All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] ntb: Add more debugfs support for ntb_perf testing options
@ 2017-05-04 16:37 Gary R Hook
  2017-05-04 16:50 ` Dave Jiang
  0 siblings, 1 reply; 3+ messages in thread
From: Gary R Hook @ 2017-05-04 16:37 UTC (permalink / raw)
  To: linux-ntb; +Cc: Allen.Hubbe, dave.jiang, jdmason

The ntb_perf tool uses module parameters to control the
characteristics of its test.  Enable the changing of these
options through debugfs, and eliminating the need to unload
and reload the module to make changes and run additional tests.

Add a new module parameter that forces the DMA channel
selection onto the same node as the NTB device (default: true).

 - seg_order: Size of the NTB memory window; power of 2.
 - run_order: Size of the data buffer; power of 2.
 - use_dma:   Use DMA or memcpy? Default: 0.
 - on_node:   Only use DMA channel(s) on the NTB node. Default: true.


Signed-off-by: Gary R Hook <gary.hook@amd.com>
---
 drivers/ntb/test/ntb_perf.c |   48 +++++++++++++++++++++++++++++++++++++++----
 1 file changed, 44 insertions(+), 4 deletions(-)

diff --git a/drivers/ntb/test/ntb_perf.c b/drivers/ntb/test/ntb_perf.c
index 42756a98a728..425c2a1b36e7 100644
--- a/drivers/ntb/test/ntb_perf.c
+++ b/drivers/ntb/test/ntb_perf.c
@@ -101,6 +101,10 @@
 module_param(use_dma, bool, 0644);
 MODULE_PARM_DESC(use_dma, "Using DMA engine to measure performance");
 
+static bool on_node = true; /* default to 1 */
+module_param(on_node, bool, 0644);
+MODULE_PARM_DESC(on_node, "Run threads only on NTB device node (default: true)");
+
 struct perf_mw {
 	phys_addr_t	phys_addr;
 	resource_size_t	phys_size;
@@ -139,6 +143,10 @@ struct perf_ctx {
 	struct dentry		*debugfs_node_dir;
 	struct dentry		*debugfs_run;
 	struct dentry		*debugfs_threads;
+	struct dentry		*debugfs_seg_order;
+	struct dentry		*debugfs_run_order;
+	struct dentry		*debugfs_use_dma;
+	struct dentry		*debugfs_on_node;
 	u8			perf_threads;
 	/* mutex ensures only one set of threads run at once */
 	struct mutex		run_mutex;
@@ -345,6 +353,10 @@ static int perf_move_data(struct pthr_ctx *pctx, char __iomem *dst, char *src,
 
 static bool perf_dma_filter_fn(struct dma_chan *chan, void *node)
 {
+	/* Is the channel required to be on the same node as the device? */
+	if (!on_node)
+		return true;
+
 	return dev_to_node(&chan->dev->device) == (int)(unsigned long)node;
 }
 
@@ -682,7 +694,8 @@ static ssize_t debugfs_run_write(struct file *filp, const char __user *ubuf,
 		pr_info("Fix run_order to %u\n", run_order);
 	}
 
-	node = dev_to_node(&perf->ntb->pdev->dev);
+	node = on_node ? dev_to_node(&perf->ntb->pdev->dev)
+		       : NUMA_NO_NODE;
 	atomic_set(&perf->tdone, 0);
 
 	/* launch kernel thread */
@@ -743,18 +756,42 @@ static int perf_debugfs_setup(struct perf_ctx *perf)
 	if (!perf->debugfs_node_dir)
 		return -ENODEV;
 
-	perf->debugfs_run = debugfs_create_file("run", S_IRUSR | S_IWUSR,
+	perf->debugfs_run = debugfs_create_file("run", 0600,
 						perf->debugfs_node_dir, perf,
 						&ntb_perf_debugfs_run);
 	if (!perf->debugfs_run)
 		return -ENODEV;
 
