linux-wireless.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Florian Fainelli <f.fainelli@gmail.com>
To: Dedy Lansky <qca_dlansky@qualcomm.com>,
	"John W . Linville" <linville@tuxdriver.com>
Cc: linux-wireless@vger.kernel.org, wil6210@qca.qualcomm.com,
	Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Subject: Re: [PATCH 04/14] wil6210: Optimization for Interrupt moderation threshold value
Date: Mon, 04 Aug 2014 10:38:50 -0700	[thread overview]
Message-ID: <53DFC52A.4030904@gmail.com> (raw)
In-Reply-To: <1407140454-31639-5-git-send-email-qca_dlansky@qca.qualcomm.com>

Hello,

On 08/04/2014 01:20 AM, Dedy Lansky wrote:
> Default value was changed to 50000 (~3 IRQs/msec) in order to reduce
> significantly CPU usage without degradation in throughput.
> Added module parameter named "itr_trsh" that allows the user to change
> default value through insmod.
> Added entry in debugfs (itr_trsh_val) for changing interrupt moderation
> threshold in a single operation.

This should be made configurable through the ethtool::set_coalesce
operations, which allows you to have separate RX and TX interrupts
coalescing parameters.

> 
> Signed-off-by: Dedy Lansky <qca_dlansky@qca.qualcomm.com>
> ---
>  drivers/net/wireless/ath/wil6210/debugfs.c   | 50 +++++++++++++++++++++++++++-
>  drivers/net/wireless/ath/wil6210/interrupt.c |  8 ++++-
>  drivers/net/wireless/ath/wil6210/wil6210.h   |  2 +-
>  3 files changed, 57 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/wireless/ath/wil6210/debugfs.c b/drivers/net/wireless/ath/wil6210/debugfs.c
> index fb802bd..3990a2d 100644
> --- a/drivers/net/wireless/ath/wil6210/debugfs.c
> +++ b/drivers/net/wireless/ath/wil6210/debugfs.c
> @@ -24,6 +24,12 @@
>  #include "wil6210.h"
>  #include "txrx.h"
>  
> +/* Max threshold supported value by wil6210 is 5sec.
> + * The interrupt moderation counter counts down every clock tick of 6nsec.
> + * Therefore max value for ITR_CNT_TRSH can be 5sec/6nsec = 833333333
> + */
> +#define WIL6210_ITR_TRSH_MAX 833333333
> +
>  /* Nasty hack. Better have per device instances */
>  static u32 mem_addr;
>  static u32 dbg_txdesc_index;
> @@ -270,6 +276,48 @@ static struct dentry *wil_debugfs_create_ulong(const char *name, umode_t mode,
>  	return debugfs_create_file(name, mode, parent, value, &wil_fops_ulong);
>  }
>  
> +static int wil_debugfs_itr_trsh_set(void *data, u64 val)
> +{
> +	struct wil6210_priv *wil = data;
> +
> +	/* Check value is smaller than max val limitation
> +	 * (covers also negative values)
> +	 */
> +	if (val > WIL6210_ITR_TRSH_MAX)
> +		return -EINVAL;
> +
> +	/* Perform interrupt moderation threshold update procedure */
> +	iowrite32(0, wil->csr + HOSTADDR(RGF_DMA_ITR_CNT_CRL));
> +	wmb(); /* make sure write propagated to HW */
> +
> +	iowrite32((u32)val, wil->csr + HOSTADDR(RGF_DMA_ITR_CNT_TRSH));
> +	wmb(); /* make sure write propagated to HW */
> +
> +	iowrite32(BIT_DMA_ITR_CNT_CRL_EN,
> +		  wil->csr + HOSTADDR(RGF_DMA_ITR_CNT_CRL));
> +	wmb(); /* make sure write propagated to HW */
> +
> +	return 0;
> +}
> +
> +static int wil_debugfs_itr_trsh_get(void *data, u64 *val)
> +{
> +	struct wil6210_priv *wil = data;
> +	*val = ioread32(wil->csr + HOSTADDR(RGF_DMA_ITR_CNT_TRSH));
> +	return 0;
> +}
> +
> +DEFINE_SIMPLE_ATTRIBUTE(fops_itr_trsh, wil_debugfs_itr_trsh_get,
> +			wil_debugfs_itr_trsh_set, "0x%08llx\n");
> +
> +static struct dentry *debugfs_create_itr_trsh(const char *name,
> +					      umode_t mode,
> +					      struct dentry *parent,
> +					      struct wil6210_priv *wil)
> +{
> +	return debugfs_create_file(name, mode, parent, wil, &fops_itr_trsh);
> +}
> +
>  static int wil6210_debugfs_create_ISR(struct wil6210_priv *wil,
>  				      const char *name,
>  				      struct dentry *parent, u32 off)
> @@ -1024,7 +1072,7 @@ int wil6210_debugfs_init(struct wil6210_priv *wil)
>  	wil_debugfs_create_iomem_x32("MAC_MTRL_COUNTER_0", S_IRUGO, dbg,
>  				     wil->csr +
>  				     HOSTADDR(RGF_MAC_MTRL_COUNTER_0));
> -
> +	debugfs_create_itr_trsh("itr_trsh_val", S_IRUGO | S_IWUSR, dbg, wil);
>  	wil_debugfs_create_iomem_x32("RGF_USER_USAGE_1", S_IRUGO, dbg,
>  				     wil->csr +
>  				     HOSTADDR(RGF_USER_USAGE_1));
> diff --git a/drivers/net/wireless/ath/wil6210/interrupt.c b/drivers/net/wireless/ath/wil6210/interrupt.c
> index 98bfbb6..1f18e86 100644
> --- a/drivers/net/wireless/ath/wil6210/interrupt.c
> +++ b/drivers/net/wireless/ath/wil6210/interrupt.c
> @@ -18,6 +18,12 @@
>  
>  #include "wil6210.h"
>  #include "trace.h"
> +#include <linux/moduleparam.h>
> +
> +static unsigned int itr_trsh = WIL6210_ITR_TRSH;
> +
> +module_param(itr_trsh, uint, S_IRUGO);
> +MODULE_PARM_DESC(itr_trsh, " Interrupt moderation threshold value.");
>  
>  /**
>   * Theory of operation:
> @@ -163,7 +169,7 @@ void wil6210_enable_irq(struct wil6210_priv *wil)
>  		 */
>  		iowrite32(0, wil->csr + HOSTADDR(RGF_DMA_ITR_CNT_CRL));
>  	} else {
> -		iowrite32(WIL6210_ITR_TRSH,
> +		iowrite32((u32)itr_trsh,
>  			  wil->csr + HOSTADDR(RGF_DMA_ITR_CNT_TRSH));
>  		iowrite32(BIT_DMA_ITR_CNT_CRL_EN,
>  			  wil->csr + HOSTADDR(RGF_DMA_ITR_CNT_CRL));
> diff --git a/drivers/net/wireless/ath/wil6210/wil6210.h b/drivers/net/wireless/ath/wil6210/wil6210.h
> index 0097300..47d656a 100644
> --- a/drivers/net/wireless/ath/wil6210/wil6210.h
> +++ b/drivers/net/wireless/ath/wil6210/wil6210.h
> @@ -47,7 +47,7 @@ static inline u32 WIL_GET_BITS(u32 x, int b0, int b1)
>  #define WIL6210_MAX_TX_RINGS	(24) /* HW limit */
>  #define WIL6210_MAX_CID		(8) /* HW limit */
>  #define WIL6210_NAPI_BUDGET	(16) /* arbitrary */
> -#define WIL6210_ITR_TRSH	(10000) /* arbitrary - about 15 IRQs/msec */
> +#define WIL6210_ITR_TRSH	(50000) /* about 3 IRQs/msec */
>  #define WIL6210_FW_RECOVERY_RETRIES	(5) /* try to recover this many times */
>  #define WIL6210_FW_RECOVERY_TO	msecs_to_jiffies(5000)
>  #define WIL6210_SCAN_TO		msecs_to_jiffies(10000)
> 


