU_FILTER_ATTR(port, config1, 0, 3);
>> +HNS3_PMU_FILTER_ATTR(tc, config1, 4, 7);
>> +HNS3_PMU_FILTER_ATTR(bdf, config1, 8, 23);
>> +HNS3_PMU_FILTER_ATTR(queue, config1, 24, 39);
>> +HNS3_PMU_FILTER_ATTR(intr, config1, 40, 51);
>> +HNS3_PMU_FILTER_ATTR(global, config1, 52, 52);
>> +
>> +#define HNS3_BW_EVT_BYTE_NUM(_name)    (&(struct hns3_pmu_event_attr) {\
>> +    HNS3_PMU_EVT_BW_##_name##_BYTE_NUM,                \
>> +    HNS3_PMU_FILTER_BW_##_name})
>> +#define HNS3_BW_EVT_TIME(_name)        (&(struct hns3_pmu_event_attr) {\
>> +    HNS3_PMU_EVT_BW_##_name##_TIME,                    \
>> +    HNS3_PMU_FILTER_BW_##_name})
>> +#define HNS3_PPS_EVT_PACKET_NUM(_name)    (&(struct hns3_pmu_event_attr) {\
>> +    HNS3_PMU_EVT_PPS_##_name##_PACKET_NUM,                \
>> +    HNS3_PMU_FILTER_PPS_##_name})
>> +#define HNS3_PPS_EVT_TIME(_name)    (&(struct hns3_pmu_event_attr) {\
>> +    HNS3_PMU_EVT_PPS_##_name##_TIME,                \
>> +    HNS3_PMU_FILTER_PPS_##_name})
>> +#define HNS3_DLY_EVT_TIME(_name)    (&(struct hns3_pmu_event_attr) {\
>> +    HNS3_PMU_EVT_DLY_##_name##_TIME,                \
>> +    HNS3_PMU_FILTER_DLY_##_name})
>> +#define HNS3_DLY_EVT_PACKET_NUM(_name)    (&(struct hns3_pmu_event_attr) {\
>> +    HNS3_PMU_EVT_DLY_##_name##_PACKET_NUM,                \
>> +    HNS3_PMU_FILTER_DLY_##_name})
>> +#define HNS3_INTR_EVT_INTR_NUM(_name)    (&(struct hns3_pmu_event_attr) {\
>> +    HNS3_PMU_EVT_PPS_##_name##_INTR_NUM,                \
>> +    HNS3_PMU_FILTER_INTR_##_name})
>> +#define HNS3_INTR_EVT_TIME(_name)    (&(struct hns3_pmu_event_attr) {\
>> +    HNS3_PMU_EVT_PPS_##_name##_TIME,                \
>> +    HNS3_PMU_FILTER_INTR_##_name})
>> +
>> +static ssize_t hns3_pmu_format_show(struct device *dev,
>> +                    struct device_attribute *attr, char *buf)
>> +{
>> +    struct dev_ext_attribute *eattr;
>> +
>> +    eattr = container_of(attr, struct dev_ext_attribute, attr);
>> +
>> +    return sysfs_emit(buf, "%s\n", (char *)eattr->var);
>> +}
>> +
>> +static ssize_t hns3_pmu_event_show(struct device *dev,
>> +                   struct device_attribute *attr, char *buf)
>> +{
>> +    struct hns3_pmu_event_attr *event;
>> +    struct dev_ext_attribute *eattr;
>> +
>> +    eattr = container_of(attr, struct dev_ext_attribute, attr);
>> +    event = (struct hns3_pmu_event_attr *)eattr->var;
>> +
>> +    return sysfs_emit(buf, "config=0x%05x\n", event->event);
>> +}
>> +
>> +static ssize_t hns3_pmu_filter_mode_show(struct device *dev,
>> +                     struct device_attribute *attr,
>> +                     char *buf)
>> +{
>> +    struct hns3_pmu_event_attr *event;
>> +    struct dev_ext_attribute *eattr;
>> +    int len;
>> +
>> +    eattr = container_of(attr, struct dev_ext_attribute, attr);
>> +    event = (struct hns3_pmu_event_attr *)eattr->var;
>> +
>> +    len = sysfs_emit_at(buf, 0, "filter mode supported: ");
>> +    if (event->filter_support & HNS3_PMU_FILTER_SUPPORT_GLOBAL)
>> +        len += sysfs_emit_at(buf, len, "global/");
>> +    if (event->filter_support & HNS3_PMU_FILTER_SUPPORT_PORT)
>> +        len += sysfs_emit_at(buf, len, "port/");
>> +    if (event->filter_support & HNS3_PMU_FILTER_SUPPORT_PORT_TC)
>> +        len += sysfs_emit_at(buf, len, "port-tc/");
>> +    if (event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC)
>> +        len += sysfs_emit_at(buf, len, "func/");
>> +    if (event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC_QUEUE)
>> +        len += sysfs_emit_at(buf, len, "func-queue/");
>> +    if (event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC_INTR)
>> +        len += sysfs_emit_at(buf, len, "func-intr/");
> 
> eh, is this final "/" intended?
> 
Here use "/" to separate each type, may be it is bette to use a space to separate them,
I will modify to space.

>> +
>> +    len += sysfs_emit_at(buf, len, "\n");
>> +
>> +    return len;
>> +}
>> +
>> +#define HNS3_PMU_ATTR(_name, _func, _config)                \
>> +    (&((struct dev_ext_attribute[]) {                \
>> +        { __ATTR(_name, 0444, _func, NULL), (void *)_config }    \
>> +    })[0].attr.attr)
>> +
>> +#define HNS3_PMU_FORMAT_ATTR(_name, _format) \
>> +    HNS3_PMU_ATTR(_name, hns3_pmu_format_show, (void *)_format)
>> +#define HNS3_PMU_EVENT_ATTR(_name, _event) \
>> +    HNS3_PMU_ATTR(_name, hns3_pmu_event_show, (void *)_event)
>> +
>> +#define HNS3_PMU_BW_EVT_PAIR(_name, _macro) \
>> +    HNS3_PMU_EVENT_ATTR(_name##_byte_num, HNS3_BW_EVT_BYTE_NUM(_macro)), \
>> +    HNS3_PMU_EVENT_ATTR(_name##_time, HNS3_BW_EVT_TIME(_macro))
>> +#define HNS3_PMU_PPS_EVT_PAIR(_name, _macro) \
>> +    HNS3_PMU_EVENT_ATTR(_name##_packet_num, HNS3_PPS_EVT_PACKET_NUM(_macro)), \
>> +    HNS3_PMU_EVENT_ATTR(_name##_time, HNS3_PPS_EVT_TIME(_macro))
>> +#define HNS3_PMU_DLY_EVT_PAIR(_name, _macro) \
>> +    HNS3_PMU_EVENT_ATTR(_name##_time, HNS3_DLY_EVT_TIME(_macro)), \
>> +    HNS3_PMU_EVENT_ATTR(_name##_packet_num, HNS3_DLY_EVT_PACKET_NUM(_macro))
>> +#define HNS3_PMU_INTR_EVT_PAIR(_name, _macro) \
>> +    HNS3_PMU_EVENT_ATTR(_name##_intr_num, HNS3_INTR_EVT_INTR_NUM(_macro)), \
>> +    HNS3_PMU_EVENT_ATTR(_name##_time, HNS3_INTR_EVT_TIME(_macro))
>> +
>> +static u8 hns3_pmu_hw_filter_modes[] = {
>> +    HNS3_PMU_HW_FILTER_GLOBAL,
>> +    HNS3_PMU_HW_FILTER_PORT,
>> +    HNS3_PMU_HW_FILTER_PORT_TC,
>> +    HNS3_PMU_HW_FILTER_FUNC,
>> +    HNS3_PMU_HW_FILTER_FUNC_QUEUE,
>> +    HNS3_PMU_HW_FILTER_FUNC_INTR
> 
> why no ','?