-	perf->debugfs_threads = debugfs_create_u8("threads", S_IRUSR | S_IWUSR,
+	perf->debugfs_threads = debugfs_create_u8("threads", 0600,
 						  perf->debugfs_node_dir,
 						  &perf->perf_threads);
 	if (!perf->debugfs_threads)
 		return -ENODEV;
 
+	perf->debugfs_seg_order = debugfs_create_u32("seg_order", 0600,
+						  perf->debugfs_node_dir,
+						  &seg_order);
+	if (!perf->debugfs_seg_order)
+		return -ENODEV;
+
+	perf->debugfs_run_order = debugfs_create_u32("run_order", 0600,
+						  perf->debugfs_node_dir,
+						  &run_order);
+	if (!perf->debugfs_run_order)
+		return -ENODEV;
+
+	perf->debugfs_use_dma = debugfs_create_u32("use_dma", 0600,
+						  perf->debugfs_node_dir,
+						  &use_dma);
+	if (!perf->debugfs_use_dma)
+		return -ENODEV;
+
+	perf->debugfs_on_node = debugfs_create_bool("on_node", 0600,
+						  perf->debugfs_node_dir,
+						  &on_node);
+	if (!perf->debugfs_on_node)
+		return -ENODEV;
+
 	return 0;
 }
 
@@ -781,7 +818,10 @@ static int perf_probe(struct ntb_client *client, struct ntb_dev *ntb)
 
 	node = dev_to_node(&pdev->dev);
 
-	perf = kzalloc_node(sizeof(*perf), GFP_KERNEL, node);
+	if (on_node)
+		perf = kzalloc_node(sizeof(*perf), GFP_KERNEL, node);
+	else
+		perf = kzalloc(sizeof(*perf), GFP_KERNEL);
 	if (!perf) {
 		rc = -ENOMEM;
 		goto err_perf;


^ permalink raw reply related	[flat|nested] 3+ messages in thread
* Re: [PATCH] ntb: Add more debugfs support for ntb_perf testing options
  2017-05-04 16:37 [PATCH] ntb: Add more debugfs support for ntb_perf testing options Gary R Hook
@ 2017-05-04 16:50 ` Dave Jiang
  2017-05-04 17:00   ` Gary R Hook
  0 siblings, 1 reply; 3+ messages in thread
From: Dave Jiang @ 2017-05-04 16:50 UTC (permalink / raw)
  To: Gary R Hook, linux-ntb; +Cc: Allen.Hubbe, jdmason



On 05/04/2017 09:37 AM, Gary R Hook wrote:
> The ntb_perf tool uses module parameters to control the
> characteristics of its test.  Enable the changing of these
> options through debugfs, and eliminating the need to unload
> and reload the module to make changes and run additional tests.
> 
> Add a new module parameter that forces the DMA channel
> selection onto the same node as the NTB device (default: true).
> 
>  - seg_order: Size of the NTB memory window; power of 2.
>  - run_order: Size of the data buffer; power of 2.
>  - use_dma:   Use DMA or memcpy? Default: 0.
>  - on_node:   Only use DMA channel(s) on the NTB node. Default: true.
> 

Do you mind breaking these into separate patches? One for the debugfs
improvements and another for the NUMA filter for the DMA? It would make
git bisect and debugging easier later on if we do one change set per
commit. Thanks!

