[PATCH 1/3] perf/amlogic: Fix config1/config2 parsing issue

[PATCH 1/3] perf/amlogic: Fix config1/config2 parsing issue

[Jiucheng Xu]

The 3th argument of for_each_set_bit is incorrect, fix them.

Fixes: 2016e2113d35 ("perf/amlogic: Add support for Amlogic meson G12 SoC DDR PMU driver")
Signed-off-by: Jiucheng Xu <jiucheng.xu@amlogic.com>
---
 drivers/perf/amlogic/meson_ddr_pmu_core.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/perf/amlogic/meson_ddr_pmu_core.c b/drivers/perf/amlogic/meson_ddr_pmu_core.c
index b84346dbac2c..0b24dee1ed3c 100644
--- a/drivers/perf/amlogic/meson_ddr_pmu_core.c
+++ b/drivers/perf/amlogic/meson_ddr_pmu_core.c
@@ -156,10 +156,14 @@ static int meson_ddr_perf_event_add(struct perf_event *event, int flags)
 	u64 config2 = event->attr.config2;
 	int i;
 
-	for_each_set_bit(i, (const unsigned long *)&config1, sizeof(config1))
+	for_each_set_bit(i,
+			 (const unsigned long *)&config1,
+			 BITS_PER_TYPE(config1))
 		meson_ddr_set_axi_filter(event, i);
 
-	for_each_set_bit(i, (const unsigned long *)&config2, sizeof(config2))
+	for_each_set_bit(i,
+			 (const unsigned long *)&config2,
+			 BITS_PER_TYPE(config2))
 		meson_ddr_set_axi_filter(event, i + 64);
 
 	if (flags & PERF_EF_START)

base-commit: e8a709dc2a9156f223ec953ae70a919e87ad7e9a
-- 
2.25.1


_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

[PATCH 2/3] perf/amlogic: Fix large number of counter issue

[Jiucheng Xu]

When use 1ms interval, very large number of counter happens
once in a while as below:

25.968654513 281474976710655.84 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
26.118657346 281474976710655.88 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
26.180137180 281474976710655.66 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/

Root cause is the race between irq handler
and pmu.read callback. Use spin lock to protect the sw&hw
counters.

Signed-off-by: Jiucheng Xu <jiucheng.xu@amlogic.com>
---
 drivers/perf/amlogic/meson_ddr_pmu_core.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/perf/amlogic/meson_ddr_pmu_core.c b/drivers/perf/amlogic/meson_ddr_pmu_core.c
index 0b24dee1ed3c..9b2e5d5c0626 100644
--- a/drivers/perf/amlogic/meson_ddr_pmu_core.c
+++ b/drivers/perf/amlogic/meson_ddr_pmu_core.c
@@ -14,6 +14,7 @@
 #include <linux/perf_event.h>
 #include <linux/platform_device.h>
 #include <linux/printk.h>
+#include <linux/spinlock.h>
 #include <linux/sysfs.h>
 #include <linux/types.h>
 
@@ -23,6 +24,7 @@ struct ddr_pmu {
 	struct pmu pmu;
 	struct dmc_info info;
 	struct dmc_counter counters;	/* save counters from hw */
+	spinlock_t lock;		/* protect hw/sw counter */
 	bool pmu_enabled;
 	struct device *dev;
 	char *name;
@@ -92,10 +94,12 @@ static void meson_ddr_perf_event_update(struct perf_event *event)
 	int idx;
 	int chann_nr = pmu->info.hw_info->chann_nr;
 
+	spin_lock(&pmu->lock);
 	/* get the remain counters in register. */
 	pmu->info.hw_info->get_counters(&pmu->info, &dc);
 
 	ddr_cnt_addition(&sum_dc, &pmu->counters, &dc, chann_nr);
+	spin_unlock(&pmu->lock);
 
 	switch (event->attr.config) {
 	case ALL_CHAN_COUNTER_ID:
@@ -355,6 +359,7 @@ static irqreturn_t dmc_irq_handler(int irq, void *dev_id)
 
 	pmu = dmc_info_to_pmu(info);
 