> 
I think it doesn't matter, I will add it.

>> +};
>> +
>> +#define HNS3_PMU_SET_HW_FILTER(_hwc, _mode) \
>> +    ((_hwc)->addr_filters = (void *)&hns3_pmu_hw_filter_modes[(_mode)])
>> +
>> +static ssize_t identifier_show(struct device *dev,
>> +                   struct device_attribute *attr, char *buf)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(dev_get_drvdata(dev));
>> +
>> +    return sysfs_emit(buf, "0x%x\n", hns3_pmu->identifier);
>> +}
>> +static DEVICE_ATTR_RO(identifier);
>> +
>> +static ssize_t cpumask_show(struct device *dev, struct device_attribute *attr,
>> +                char *buf)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(dev_get_drvdata(dev));
>> +
>> +    return sysfs_emit(buf, "%d\n", hns3_pmu->on_cpu);
>> +}
>> +static DEVICE_ATTR_RO(cpumask);
>> +
>> +static ssize_t bdf_min_show(struct device *dev, struct device_attribute *attr,
>> +                char *buf)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(dev_get_drvdata(dev));
>> +
>> +    return sysfs_emit(buf, "0x%4x\n", hns3_pmu->bdf_min);
>> +}
>> +static DEVICE_ATTR_RO(bdf_min);
>> +
>> +static ssize_t bdf_max_show(struct device *dev, struct device_attribute *attr,
>> +                char *buf)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(dev_get_drvdata(dev));
>> +
>> +    return sysfs_emit(buf, "0x%4x\n", hns3_pmu->bdf_max);
>> +}
>> +static DEVICE_ATTR_RO(bdf_max);
>> +
>> +static ssize_t
>> +hw_clk_freq_show(struct device *dev, struct device_attribute *attr, char *buf)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(dev_get_drvdata(dev));
>> +
>> +    return sysfs_emit(buf, "%u\n", hns3_pmu->hw_clk_freq);
>> +}
>> +static DEVICE_ATTR_RO(hw_clk_freq);
>> +
> 
> ...
> 
>> +
>> +static int hns3_pmu_init_filter_mode_group(void)
>> +{
>> +    u32 group_size = ARRAY_SIZE(hns3_pmu_events_attr);
>> +    struct dev_ext_attribute *eattr_filter;
>> +    struct dev_ext_attribute *eattr_event;
>> +    struct device_attribute *dattr_event;
>> +    struct attribute **attr_filter;
>> +    struct attribute *attr_event;
>> +    u32 i;
>> +
>> +    attr_filter = kcalloc(group_size, sizeof(struct attribute *),
>> +                  GFP_KERNEL);
>> +    if (!attr_filter)
>> +        return -ENOMEM;
>> +
>> +    eattr_filter = kcalloc(group_size - 1, sizeof(struct dev_ext_attribute),
>> +                   GFP_KERNEL);
>> +    if (!eattr_filter) {
> 
> I might have said this before - I'm not sure why these need to be dynamically allocated. All the sizes seem fixed at compile time.
> 
Yes, the sizes is fixed. I just don't want to define a new array as big as hns3_pmu_events_attr[], and the only difference
between the new array and array hns3_pmu_events_attr[] is the attr.show() callback of each event.

However, if all of you don't dynamically allocation, I will modify it on next version.

>> +        kfree(attr_filter);
>> +        return -ENOMEM;
>> +    }
>> +
>> +    for (i = 0; i < group_size - 1; i++) {
>> +        attr_event = hns3_pmu_events_attr[i];
>> +        dattr_event = attr_to_dattr(attr_event);
>> +        eattr_event = dattr_to_eattr(dattr_event);
>> +
>> +        memcpy(&eattr_filter[i], eattr_event,
>> +               sizeof(struct dev_ext_attribute));
>> +        eattr_filter[i].attr.show = hns3_pmu_filter_mode_show;
>> +
>> +        attr_filter[i] = &eattr_filter[i].attr.attr;
>> +    }
>> +
>> +    hns3_pmu_filter_mode_group.attrs = attr_filter;
>> +
>> +    return 0;
>> +}
>> +
>> +static void hns3_pmu_uinit_filter_mode_group(void)
>> +{
>> +    struct dev_ext_attribute *eattr;
>> +    struct device_attribute *dattr;
>> +    struct attribute **attrs;
>> +
>> +    attrs = hns3_pmu_filter_mode_group.attrs;
>> +    dattr = attr_to_dattr(attrs[0]);
>> +    eattr = dattr_to_eattr(dattr);
>> +
>> +    kfree(eattr);
>> +    kfree(attrs);
>> +
>> +    hns3_pmu_filter_mode_group.attrs = NULL;
>> +}
>> +
>> +static u32 hns3_pmu_get_offset(u32 offset, u32 idx)
>> +{
>> +    return offset + HNS3_PMU_REG_EVENT_OFFSET +
>> +           HNS3_PMU_REG_EVENT_SIZE * idx;
>> +}
>> +
>> +static u32 hns3_pmu_readl(struct hns3_pmu *hns3_pmu, u32 reg_offset, u32 idx)
>> +{
>> +    u32 offset = hns3_pmu_get_offset(reg_offset, idx);
>> +
>> +    return readl(hns3_pmu->base + offset);
>> +}
>> +
>> +static void hns3_pmu_writel(struct hns3_pmu *hns3_pmu, u32 reg_offset, u32 idx,
>> +                u32 val)
>> +{
>> +    u32 offset = hns3_pmu_get_offset(reg_offset, idx);
>> +
>> +    writel(val, hns3_pmu->base + offset);
>> +}
>> +
>> +static u64 hns3_pmu_readq(struct hns3_pmu *hns3_pmu, u32 reg_offset, u32 idx)
>> +{
>> +    u32 offset = hns3_pmu_get_offset(reg_offset, idx);
>> +
>> +    return readq(hns3_pmu->base + offset);
>> +}
>> +
>> +static void hns3_pmu_writeq(struct hns3_pmu *hns3_pmu, u32 reg_offset, u32 idx,
>> +                u64 val)
>> +{
>> +    u32 offset = hns3_pmu_get_offset(reg_offset, idx);
>> +
>> +    writeq(val, hns3_pmu->base + offset);
>> +}
>> +
>> +static bool hns3_pmu_cmp_event(struct perf_event *target,
>> +                   struct perf_event *event)
>> +{
>> +    return hns3_get_real_event(target) == hns3_get_real_event(event);
>> +}
>> +
>> +static int hns3_pmu_find_related_event(struct hns3_pmu *hns3_pmu,
>> +                       struct perf_event *event)
>> +{
>> +    struct perf_event *sibling;
>> +    int idx;
>> +
>> +    for (idx = 0; idx < HNS3_PMU_MAX_HW_EVENTS; idx++) {
>> +        sibling = hns3_pmu->hw_events[idx];
>> +        if (!sibling)
>> +            continue;