> 
> Signed-off-by: Gary R Hook <gary.hook@amd.com>
> ---
>  drivers/ntb/test/ntb_perf.c |   48 +++++++++++++++++++++++++++++++++++++++----
>  1 file changed, 44 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/ntb/test/ntb_perf.c b/drivers/ntb/test/ntb_perf.c
> index 42756a98a728..425c2a1b36e7 100644
> --- a/drivers/ntb/test/ntb_perf.c
> +++ b/drivers/ntb/test/ntb_perf.c
> @@ -101,6 +101,10 @@
>  module_param(use_dma, bool, 0644);
>  MODULE_PARM_DESC(use_dma, "Using DMA engine to measure performance");
>  
> +static bool on_node = true; /* default to 1 */
> +module_param(on_node, bool, 0644);
> +MODULE_PARM_DESC(on_node, "Run threads only on NTB device node (default: true)");
> +
>  struct perf_mw {
>  	phys_addr_t	phys_addr;
>  	resource_size_t	phys_size;
> @@ -139,6 +143,10 @@ struct perf_ctx {
>  	struct dentry		*debugfs_node_dir;
>  	struct dentry		*debugfs_run;
>  	struct dentry		*debugfs_threads;
> +	struct dentry		*debugfs_seg_order;
> +	struct dentry		*debugfs_run_order;
> +	struct dentry		*debugfs_use_dma;
> +	struct dentry		*debugfs_on_node;
>  	u8			perf_threads;
>  	/* mutex ensures only one set of threads run at once */
>  	struct mutex		run_mutex;
> @@ -345,6 +353,10 @@ static int perf_move_data(struct pthr_ctx *pctx, char __iomem *dst, char *src,
>  
>  static bool perf_dma_filter_fn(struct dma_chan *chan, void *node)
>  {
> +	/* Is the channel required to be on the same node as the device? */
> +	if (!on_node)
> +		return true;
> +
>  	return dev_to_node(&chan->dev->device) == (int)(unsigned long)node;
>  }
>  
> @@ -682,7 +694,8 @@ static ssize_t debugfs_run_write(struct file *filp, const char __user *ubuf,
>  		pr_info("Fix run_order to %u\n", run_order);
>  	}
>  
> -	node = dev_to_node(&perf->ntb->pdev->dev);
> +	node = on_node ? dev_to_node(&perf->ntb->pdev->dev)
> +		       : NUMA_NO_NODE;
>  	atomic_set(&perf->tdone, 0);
>  
>  	/* launch kernel thread */
> @@ -743,18 +756,42 @@ static int perf_debugfs_setup(struct perf_ctx *perf)
>  	if (!perf->debugfs_node_dir)
>  		return -ENODEV;
>  
> -	perf->debugfs_run = debugfs_create_file("run", S_IRUSR | S_IWUSR,
> +	perf->debugfs_run = debugfs_create_file("run", 0600,
>  						perf->debugfs_node_dir, perf,
>  						&ntb_perf_debugfs_run);
>  	if (!perf->debugfs_run)
>  		return -ENODEV;
>  
> -	perf->debugfs_threads = debugfs_create_u8("threads", S_IRUSR | S_IWUSR,
> +	perf->debugfs_threads = debugfs_create_u8("threads", 0600,
>  						  perf->debugfs_node_dir,
>  						  &perf->perf_threads);
>  	if (!perf->debugfs_threads)
>  		return -ENODEV;
>  
> +	perf->debugfs_seg_order = debugfs_create_u32("seg_order", 0600,
> +						  perf->debugfs_node_dir,
> +						  &seg_order);
> +	if (!perf->debugfs_seg_order)
> +		return -ENODEV;
> +
> +	perf->debugfs_run_order = debugfs_create_u32("run_order", 0600,
> +						  perf->debugfs_node_dir,
> +						  &run_order);
> +	if (!perf->debugfs_run_order)
> +		return -ENODEV;
> +
> +	perf->debugfs_use_dma = debugfs_create_u32("use_dma", 0600,
> +						  perf->debugfs_node_dir,
> +						  &use_dma);
> +	if (!perf->debugfs_use_dma)
> +		return -ENODEV;
> +
> +	perf->debugfs_on_node = debugfs_create_bool("on_node", 0600,
> +						  perf->debugfs_node_dir,
> +						  &on_node);
> +	if (!perf->debugfs_on_node)
> +		return -ENODEV;
> +
>  	return 0;
>  }
>  
> @@ -781,7 +818,10 @@ static int perf_probe(struct ntb_client *client, struct ntb_dev *ntb)
>  
>  	node = dev_to_node(&pdev->dev);
>  
> -	perf = kzalloc_node(sizeof(*perf), GFP_KERNEL, node);
> +	if (on_node)
> +		perf = kzalloc_node(sizeof(*perf), GFP_KERNEL, node);
> +	else
> +		perf = kzalloc(sizeof(*perf), GFP_KERNEL);
>  	if (!perf) {
>  		rc = -ENOMEM;
>  		goto err_perf;
> 