  reply	other threads:[~2014-08-04 17:39 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-08-04  8:20 [PATCH 00/14] wil6210 misc updates Dedy Lansky
2014-08-04  8:20 ` [PATCH 01/14] wil6210: map MAC timer for packet lifetime into debugfs Dedy Lansky
2014-08-04  8:20 ` [PATCH 02/14] wil6210: fix race in reset Dedy Lansky
2014-08-04  8:20 ` [PATCH 03/14] wil6210: update copyright year 2014 Dedy Lansky
2014-08-04  8:20 ` [PATCH 04/14] wil6210: Optimization for Interrupt moderation threshold value Dedy Lansky
2014-08-04 17:38   ` Florian Fainelli [this message]
2014-08-05  6:15     ` qca_dlansky
2014-08-04  8:20 ` [PATCH 05/14] wil6210: check error in wil_target_reset() Dedy Lansky
2014-08-04  8:20 ` [PATCH 06/14] wil6210: wait longer for hardware reset completion Dedy Lansky
2014-08-04  8:20 ` [PATCH 07/14] wil6210: Workaround for Sparrow with bad device id Dedy Lansky
2014-08-04  8:20 ` [PATCH 08/14] wil6210: convert debugfs to the table mode Dedy Lansky
2014-08-04  8:20 ` [PATCH 09/14] wil6210: fix beamforming data reporting Dedy Lansky
2014-08-04  8:20 ` [PATCH 10/14] wil6210: fix false "scan timeout" Dedy Lansky
2014-08-04  8:20 ` [PATCH 11/14] wil6210: Limit max number of associated stations Dedy Lansky
2014-08-04  8:20 ` [PATCH 12/14] wil6210: fix free'd memory access in wil_if_free() Dedy Lansky
2014-08-04  8:20 ` [PATCH 13/14] wil6210: cfg80211_rx_mgmt to use GFP_ATOMIC Dedy Lansky
2014-08-04  8:20 ` [PATCH 14/14] wil6210: fix access after free in wil_pcie_remove() Dedy Lansky

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=53DFC52A.4030904@gmail.com \
    --to=f.fainelli@gmail.com \
    --cc=linux-wireless@vger.kernel.org \
    --cc=linville@tuxdriver.com \
    --cc=qca_dlansky@qualcomm.com \
    --cc=qca_vkondrat@qca.qualcomm.com \
    --cc=wil6210@qca.qualcomm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).