>> +
>> +        if (!hns3_pmu_cmp_event(sibling, event))
>> +            continue;
>> +
>> +        /* Related events must be used in group */
>> +        if (sibling->group_leader == event->group_leader)
>> +            return idx;
>> +        else
>> +            return -EINVAL;
>> +    }
>> +
>> +    return idx;
>> +}
>> +
>> +static int hns3_pmu_get_event_idx(struct hns3_pmu *hns3_pmu)
>> +{
>> +    int idx;
>> +
>> +    for (idx = 0; idx < HNS3_PMU_MAX_HW_EVENTS; idx++)
>> +        if (!hns3_pmu->hw_events[idx])
>> +            return idx;
> 
> nit: for multi-line statements it's better to use {} around them
> 
Ok.

>> +
>> +    return -EBUSY;
>> +}
>> +
>> +static bool hns3_pmu_valid_bdf(struct hns3_pmu *hns3_pmu, u16 bdf)
>> +{
>> +    struct pci_dev *pdev;
>> +
>> +    if (bdf < hns3_pmu->bdf_min || bdf > hns3_pmu->bdf_max) {
> 
> why not inclusive? If I say that something is the minimum, it means inclusive. By definition, that is what minimum means.
> 
>> +        pci_err(hns3_pmu->pdev, "Invalid EP device: %#x!\n", bdf);
>> +        return false;
>> +    }
>> +
>> +    pdev = pci_get_domain_bus_and_slot(pci_domain_nr(hns3_pmu->pdev->bus),
>> +                       PCI_BUS_NUM(bdf),
>> +                       GET_PCI_DEVFN(bdf));
>> +    if (!pdev) {
>> +        pci_err(hns3_pmu->pdev, "Nonexistent EP device: %#x!\n", bdf);
>> +        return false;
>> +    }
>> +
>> +    pci_dev_put(pdev);
>> +    return true;
>> +}
>> +
>> +static void hns3_pmu_set_qid_para(struct hns3_pmu *hns3_pmu, u32 idx, u16 bdf,
>> +                  u16 queue)
>> +{
>> +    u32 val;
>> +
>> +    val = GET_PCI_DEVFN(bdf);
>> +    val |= (u32)queue << HNS3_PMU_QID_PARA_QUEUE_S;
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_QID_PARA, idx, val);
>> +}
>> +
>> +static bool hns3_pmu_qid_req_start(struct hns3_pmu *hns3_pmu, u32 idx)
>> +{
>> +    bool queue_id_valid = false;
>> +    u32 reg_qid_ctrl, val;
>> +    int err;
>> +
>> +    /* enable queue id request */
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_QID_CTRL, idx,
>> +            HNS3_PMU_QID_CTRL_REQ_ENABLE);
>> +
>> +    reg_qid_ctrl = hns3_pmu_get_offset(HNS3_PMU_REG_EVENT_QID_CTRL, idx);
>> +    err = readl_poll_timeout(hns3_pmu->base + reg_qid_ctrl, val,
>> +                 val & HNS3_PMU_QID_CTRL_DONE, 1, 1000);
>> +    if (err == -ETIMEDOUT) {
>> +        pci_err(hns3_pmu->pdev, "QID request timeout!\n");
>> +        goto out;
>> +    }
>> +
>> +    queue_id_valid = (val & HNS3_PMU_QID_CTRL_MISS) == 0;
>> +
>> +out:
>> +    /* disable qid request and clear status */
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_QID_CTRL, idx, 0);
>> +
>> +    return queue_id_valid;
>> +}
>> +
>> +static bool hns3_pmu_valid_queue(struct hns3_pmu *hns3_pmu, u32 idx, u16 bdf,
>> +                 u16 queue)
>> +{
>> +    hns3_pmu_set_qid_para(hns3_pmu, idx, bdf, queue);
>> +
>> +    return hns3_pmu_qid_req_start(hns3_pmu, idx);
>> +}
>> +
>> +static struct hns3_pmu_event_attr *hns3_pmu_get_pmu_event(u32 event)
>> +{
>> +    struct hns3_pmu_event_attr *pmu_event;
>> +    struct dev_ext_attribute *eattr;
>> +    struct device_attribute *dattr;
>> +    struct attribute *attr;
>> +    u32 i;
>> +
>> +    for (i = 0; i < ARRAY_SIZE(hns3_pmu_events_attr) - 1; i++) {
>> +        attr = hns3_pmu_events_attr[i];
>> +        dattr = container_of(attr, struct device_attribute, attr);
>> +        eattr = container_of(dattr, struct dev_ext_attribute, attr);
>> +        pmu_event = (struct hns3_pmu_event_attr *)eattr->var;
>> +
>> +        if (event == pmu_event->event)
>> +            return pmu_event;
>> +    }
>> +
>> +    return NULL;
>> +}
>> +
>> +static int hns3_pmu_set_func_mode(struct perf_event *event,
>> +                  struct hns3_pmu *hns3_pmu)
>> +{
>> +    struct hw_perf_event *hwc = &event->hw;
>> +    u16 bdf = hns3_get_bdf(event);
>> +
>> +    if (!hns3_pmu_valid_bdf(hns3_pmu, bdf))
>> +        return -ENOENT;
>> +
>> +    HNS3_PMU_SET_HW_FILTER(hwc, HNS3_PMU_HW_FILTER_FUNC);
>> +
>> +    return 0;
>> +}
>> +
>> +static int hns3_pmu_set_func_queue_mode(struct perf_event *event,
>> +                    struct hns3_pmu *hns3_pmu)
>> +{
>> +    struct hw_perf_event *hwc = &event->hw;
>> +    u16 queue_id = hns3_get_queue(event);
>> +    u16 bdf = hns3_get_bdf(event);
>> +
>> +    if (!hns3_pmu_valid_bdf(hns3_pmu, bdf))
>> +        return -ENOENT;
>> +
>> +    if (!hns3_pmu_valid_queue(hns3_pmu, hwc->idx, bdf, queue_id)) {
>> +        pci_err(hns3_pmu->pdev, "Invalid queue: %u\n", queue_id);
>> +        return -ENOENT;
>> +    }
>> +
>> +    HNS3_PMU_SET_HW_FILTER(hwc, HNS3_PMU_HW_FILTER_FUNC_QUEUE);
>> +
>> +    return 0;
>> +}
>> +
>> +static bool
>> +hns3_pmu_is_enabled_global_mode(struct perf_event *event,
>> +                struct hns3_pmu_event_attr *pmu_event)
>> +{
>> +    u8 global = hns3_get_global(event);
>> +
>> +    return pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_GLOBAL &&
>> +           !!global;
>> +}
>> +
>> +static bool hns3_pmu_is_enabled_func_mode(struct perf_event *event,
>> +                      struct hns3_pmu_event_attr *pmu_event)
>> +{
>> +    u16 queue_id = hns3_get_queue(event);
>> +    u16 bdf = hns3_get_bdf(event);
>> +
>> +    return pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC &&
>> +           pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC && !!bdf;
> 
> This is hard to read. And pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC seems to be tested twice
> 
Ok.