^ permalink raw reply	[flat|nested] 3+ messages in thread
* Re: [PATCH] ntb: Add more debugfs support for ntb_perf testing options
  2017-05-04 16:50 ` Dave Jiang
@ 2017-05-04 17:00   ` Gary R Hook
  0 siblings, 0 replies; 3+ messages in thread
From: Gary R Hook @ 2017-05-04 17:00 UTC (permalink / raw)
  To: Dave Jiang, Gary R Hook, linux-ntb; +Cc: Allen.Hubbe, jdmason

On 05/04/2017 11:50 AM, Dave Jiang wrote:
>
>
> On 05/04/2017 09:37 AM, Gary R Hook wrote:
>> The ntb_perf tool uses module parameters to control the
>> characteristics of its test.  Enable the changing of these
>> options through debugfs, and eliminating the need to unload
>> and reload the module to make changes and run additional tests.
>>
>> Add a new module parameter that forces the DMA channel
>> selection onto the same node as the NTB device (default: true).
>>
>>  - seg_order: Size of the NTB memory window; power of 2.
>>  - run_order: Size of the data buffer; power of 2.
>>  - use_dma:   Use DMA or memcpy? Default: 0.
>>  - on_node:   Only use DMA channel(s) on the NTB node. Default: true.
>>
>
> Do you mind breaking these into separate patches? One for the debugfs
> improvements and another for the NUMA filter for the DMA? It would make
> git bisect and debugging easier later on if we do one change set per
> commit. Thanks!

I would be happy to. I was going to have to negate this one anyway, as I 
found
a problem right after submitting.

This will end up as a (short) series, with a "V2".