+	spin_lock(&pmu->lock);
 	if (info->hw_info->irq_handler(info, &counters) != 0)
 		goto out;
 
@@ -372,6 +377,8 @@ static irqreturn_t dmc_irq_handler(int irq, void *dev_id)
 		 * it in ISR to support continue mode.
 		 */
 		info->hw_info->enable(info);
+out:
+	spin_unlock(&pmu->lock);
 
 	dev_dbg(pmu->dev, "counts: %llu %llu %llu, %llu, %llu, %llu\t\t"
 			"sum: %llu %llu %llu, %llu, %llu, %llu\n",
@@ -388,7 +395,7 @@ static irqreturn_t dmc_irq_handler(int irq, void *dev_id)
 			pmu->counters.channel_cnt[1],
 			pmu->counters.channel_cnt[2],
 			pmu->counters.channel_cnt[3]);
-out:
+
 	return IRQ_HANDLED;
 }
 
@@ -539,6 +546,7 @@ int meson_ddr_pmu_create(struct platform_device *pdev)
 	pmu->name = name;
 	pmu->dev = &pdev->dev;
 	pmu->pmu_enabled = false;
+	spin_lock_init(&pmu->lock);
 
 	platform_set_drvdata(pdev, pmu);
 
-- 
2.25.1


_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

[PATCH 3/3] arm64: defconfig: Add Meson DDR PMU as build-in

[Jiucheng Xu]

Add Meson DDR PMU to defconfig so that build errors are caught.

Signed-off-by: Jiucheng Xu <jiucheng.xu@amlogic.com>
---
 arch/arm64/configs/defconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
index 851e8f9be06d..9fefe659ade7 100644
--- a/arch/arm64/configs/defconfig
+++ b/arch/arm64/configs/defconfig
@@ -1296,6 +1296,7 @@ CONFIG_ARM_DMC620_PMU=m
 CONFIG_QCOM_L2_PMU=y
 CONFIG_QCOM_L3_PMU=y
 CONFIG_HISI_PMU=y
+CONFIG_MESON_DDR_PMU=y
 CONFIG_NVMEM_IMX_OCOTP=y
 CONFIG_NVMEM_IMX_OCOTP_SCU=y
 CONFIG_NVMEM_MTK_EFUSE=y
-- 
2.25.1


_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

Re: [PATCH 3/3] arm64: defconfig: Add Meson DDR PMU as build-in

[Chris Healy]

Reviewed-by: Chris Healy <healych@amazon.com>

On Thu, Feb 9, 2023 at 3:54 AM Jiucheng Xu <jiucheng.xu@amlogic.com> wrote:
>
> Add Meson DDR PMU to defconfig so that build errors are caught.
>
> Signed-off-by: Jiucheng Xu <jiucheng.xu@amlogic.com>
> ---
>  arch/arm64/configs/defconfig | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
> index 851e8f9be06d..9fefe659ade7 100644
> --- a/arch/arm64/configs/defconfig
> +++ b/arch/arm64/configs/defconfig
> @@ -1296,6 +1296,7 @@ CONFIG_ARM_DMC620_PMU=m
>  CONFIG_QCOM_L2_PMU=y
>  CONFIG_QCOM_L3_PMU=y
>  CONFIG_HISI_PMU=y
> +CONFIG_MESON_DDR_PMU=y
>  CONFIG_NVMEM_IMX_OCOTP=y
>  CONFIG_NVMEM_IMX_OCOTP_SCU=y
>  CONFIG_NVMEM_MTK_EFUSE=y
> --
> 2.25.1
>

_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

Re: [PATCH 2/3] perf/amlogic: Fix large number of counter issue

[Will Deacon]