>> +}
>> +
>> +static bool
>> +hns3_pmu_is_enabled_func_queue_mode(struct perf_event *event,
>> +                    struct hns3_pmu_event_attr *pmu_event)
>> +{
>> +    u16 queue_id = hns3_get_queue(event);
>> +    u16 bdf = hns3_get_bdf(event);
>> +
>> +    return pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC_QUEUE &&
>> +           queue_id != HNS3_PMU_FILTER_ALL_QUEUE && !!bdf;
> 
> again, hard to read.
> 
> maybe
> 
>      if (!(pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC))
>          return false;
>      if ( queue_id != HNS3_PMU_FILTER_ALL_QUEUE)
>          return false
>      return !bdf.
> 
> I think that you get the idea.
> 
Ok.

>> +}
>> +
>> +static bool hns3_pmu_is_enabled_port_mode(struct perf_event *event,
>> +                      struct hns3_pmu_event_attr *pmu_event)
>> +{
>> +    u8 tc_id = hns3_get_tc(event);
>> +
>> +    return pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_PORT &&
>> +           tc_id == HNS3_PMU_FILTER_ALL_TC;
>> +}
>> +
>> +static bool
>> +hns3_pmu_is_enabled_port_tc_mode(struct perf_event *event,
>> +                 struct hns3_pmu_event_attr *pmu_event)
>> +{
>> +    u8 tc_id = hns3_get_tc(event);
>> +
>> +    return pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_PORT_TC &&
>> +           tc_id != HNS3_PMU_FILTER_ALL_TC;
>> +}
>> +
>> +static bool
>> +hns3_pmu_is_enabled_func_intr_mode(struct perf_event *event,
>> +                   struct hns3_pmu *hns3_pmu,
>> +                   struct hns3_pmu_event_attr *pmu_event)
>> +{
>> +    u16 bdf = hns3_get_bdf(event);
>> +
>> +    return pmu_event->filter_support & HNS3_PMU_FILTER_SUPPORT_FUNC_INTR &&
>> +           hns3_pmu_valid_bdf(hns3_pmu, bdf);
>> +}
>> +
>> +static int hns3_pmu_select_filter_mode(struct perf_event *event,
>> +                       struct hns3_pmu *hns3_pmu)
>> +{
>> +    struct hns3_pmu_event_attr *pmu_event;
>> +    struct hw_perf_event *hwc = &event->hw;
>> +    u32 event_id = hns3_get_event(event);
>> +
>> +    pmu_event = hns3_pmu_get_pmu_event(event_id);
>> +    if (!pmu_event) {
>> +        pci_err(hns3_pmu->pdev, "Invalid pmu event\n");
>> +        return -ENOENT;
>> +    }
>> +
>> +    if (hns3_pmu_is_enabled_global_mode(event, pmu_event)) {
>> +        HNS3_PMU_SET_HW_FILTER(hwc, HNS3_PMU_HW_FILTER_GLOBAL);
>> +        return 0;
>> +    }
>> +
>> +    if (hns3_pmu_is_enabled_func_mode(event, pmu_event))
>> +        return hns3_pmu_set_func_mode(event, hns3_pmu);
>> +
>> +    if (hns3_pmu_is_enabled_func_queue_mode(event, pmu_event))
>> +        return hns3_pmu_set_func_queue_mode(event, hns3_pmu);
>> +
>> +    if (hns3_pmu_is_enabled_port_mode(event, pmu_event)) {
>> +        HNS3_PMU_SET_HW_FILTER(hwc, HNS3_PMU_HW_FILTER_PORT);
>> +        return 0;
>> +    }
>> +
>> +    if (hns3_pmu_is_enabled_port_tc_mode(event, pmu_event)) {
>> +        HNS3_PMU_SET_HW_FILTER(hwc, HNS3_PMU_HW_FILTER_PORT_TC);
>> +        return 0;
>> +    }
>> +
>> +    if (hns3_pmu_is_enabled_func_intr_mode(event, hns3_pmu, pmu_event)) {
>> +        HNS3_PMU_SET_HW_FILTER(hwc, HNS3_PMU_HW_FILTER_FUNC_INTR);
>> +        return 0;
>> +    }
>> +
>> +    return -ENOENT;
>> +}
>> +
>> +static bool hns3_pmu_validate_event_group(struct perf_event *event)
>> +{
>> +    struct perf_event *sibling, *leader = event->group_leader;
>> +    struct perf_event *event_group[HNS3_PMU_MAX_HW_EVENTS];
>> +    int counters = 1;
>> +    int num;
>> +
>> +    event_group[0] = leader;
>> +    if (!is_software_event(leader)) {
>> +        if (leader->pmu != event->pmu)
>> +            return false;
>> +
>> +        if (leader != event && !hns3_pmu_cmp_event(leader, event))
>> +            event_group[counters++] = event;
>> +    }
>> +
>> +    for_each_sibling_event(sibling, event->group_leader) {
>> +        if (is_software_event(sibling))
>> +            continue;
>> +
>> +        if (sibling->pmu != event->pmu)
>> +            return false;
>> +
>> +        for (num = 0; num < counters; num++) {
>> +            if (hns3_pmu_cmp_event(event_group[num], sibling))
>> +                break;
>> +        }
>> +
>> +        if (num == counters)
>> +            event_group[counters++] = sibling;
>> +    }
>> +
>> +    return counters <= HNS3_PMU_MAX_HW_EVENTS;
>> +}
>> +
>> +static u32 hns3_pmu_get_filter_condition(struct perf_event *event)
>> +{
>> +    struct hw_perf_event *hwc = &event->hw;
>> +    u16 intr_id = hns3_get_intr(event);
>> +    u8 port_id = hns3_get_port(event);
>> +    u16 bdf = hns3_get_bdf(event);
>> +    u8 tc_id = hns3_get_tc(event);
>> +    u8 filter_mode;
>> +    u32 filter = 0;
>> +
>> +    filter_mode = *(u8 *)hwc->addr_filters;
>> +    switch (filter_mode) {
>> +    case HNS3_PMU_HW_FILTER_PORT:
>> +        filter = FILTER_CONDITION_PORT(port_id);
>> +        break;
>> +    case HNS3_PMU_HW_FILTER_PORT_TC:
>> +        filter = FILTER_CONDITION_PORT_TC(port_id, tc_id);
>> +        break;
>> +    case HNS3_PMU_HW_FILTER_FUNC:
>> +    case HNS3_PMU_HW_FILTER_FUNC_QUEUE:
>> +        filter = GET_PCI_DEVFN(bdf);
>> +        break;
>> +    case HNS3_PMU_HW_FILTER_FUNC_INTR:
>> +        filter = FILTER_CONDITION_FUNC_INTR(GET_PCI_DEVFN(bdf),
>> +                            intr_id);
>> +        break;
>> +    default:
>> +        break;
>> +    }
>> +
>> +    return filter;
>> +}
>> +
>> +static void hns3_pmu_config_filter(struct perf_event *event)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(event->pmu);
>> +    u8 event_type = hns3_get_event_type(event);
>> +    u8 subevent_id = hns3_get_subevent(event);
>> +    struct hw_perf_event *hwc = &event->hw;
>> +    u8 filter_mode = *(u8 *)hwc->addr_filters;
>> +    u16 queue_id = hns3_get_queue(event);
>> +    u16 bdf = hns3_get_bdf(event);
>> +    u32 idx = hwc->idx;
>> +    u32 val;
>> +
>> +    val = event_type;
>> +    val |= subevent_id << HNS3_PMU_CTRL_SUBEVENT_S;
>> +    val |= filter_mode << HNS3_PMU_CTRL_FILTER_MODE_S;
>> +    val |= HNS3_PMU_EVENT_OVERFLOW_RESTART;
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx, val);
>> +
>> +    val = hns3_pmu_get_filter_condition(event);
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_HIGH, idx, val);
>> +
>> +    if (filter_mode == HNS3_PMU_HW_FILTER_FUNC_QUEUE)
>> +        hns3_pmu_set_qid_para(hns3_pmu, idx, bdf, queue_id);
>> +}
>> +
>> +static void hns3_pmu_enable_counter(struct hns3_pmu *hns3_pmu,
>> +                    struct hw_perf_event *hwc)
>> +{
>> +    u32 idx = hwc->idx;
>> +    u32 val;
>> +
>> +    val = hns3_pmu_readl(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx);
>> +    val |= HNS3_PMU_EVENT_EN;
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx, val);
>> +}
>> +
>> +static void hns3_pmu_disable_counter(struct hns3_pmu *hns3_pmu,
>> +                     struct hw_perf_event *hwc)
>> +{
>> +    u32 idx = hwc->idx;
>> +    u32 val;
>> +
>> +    val = hns3_pmu_readl(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx);
>> +    val &= ~HNS3_PMU_EVENT_EN;
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx, val);
>> +}
>> +
>> +static void hns3_pmu_enable_intr(struct hns3_pmu *hns3_pmu,
>> +                 struct hw_perf_event *hwc)
>> +{
>> +    u32 idx = hwc->idx;
>> +    u32 val;
>> +
>> +    val = hns3_pmu_readl(hns3_pmu, HNS3_PMU_REG_EVENT_INTR_MASK, idx);
>> +    val &= ~HNS3_PMU_INTR_MASK_OVERFLOW;
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_INTR_MASK, idx, val);
>> +}
>> +
>> +static void hns3_pmu_disable_intr(struct hns3_pmu *hns3_pmu,
>> +                  struct hw_perf_event *hwc)
>> +{
>> +    u32 idx = hwc->idx;
>> +    u32 val;
>> +
>> +    val = hns3_pmu_readl(hns3_pmu, HNS3_PMU_REG_EVENT_INTR_MASK, idx);
>> +    val |= HNS3_PMU_INTR_MASK_OVERFLOW;
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_INTR_MASK, idx, val);
>> +}
>> +
>> +static void hns3_pmu_clear_intr_status(struct hns3_pmu *hns3_pmu, u32 idx)
>> +{
>> +    u32 val;
>> +
>> +    val = hns3_pmu_readl(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx);
>> +    val |= HNS3_PMU_EVENT_STATUS_RESET;
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx, val);
>> +
>> +    val = hns3_pmu_readl(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx);
>> +    val &= ~HNS3_PMU_EVENT_STATUS_RESET;
>> +    hns3_pmu_writel(hns3_pmu, HNS3_PMU_REG_EVENT_CTRL_LOW, idx, val);
>> +}
>> +
>> +static u64 hns3_pmu_read_counter(struct perf_event *event)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(event->pmu);
>> +
>> +    return hns3_pmu_readq(hns3_pmu, event->hw.event_base, event->hw.idx);
>> +}
>> +
>> +static void hns3_pmu_write_counter(struct perf_event *event, u64 value)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(event->pmu);
>> +    u32 idx = event->hw.idx;
>> +
>> +    hns3_pmu_writeq(hns3_pmu, HNS3_PMU_REG_EVENT_COUNTER, idx, value);
>> +    hns3_pmu_writeq(hns3_pmu, HNS3_PMU_REG_EVENT_EXT_COUNTER, idx, value);
>> +}
>> +
>> +static void hns3_pmu_init_counter(struct perf_event *event)
>> +{
>> +    struct hw_perf_event *hwc = &event->hw;
>> +
>> +    local64_set(&hwc->prev_count, 0);
>> +    hns3_pmu_write_counter(event, 0);
>> +}
>> +
>> +static int hns3_pmu_event_init(struct perf_event *event)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(event->pmu);
>> +    struct hw_perf_event *hwc = &event->hw;
>> +    int idx;
>> +    int ret;
>> +
>> +    if (event->attr.type != event->pmu->type)
>> +        return -ENOENT;
>> +
>> +    /* Sampling is not supported */
>> +    if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
>> +        return -EOPNOTSUPP;
>> +
>> +    event->cpu = hns3_pmu->on_cpu;
>> +
>> +    idx = hns3_pmu_get_event_idx(hns3_pmu);
>> +    if (idx < 0) {
>> +        pci_err(hns3_pmu->pdev, "Up to %u events are supported!\n",
>> +            HNS3_PMU_MAX_HW_EVENTS);
>> +        return -EBUSY;
>> +    }
>> +
>> +    hwc->idx = idx;
>> +
>> +    ret = hns3_pmu_select_filter_mode(event, hns3_pmu);
>> +    if (ret) {
>> +        pci_err(hns3_pmu->pdev, "Invalid filter, ret = %d.\n", ret);
>> +        return ret;
>> +    }
>> +
>> +    if (!hns3_pmu_validate_event_group(event)) {
>> +        pci_err(hns3_pmu->pdev, "Invalid event group.\n");
>> +        return -EINVAL;
>> +    }
>> +
>> +    if (hns3_get_ext_counter_used(event))
>> +        hwc->event_base = HNS3_PMU_REG_EVENT_EXT_COUNTER;
>> +    else
>> +        hwc->event_base = HNS3_PMU_REG_EVENT_COUNTER;
>> +
>> +    return 0;
>> +}
>> +
>> +static void hns3_pmu_read(struct perf_event *event)
>> +{
>> +    struct hw_perf_event *hwc = &event->hw;
>> +    u64 new_cnt, prev_cnt, delta;
>> +
>> +    do {
>> +        prev_cnt = local64_read(&hwc->prev_count);
>> +        new_cnt = hns3_pmu_read_counter(event);
>> +    } while (local64_cmpxchg(&hwc->prev_count, prev_cnt, new_cnt) !=
>> +         prev_cnt);
>> +
>> +    delta = new_cnt - prev_cnt;
>> +    local64_add(delta, &event->count);
>> +}
>> +
>> +static void hns3_pmu_start(struct perf_event *event, int flags)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(event->pmu);
>> +    struct hw_perf_event *hwc = &event->hw;
>> +
>> +    if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
>> +        return;
>> +
>> +    WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
>> +    hwc->state = 0;
>> +
>> +    hns3_pmu_config_filter(event);
>> +    hns3_pmu_init_counter(event);
>> +    hns3_pmu_enable_intr(hns3_pmu, hwc);
>> +    hns3_pmu_enable_counter(hns3_pmu, hwc);
>> +
>> +    perf_event_update_userpage(event);
>> +}
>> +
>> +static void hns3_pmu_stop(struct perf_event *event, int flags)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(event->pmu);
>> +    struct hw_perf_event *hwc = &event->hw;
>> +
>> +    hns3_pmu_disable_counter(hns3_pmu, hwc);
>> +    hns3_pmu_disable_intr(hns3_pmu, hwc);
>> +
>> +    WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
>> +    hwc->state |= PERF_HES_STOPPED;
>> +
>> +    if (hwc->state & PERF_HES_UPTODATE)
>> +        return;
>> +
>> +    /* Read hardware counter and update the perf counter statistics */
>> +    hns3_pmu_read(event);
>> +    hwc->state |= PERF_HES_UPTODATE;
>> +}
>> +
>> +static int hns3_pmu_add(struct perf_event *event, int flags)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(event->pmu);
>> +    struct hw_perf_event *hwc = &event->hw;
>> +    int idx;
>> +
>> +    hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
>> +
>> +    /* Check all working events to find a related event. */
>> +    idx = hns3_pmu_find_related_event(hns3_pmu, event);
>> +    if (idx < 0)
>> +        return idx;
>> +
>> +    /* Current event shares an enabled hardware event with related event */
>> +    if (idx < HNS3_PMU_MAX_HW_EVENTS) {
>> +        hwc->idx = idx;
>> +        goto start_count;
>> +    }
>> +
>> +    idx = hns3_pmu_get_event_idx(hns3_pmu);
>> +    if (idx < 0)
>> +        return idx;
>> +
>> +    hwc->idx = idx;
>> +    hns3_pmu->hw_events[idx] = event;
>> +
>> +start_count:
>> +    if (flags & PERF_EF_START)
>> +        hns3_pmu_start(event, PERF_EF_RELOAD);
>> +
>> +    return 0;
>> +}
>> +
>> +static void hns3_pmu_del(struct perf_event *event, int flags)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(event->pmu);
>> +    struct hw_perf_event *hwc = &event->hw;
>> +
>> +    hns3_pmu_stop(event, PERF_EF_UPDATE);
>> +    hns3_pmu->hw_events[hwc->idx] = NULL;
>> +    perf_event_update_userpage(event);
>> +}
>> +
>> +static void hns3_pmu_enable(struct pmu *pmu)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(pmu);
>> +    u32 val;
>> +
>> +    val = readl(hns3_pmu->base + HNS3_PMU_REG_GLOBAL_CTRL);
>> +    val |= HNS3_PMU_GLOBAL_START;
>> +    writel(val, hns3_pmu->base + HNS3_PMU_REG_GLOBAL_CTRL);
>> +}
>> +
>> +static void hns3_pmu_disable(struct pmu *pmu)
>> +{
>> +    struct hns3_pmu *hns3_pmu = to_hns3_pmu(pmu);
>> +    u32 val;
>> +
>> +    val = readl(hns3_pmu->base + HNS3_PMU_REG_GLOBAL_CTRL);
>> +    val &= ~HNS3_PMU_GLOBAL_START;
>> +    writel(val, hns3_pmu->base + HNS3_PMU_REG_GLOBAL_CTRL);
>> +}
>> +
>> +static int hns3_pmu_alloc_pmu(struct pci_dev *pdev, struct hns3_pmu *hns3_pmu)
>> +{
>> +    u16 device_id;
>> +    char *name;
>> +    u32 val;
>> +    int ret;
>> +
>> +    hns3_pmu->base = pci_ioremap_bar(pdev, 2);
> 
> Can you use pcim_ioremap_regions()?
> 
ok.

>> +    if (!hns3_pmu->base) {
>> +        pci_err(pdev, "ioremap failed for hns3_pmu resource\n");
>> +        ret = -ENOMEM;
>> +        goto err_ioremap_bar;
>> +    }
>> +
>> +    hns3_pmu->hw_clk_freq = readl(hns3_pmu->base + HNS3_PMU_REG_CLOCK_FREQ);
>> +
>> +    val = readl(hns3_pmu->base + HNS3_PMU_REG_BDF);
>> +    hns3_pmu->bdf_min = val & 0xffff;
>> +    hns3_pmu->bdf_max = val >> 16;
>> +
>> +    val = readl(hns3_pmu->base + HNS3_PMU_REG_DEVICE_ID);
>> +    device_id = val & 0xffff;
>> +    name = devm_kasprintf(&pdev->dev, GFP_KERNEL, "hns3_pmu_sicl_%u", device_id);
>> +    if (!name) {
>> +        ret = -ENOMEM;
>> +        goto err_alloc_dev_name;
>> +    }
>> +
>> +    hns3_pmu->pdev = pdev;
>> +    hns3_pmu->on_cpu = -1;
>> +    hns3_pmu->identifier = readl(hns3_pmu->base + HNS3_PMU_REG_VERSION);
>> +    hns3_pmu->pmu = (struct pmu) {
>> +        .name        = name,
>> +        .module        = THIS_MODULE,
>> +        .event_init    = hns3_pmu_event_init,
>> +        .pmu_enable    = hns3_pmu_enable,
>> +        .pmu_disable    = hns3_pmu_disable,
>> +        .add        = hns3_pmu_add,
>> +        .del        = hns3_pmu_del,
>> +        .start        = hns3_pmu_start,
>> +        .stop        = hns3_pmu_stop,
>> +        .read        = hns3_pmu_read,
>> +        .task_ctx_nr    = perf_invalid_context,
>> +        .attr_groups    = hns3_pmu_attr_groups,
>> +        .capabilities    = PERF_PMU_CAP_NO_EXCLUDE,
>> +    };
>> +
>> +    return 0;
>> +
>> +err_alloc_dev_name:
>> +    iounmap(hns3_pmu->base);
>> +err_ioremap_bar:
>> +    pci_release_regions(pdev);
> 
> I don't see what this is un-doing in this function. You should only un-do what is done in this function
> 
Ok, thanks.

>> +
>> +    return ret;
>> +}
>> +
>> +static irqreturn_t hns3_pmu_irq(int irq, void *data)
>> +{
>> +    struct hns3_pmu *hns3_pmu = data;
>> +    u32 intr_status, idx;
>> +
>> +    for (idx = 0; idx < HNS3_PMU_MAX_HW_EVENTS; idx++) {
>> +        intr_status = hns3_pmu_readl(hns3_pmu,
>> +                         HNS3_PMU_REG_EVENT_INTR_STATUS,
>> +                         idx);
>> +
>> +        /*
>> +         * As each counter will restart from 0 when it is overflowed,
>> +         * extra processing is no need, just clear interrupt status.
>> +         */
>> +        if (intr_status)
>> +            hns3_pmu_clear_intr_status(hns3_pmu, idx);
>> +    }
>> +
>> +    return IRQ_HANDLED;
>> +}
>> +
>> +static int hns3_pmu_online_cpu(unsigned int cpu, struct hlist_node *node)
>> +{
>> +    struct hns3_pmu *hns3_pmu;
>> +
>> +    hns3_pmu = hlist_entry_safe(node, struct hns3_pmu, node);
>> +    if (!hns3_pmu)
>> +        return -ENODEV;
>> +
>> +    if (hns3_pmu->on_cpu == -1) {
>> +        hns3_pmu->on_cpu = cpu;
>> +        irq_set_affinity(hns3_pmu->irq, cpumask_of(cpu));
>> +    }
>> +
>> +    return 0;
>> +}
>> +
>> +static int hns3_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
>> +{
>> +    struct hns3_pmu *hns3_pmu;
>> +    unsigned int target;
>> +
>> +    hns3_pmu = hlist_entry_safe(node, struct hns3_pmu, node);
>> +    if (!hns3_pmu)
>> +        return -ENODEV;
>> +
>> +    /* Nothing to do if this CPU doesn't own the PMU */
>> +    if (hns3_pmu->on_cpu != cpu)
>> +        return 0;
>> +
>> +    /* Choose a new CPU from all online cpus */
>> +    target = cpumask_any_but(cpu_online_mask, cpu);
>> +    if (target >= nr_cpu_ids)
>> +        return 0;
> 
> I hope that some CPU is online...
> 
I don't understand here, can you explain in detail?

>> +
>> +    perf_pmu_migrate_context(&hns3_pmu->pmu, cpu, target);
>> +    hns3_pmu->on_cpu = target;
>> +    irq_set_affinity(hns3_pmu->irq, cpumask_of(target));
>> +
>> +    return 0;
>> +}
>> +
>> +static void hns3_pmu_free_irq(void *data)
>> +{
>> +    struct pci_dev *pdev = data;
>> +
>> +    pci_free_irq_vectors(pdev);
>> +}
>> +
>> +static int hns3_pmu_irq_register(struct pci_dev *pdev,
>> +                 struct hns3_pmu *hns3_pmu)
>> +{
>> +    int irq, ret;
>> +
>> +    ret = pci_alloc_irq_vectors(pdev, 1, 1, PCI_IRQ_MSI);
>> +    if (ret < 0) {
>> +        pci_err(pdev, "failed to enable MSI vectors, ret = %d.\n", ret);
>> +        return ret;
>> +    }
>> +
>> +    ret = devm_add_action(&pdev->dev, hns3_pmu_free_irq, pdev);
>> +    if (ret) {
>> +        pci_err(pdev, "failed to add free irq action, ret = %d.\n", ret);
>> +        return ret;
>> +    }
>> +
>> +    irq = pci_irq_vector(pdev, 0);
>> +    ret = devm_request_irq(&pdev->dev, irq, hns3_pmu_irq, 0,
>> +                   hns3_pmu->pmu.name, hns3_pmu);
>> +    if (ret) {
>> +        pci_err(pdev, "failed to register irq, ret = %d.\n", ret);
>> +        return ret;
>> +    }
>> +
>> +    hns3_pmu->irq = irq;
>> +
>> +    return 0;
>> +}
>> +
>> +static int hns3_pmu_init(struct pci_dev *pdev, struct hns3_pmu *hns3_pmu)
> 
> nit: "init" generally is for module init, so I would use "probe" or similar
> 
The function hns3_pmu_probe() already exists, how about rename it as hns3_pmu_init_pmu()?
Just refer to PCIe PMU driver.


>> +{
>> +    int ret;
>> +
>> +    ret = hns3_pmu_alloc_pmu(pdev, hns3_pmu);
>> +    if (ret)
>> +        return ret;
>> +
>> +    ret = hns3_pmu_irq_register(pdev, hns3_pmu);
>> +    if (ret)
>> +        goto err_irq_register;
>> +
>> +    ret = cpuhp_state_add_instance(CPUHP_AP_PERF_ARM_HNS3_PMU_ONLINE,
>> +                       &hns3_pmu->node);
>> +    if (ret) {
>> +        pci_err(pdev, "failed to register hotplug, ret = %d.\n", ret);
>> +        goto err_irq_register;
>> +    }
>> +
>> +    ret = perf_pmu_register(&hns3_pmu->pmu, hns3_pmu->pmu.name, -1);
>> +    if (ret) {
>> +        pci_err(pdev, "failed to register perf PMU, ret = %d.\n", ret);
>> +        goto err_register_pmu;
>> +    }
>> +
>> +    return ret;
>> +
>> +err_register_pmu:
>> +    cpuhp_state_remove_instance(CPUHP_AP_PERF_ARM_HNS3_PMU_ONLINE,
>> +                    &hns3_pmu->node);
>> +
>> +err_irq_register:
>> +    iounmap(hns3_pmu->base);
>> +
>> +    return ret;
>> +}
>> +
>> +static void hns3_pmu_uninit(struct pci_dev *pdev)
>> +{
>> +    struct hns3_pmu *hns3_pmu = pci_get_drvdata(pdev);
>> +
>> +    perf_pmu_unregister(&hns3_pmu->pmu);
>> +    cpuhp_state_remove_instance(CPUHP_AP_PERF_ARM_HNS3_PMU_ONLINE,
>> +                    &hns3_pmu->node);
>> +    iounmap(hns3_pmu->base);
>> +}
>> +
>> +static int hns3_pmu_init_dev(struct pci_dev *pdev)
>> +{
>> +    int ret;
>> +
>> +    ret = pcim_enable_device(pdev);
>> +    if (ret) {
>> +        pci_err(pdev, "failed to enable pci device, ret = %d.\n", ret);
>> +        return ret;
>> +    }
>> +
>> +    ret = pci_request_mem_regions(pdev, "hns3_pmu");
>> +    if (ret < 0) {
>> +        pci_err(pdev, "failed to request pci mem regions, ret = %d.\n",
>> +            ret);
>> +        return ret;
>> +    }
>> +
>> +    pci_set_master(pdev);
>> +
>> +    return 0;
>> +}
>> +
>> +static void hns3_pmu_uninit_dev(struct pci_dev *pdev)
>> +{
>> +    pci_clear_master(pdev);
>> +    pci_release_mem_regions(pdev);
>> +}
>> +
>> +static int hns3_pmu_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>> +{
>> +    struct hns3_pmu *hns3_pmu;
>> +    int ret;
>> +
>> +    hns3_pmu = devm_kzalloc(&pdev->dev, sizeof(*hns3_pmu), GFP_KERNEL);
>> +    if (!hns3_pmu)
>> +        return -ENOMEM;
>> +
>> +    ret = hns3_pmu_init_dev(pdev);
>> +    if (ret)
>> +        return ret;
>> +
>> +    ret = hns3_pmu_init(pdev, hns3_pmu);
>> +    if (ret) {
>> +        hns3_pmu_uninit_dev(pdev);
>> +        return ret;
>> +    }
>> +
>> +    pci_set_drvdata(pdev, hns3_pmu);
>> +
>> +    return ret;
>> +}
>> +
>> +static void hns3_pmu_remove(struct pci_dev *pdev)
>> +{
>> +    hns3_pmu_uninit(pdev);
>> +    hns3_pmu_uninit_dev(pdev);
>> +    pci_set_drvdata(pdev, NULL);
>> +}
>> +
>> +static const struct pci_device_id hns3_pmu_ids[] = {
>> +    { PCI_DEVICE(PCI_VENDOR_ID_HUAWEI, 0xA22B) },
>> +    { 0, }
>> +};
>> +MODULE_DEVICE_TABLE(pci, hns3_pmu_ids);
>> +
>> +static struct pci_driver hns3_pmu_driver = {
>> +    .name = "hns3_pmu",
>> +    .id_table = hns3_pmu_ids,
>> +    .probe = hns3_pmu_probe,
>> +    .remove = hns3_pmu_remove,
>> +};
>> +
>> +static int __init hns3_pmu_module_init(void)
>> +{
>> +    int ret;
>> +
>> +    ret = hns3_pmu_init_filter_mode_group();
>> +    if (ret)
>> +        return ret;
>> +
>> +    ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_HNS3_PMU_ONLINE,
>> +                      "AP_PERF_ARM_HNS3_PMU_ONLINE",
>> +                      hns3_pmu_online_cpu,
>> +                      hns3_pmu_offline_cpu);
>> +    if (ret) {
>> +        pr_err("failed to setup HNS3 PMU hotplug, ret = %d.\n", ret);
>> +        goto err_init;
>> +    }
>> +
>> +    ret = pci_register_driver(&hns3_pmu_driver);
>> +    if (ret) {
>> +        pr_err("failed to register pci driver, ret = %d.\n", ret);
>> +        goto err_hp;
>> +    }
>> +
>> +    return 0;
>> +
>> +err_hp:
>> +    cpuhp_remove_multi_state(CPUHP_AP_PERF_ARM_HNS3_PMU_ONLINE);
>> +err_init:
>> +    hns3_pmu_uinit_filter_mode_group();
>> +    return ret;
>> +}
>> +module_init(hns3_pmu_module_init);
>> +
>> +static void __exit hns3_pmu_module_exit(void)
>> +{
>> +    pci_unregister_driver(&hns3_pmu_driver);
>> +    cpuhp_remove_multi_state(CPUHP_AP_PERF_ARM_HNS3_PMU_ONLINE);
>> +    hns3_pmu_uinit_filter_mode_group();
>> +}
>> +module_exit(hns3_pmu_module_exit);
>> +
>> +MODULE_DESCRIPTION("HNS3 PMU driver");
>> +MODULE_LICENSE("GPL v2");
>> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
>> index 411a428ace4d..9c2cb0600740 100644
>> --- a/include/linux/cpuhotplug.h
>> +++ b/include/linux/cpuhotplug.h
>> @@ -226,6 +226,7 @@ enum cpuhp_state {
>>       CPUHP_AP_PERF_ARM_HISI_PA_ONLINE,
>>       CPUHP_AP_PERF_ARM_HISI_SLLC_ONLINE,
>>       CPUHP_AP_PERF_ARM_HISI_PCIE_PMU_ONLINE,
>> +    CPUHP_AP_PERF_ARM_HNS3_PMU_ONLINE,
>>       CPUHP_AP_PERF_ARM_L2X0_ONLINE,
>>       CPUHP_AP_PERF_ARM_QCOM_L2_ONLINE,
>>       CPUHP_AP_PERF_ARM_QCOM_L3_ONLINE,
> 
> .