>>
>> Signed-off-by: Gary R Hook <gary.hook@amd.com>
>> ---
>>  drivers/ntb/test/ntb_perf.c |   48 +++++++++++++++++++++++++++++++++++++++----
>>  1 file changed, 44 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/ntb/test/ntb_perf.c b/drivers/ntb/test/ntb_perf.c
>> index 42756a98a728..425c2a1b36e7 100644
>> --- a/drivers/ntb/test/ntb_perf.c
>> +++ b/drivers/ntb/test/ntb_perf.c
>> @@ -101,6 +101,10 @@
>>  module_param(use_dma, bool, 0644);
>>  MODULE_PARM_DESC(use_dma, "Using DMA engine to measure performance");
>>
>> +static bool on_node = true; /* default to 1 */
>> +module_param(on_node, bool, 0644);
>> +MODULE_PARM_DESC(on_node, "Run threads only on NTB device node (default: true)");
>> +
>>  struct perf_mw {
>>  	phys_addr_t	phys_addr;
>>  	resource_size_t	phys_size;
>> @@ -139,6 +143,10 @@ struct perf_ctx {
>>  	struct dentry		*debugfs_node_dir;
>>  	struct dentry		*debugfs_run;
>>  	struct dentry		*debugfs_threads;
>> +	struct dentry		*debugfs_seg_order;
>> +	struct dentry		*debugfs_run_order;
>> +	struct dentry		*debugfs_use_dma;
>> +	struct dentry		*debugfs_on_node;
>>  	u8			perf_threads;
>>  	/* mutex ensures only one set of threads run at once */
>>  	struct mutex		run_mutex;
>> @@ -345,6 +353,10 @@ static int perf_move_data(struct pthr_ctx *pctx, char __iomem *dst, char *src,
>>
>>  static bool perf_dma_filter_fn(struct dma_chan *chan, void *node)
>>  {
>> +	/* Is the channel required to be on the same node as the device? */
>> +	if (!on_node)
>> +		return true;
>> +
>>  	return dev_to_node(&chan->dev->device) == (int)(unsigned long)node;
>>  }
>>
>> @@ -682,7 +694,8 @@ static ssize_t debugfs_run_write(struct file *filp, const char __user *ubuf,
>>  		pr_info("Fix run_order to %u\n", run_order);
>>  	}
>>
>> -	node = dev_to_node(&perf->ntb->pdev->dev);
>> +	node = on_node ? dev_to_node(&perf->ntb->pdev->dev)
>> +		       : NUMA_NO_NODE;
>>  	atomic_set(&perf->tdone, 0);
>>
>>  	/* launch kernel thread */
>> @@ -743,18 +756,42 @@ static int perf_debugfs_setup(struct perf_ctx *perf)
>>  	if (!perf->debugfs_node_dir)
>>  		return -ENODEV;
>>
>> -	perf->debugfs_run = debugfs_create_file("run", S_IRUSR | S_IWUSR,
>> +	perf->debugfs_run = debugfs_create_file("run", 0600,
>>  						perf->debugfs_node_dir, perf,
>>  						&ntb_perf_debugfs_run);
>>  	if (!perf->debugfs_run)
>>  		return -ENODEV;
>>
>> -	perf->debugfs_threads = debugfs_create_u8("threads", S_IRUSR | S_IWUSR,
>> +	perf->debugfs_threads = debugfs_create_u8("threads", 0600,
>>  						  perf->debugfs_node_dir,
>>  						  &perf->perf_threads);
>>  	if (!perf->debugfs_threads)
>>  		return -ENODEV;
>>
>> +	perf->debugfs_seg_order = debugfs_create_u32("seg_order", 0600,
>> +						  perf->debugfs_node_dir,
>> +						  &seg_order);
>> +	if (!perf->debugfs_seg_order)
>> +		return -ENODEV;
>> +
>> +	perf->debugfs_run_order = debugfs_create_u32("run_order", 0600,
>> +						  perf->debugfs_node_dir,
>> +						  &run_order);
>> +	if (!perf->debugfs_run_order)
>> +		return -ENODEV;
>> +
>> +	perf->debugfs_use_dma = debugfs_create_u32("use_dma", 0600,
>> +						  perf->debugfs_node_dir,
>> +						  &use_dma);
>> +	if (!perf->debugfs_use_dma)
>> +		return -ENODEV;
>> +
>> +	perf->debugfs_on_node = debugfs_create_bool("on_node", 0600,
>> +						  perf->debugfs_node_dir,
>> +						  &on_node);
>> +	if (!perf->debugfs_on_node)
>> +		return -ENODEV;
>> +
>>  	return 0;
>>  }
>>
>> @@ -781,7 +818,10 @@ static int perf_probe(struct ntb_client *client, struct ntb_dev *ntb)
>>
>>  	node = dev_to_node(&pdev->dev);
>>
>> -	perf = kzalloc_node(sizeof(*perf), GFP_KERNEL, node);
>> +	if (on_node)
>> +		perf = kzalloc_node(sizeof(*perf), GFP_KERNEL, node);
>> +	else
>> +		perf = kzalloc(sizeof(*perf), GFP_KERNEL);
>>  	if (!perf) {
>>  		rc = -ENOMEM;
>>  		goto err_perf;
>>

-- 
This is my day job. Follow me at:
IG/Twitter/Facebook: @grhookphoto
IG/Twitter/Facebook: @grhphotographer

^ permalink raw reply	[flat|nested] 3+ messages in thread
end of thread, other threads:[~2017-05-04 17:00 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-05-04 16:37 [PATCH] ntb: Add more debugfs support for ntb_perf testing options Gary R Hook
2017-05-04 16:50 ` Dave Jiang
2017-05-04 17:00   ` Gary R Hook

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.