On Thu, Feb 09, 2023 at 07:54:02PM +0800, Jiucheng Xu wrote:
> When use 1ms interval, very large number of counter happens
> once in a while as below:
> 
> 25.968654513 281474976710655.84 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
> 26.118657346 281474976710655.88 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
> 26.180137180 281474976710655.66 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
> 
> Root cause is the race between irq handler
> and pmu.read callback. Use spin lock to protect the sw&hw
> counters.
> 
> Signed-off-by: Jiucheng Xu <jiucheng.xu@amlogic.com>
> ---
>  drivers/perf/amlogic/meson_ddr_pmu_core.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/perf/amlogic/meson_ddr_pmu_core.c b/drivers/perf/amlogic/meson_ddr_pmu_core.c
> index 0b24dee1ed3c..9b2e5d5c0626 100644
> --- a/drivers/perf/amlogic/meson_ddr_pmu_core.c
> +++ b/drivers/perf/amlogic/meson_ddr_pmu_core.c
> @@ -14,6 +14,7 @@
>  #include <linux/perf_event.h>
>  #include <linux/platform_device.h>
>  #include <linux/printk.h>
> +#include <linux/spinlock.h>
>  #include <linux/sysfs.h>
>  #include <linux/types.h>
>  
> @@ -23,6 +24,7 @@ struct ddr_pmu {
>  	struct pmu pmu;
>  	struct dmc_info info;
>  	struct dmc_counter counters;	/* save counters from hw */
> +	spinlock_t lock;		/* protect hw/sw counter */
>  	bool pmu_enabled;
>  	struct device *dev;
>  	char *name;
> @@ -92,10 +94,12 @@ static void meson_ddr_perf_event_update(struct perf_event *event)
>  	int idx;
>  	int chann_nr = pmu->info.hw_info->chann_nr;
>  
> +	spin_lock(&pmu->lock);

Why doesn't this need the _irqsave() variant if we're racing with the irq
handler?

Will

_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

Re: [PATCH 1/3] perf/amlogic: Fix config1/config2 parsing issue

[Will Deacon]

On Thu, 9 Feb 2023 19:54:01 +0800, Jiucheng Xu wrote:
> The 3th argument of for_each_set_bit is incorrect, fix them.
> 
> 

Applied first patch only to will (for-next/perf), thanks!

[1/3] perf/amlogic: Fix config1/config2 parsing issue
      https://git.kernel.org/will/c/c61e5720f232

Cheers,
-- 
Will

https://fixes.arm64.dev
https://next.arm64.dev
https://will.arm64.dev

_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

Re: [PATCH 2/3] perf/amlogic: Fix large number of counter issue

[Jiucheng Xu]

My Amlogic email box has some issues. Use my personal email 
<jiucheng.xu@163.com> to reply.

On 2023/3/27 22:10, Will Deacon wrote:
> [ EXTERNAL EMAIL ]
>
> On Thu, Feb 09, 2023 at 07:54:02PM +0800, Jiucheng Xu wrote:
>> When use 1ms interval, very large number of counter happens
>> once in a while as below:
>>
>> 25.968654513 281474976710655.84 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
>> 26.118657346 281474976710655.88 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
>> 26.180137180 281474976710655.66 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
>>
>> Root cause is the race between irq handler
>> and pmu.read callback. Use spin lock to protect the sw&hw
>> counters.
>>
>> Signed-off-by: Jiucheng Xu <jiucheng.xu@amlogic.com>
>> ---
>>   drivers/perf/amlogic/meson_ddr_pmu_core.c | 10 +++++++++-
>>   1 file changed, 9 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/perf/amlogic/meson_ddr_pmu_core.c b/drivers/perf/amlogic/meson_ddr_pmu_core.c
>> index 0b24dee1ed3c..9b2e5d5c0626 100644
>> --- a/drivers/perf/amlogic/meson_ddr_pmu_core.c
>> +++ b/drivers/perf/amlogic/meson_ddr_pmu_core.c
>> @@ -14,6 +14,7 @@
>>   #include <linux/perf_event.h>
>>   #include <linux/platform_device.h>
>>   #include <linux/printk.h>
>> +#include <linux/spinlock.h>
>>   #include <linux/sysfs.h>
>>   #include <linux/types.h>
>>   
>> @@ -23,6 +24,7 @@ struct ddr_pmu {
>>   	struct pmu pmu;
>>   	struct dmc_info info;
>>   	struct dmc_counter counters;	/* save counters from hw */
>> +	spinlock_t lock;		/* protect hw/sw counter */
>>   	bool pmu_enabled;
>>   	struct device *dev;
>>   	char *name;
>> @@ -92,10 +94,12 @@ static void meson_ddr_perf_event_update(struct perf_event *event)
>>   	int idx;
>>   	int chann_nr = pmu->info.hw_info->chann_nr;
>>   
>> +	spin_lock(&pmu->lock);
> Why doesn't this need the _irqsave() variant if we're racing with the irq
> handler?
>
> Will
I think meson_ddr_perf_event_update function is called with hard irq off.
So update function couldn't be interrupted by irq handler. Right?

Thanks,
Jiucheng
>


_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

Re: [PATCH 2/3] perf/amlogic: Fix large number of counter issue

[Will Deacon]

On Tue, Mar 28, 2023 at 10:29:04AM +0800, Jiucheng Xu wrote:
> 
> My Amlogic email box has some issues. Use my personal email
> <jiucheng.xu@163.com> to reply.
> 
> On 2023/3/27 22:10, Will Deacon wrote:
> > [ EXTERNAL EMAIL ]
> > 
> > On Thu, Feb 09, 2023 at 07:54:02PM +0800, Jiucheng Xu wrote:
> > > When use 1ms interval, very large number of counter happens
> > > once in a while as below:
> > > 
> > > 25.968654513 281474976710655.84 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
> > > 26.118657346 281474976710655.88 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
> > > 26.180137180 281474976710655.66 MB meson_ddr_bw/chan_1_rw_bytes,arm=1/
> > > 
> > > Root cause is the race between irq handler
> > > and pmu.read callback. Use spin lock to protect the sw&hw
> > > counters.
> > > 
> > > Signed-off-by: Jiucheng Xu <jiucheng.xu@amlogic.com>
> > > ---
> > >   drivers/perf/amlogic/meson_ddr_pmu_core.c | 10 +++++++++-
> > >   1 file changed, 9 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/perf/amlogic/meson_ddr_pmu_core.c b/drivers/perf/amlogic/meson_ddr_pmu_core.c
> > > index 0b24dee1ed3c..9b2e5d5c0626 100644
> > > --- a/drivers/perf/amlogic/meson_ddr_pmu_core.c
> > > +++ b/drivers/perf/amlogic/meson_ddr_pmu_core.c
> > > @@ -14,6 +14,7 @@
> > >   #include <linux/perf_event.h>
> > >   #include <linux/platform_device.h>
> > >   #include <linux/printk.h>
> > > +#include <linux/spinlock.h>
> > >   #include <linux/sysfs.h>
> > >   #include <linux/types.h>
> > > @@ -23,6 +24,7 @@ struct ddr_pmu {
> > >   	struct pmu pmu;
> > >   	struct dmc_info info;
> > >   	struct dmc_counter counters;	/* save counters from hw */
> > > +	spinlock_t lock;		/* protect hw/sw counter */
> > >   	bool pmu_enabled;
> > >   	struct device *dev;
> > >   	char *name;
> > > @@ -92,10 +94,12 @@ static void meson_ddr_perf_event_update(struct perf_event *event)
> > >   	int idx;
> > >   	int chann_nr = pmu->info.hw_info->chann_nr;
> > > +	spin_lock(&pmu->lock);
> > Why doesn't this need the _irqsave() variant if we're racing with the irq
> > handler?
> > 
> > Will
> I think meson_ddr_perf_event_update function is called with hard irq off.
> So update function couldn't be interrupted by irq handler. Right?

I'm just confused about the race, then. The commit message says you have a
race between an irq handler and a callback, which you fix with a spinlock
that isn't irq safe. So either the race is real and the lock needs to be
irqsafe, or the race is something else entirely, no?

Will

_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic