ice: lower CPU usage with GNSS
Subsystems (3)
v2 cover ice: lower CPU usage with GNSS
This series lowers the CPU usage of the ice driver when using its provided /dev/gnss*.
v2:
- Changed subject of patch 1. Requested by Andrew Lunn.
- Added patch 2 to change the polling interval as recommended by Intel.
- Added patch 3 to remove sq_cmd_timeout as suggested by Simon Horman.
Michal Schmidt (6):
ice: do not busy-wait to read GNSS data
ice: increase the GNSS data polling interval to 20 ms
ice: remove ice_ctl_q_info::sq_cmd_timeout
ice: sleep, don't busy-wait, for ICE_CTL_Q_SQ_CMD_TIMEOUT
ice: remove unused buffer copy code in ice_sq_send_cmd_retry()
ice: sleep, don't busy-wait, in the SQ send retry loop
drivers/net/ethernet/intel/ice/ice_common.c | 29 +++++-------- drivers/net/ethernet/intel/ice/ice_controlq.c | 12 +++--- drivers/net/ethernet/intel/ice/ice_controlq.h | 3 +- drivers/net/ethernet/intel/ice/ice_gnss.c | 42 +++++++++---------- drivers/net/ethernet/intel/ice/ice_gnss.h | 3 +- 5 files changed, 36 insertions(+), 53 deletions(-)
Signature · 6 lines
2.39.2 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
PATCH v2 1/6 ice: do not busy-wait to read GNSS data
The ice-gnss-<dev_name> kernel thread, which reads data from the u-blox GNSS module, keep a CPU core almost 100% busy. The main reason is that it busy-waits for data to become available.
A simple improvement would be to replace the "mdelay(10);" in ice_gnss_read() with sleeping. A better fix is to not do any waiting directly in the function and just requeue this delayed work as needed. The advantage is that canceling the work from ice_gnss_exit() becomes immediate, rather than taking up to ~2.5 seconds (ICE_MAX_UBX_READ_TRIES
* 10 ms).
This lowers the CPU usage of the ice-gnss-<dev_name> thread on my system from ~90 % to ~8 %.
I am not sure if the larger 0.1 s pause after inserting data into the gnss subsystem is really necessary, but I'm keeping that as it was.
Of course, ideally the driver would not have to poll at all, but I don't know if the E810 can watch for GNSS data availability over the i2c bus by itself and notify the driver.
--- drivers/net/ethernet/intel/ice/ice_gnss.c | 42 ++++++++++------------- drivers/net/ethernet/intel/ice/ice_gnss.h | 3 +- 2 files changed, 20 insertions(+), 25 deletions(-)
drivers/net/ethernet/intel/ice/ice_gnss.c +19 -23
@@ -117,6 +117,7 @@ static void ice_gnss_read(struct kthread_work *work) { struct gnss_serial *gnss = container_of(work, struct gnss_serial, read_work.work); +unsigned long delay =ICE_GNSS_POLL_DATA_DELAY_TIME ;unsigned int i, bytes_read, data_len, count;struct ice_aqc_link_topo_addr link_topo;struct ice_pf *pf;
@@ -136,11 +137,6 @@ static void ice_gnss_read(struct kthread_work *work)return ; hw = &pf->hw; - buf = (char *)get_zeroed_page (GFP_KERNEL ); -if (!buf) { - err = -ENOMEM ; - goto exit; - }memset (&link_topo,0 ,sizeof (struct ice_aqc_link_topo_addr )); link_topo.topo_params.index =ICE_E810T_GNSS_I2C_BUS ;
@@ -151,25 +147,24 @@ static void ice_gnss_read(struct kthread_work *work) i2c_params =ICE_GNSS_UBX_DATA_LEN_WIDTH |ICE_AQC_I2C_USE_REPEATED_START ; -/* Read data length in a loop, when it's not 0 the data is ready */ -for (i =0 ; i <ICE_MAX_UBX_READ_TRIES ; i++) { - err =ice_aq_read_i2c (hw, link_topo,ICE_GNSS_UBX_I2C_BUS_ADDR , -cpu_to_le16 (ICE_GNSS_UBX_DATA_LEN_H ), - i2c_params, (u8 *)&data_len_b,NULL ); -if (err) - goto exit_buf; + err =ice_aq_read_i2c (hw, link_topo,ICE_GNSS_UBX_I2C_BUS_ADDR , +cpu_to_le16 (ICE_GNSS_UBX_DATA_LEN_H ), + i2c_params, (u8 *)&data_len_b,NULL ); +if (err) + goto requeue; - data_len =be16_to_cpu (data_len_b); -if (data_len !=0 && data_len !=U16_MAX ) -break ; + data_len =be16_to_cpu (data_len_b); +if (data_len ==0 || data_len ==U16_MAX ) + goto requeue; -mdelay (10 ); - } +/* The u-blox has data_len bytes for us to read */ data_len =min_t (typeof (data_len), data_len,PAGE_SIZE ); -if (!data_len) { + + buf = (char *)get_zeroed_page (GFP_KERNEL ); +if (!buf) { err = -ENOMEM ; - goto exit_buf; + goto requeue; }/* Read received data */
@@ -183,7 +178,7 @@ static void ice_gnss_read(struct kthread_work *work)cpu_to_le16 (ICE_GNSS_UBX_EMPTY_DATA ), bytes_read, &buf[i], NULL);if (err) - goto exit_buf; + goto free_buf; } count =gnss_insert_raw (pf->gnss_dev, buf, i);
@@ -191,10 +186,11 @@ static void ice_gnss_read(struct kthread_work *work)dev_warn (ice_pf_to_dev (pf),"gnss_insert_raw ret=%d size=%d\n" , count, i); -exit_buf: + delay =ICE_GNSS_TIMER_DELAY_TIME ; +free_buf:free_page ((unsigned long )buf); -kthread_queue_delayed_work (gnss->kworker, &gnss->read_work, -ICE_GNSS_TIMER_DELAY_TIME ); +requeue: +kthread_queue_delayed_work (gnss->kworker, &gnss->read_work, delay); exit:if (err)dev_dbg (ice_pf_to_dev (pf),"GNSS failed to read err=%d\n" , err);
drivers/net/ethernet/intel/ice/ice_gnss.h +1 -2
@@ -5,6 +5,7 @@#define _ICE_GNSS_H_#define ICE_E810T_GNSS_I2C_BUS 0x2 +#define ICE_GNSS_POLL_DATA_DELAY_TIME (HZ / 100)/* poll every 10 ms */ #define ICE_GNSS_TIMER_DELAY_TIME (HZ / 10)/* 0.1 second per message */ #define ICE_GNSS_TTY_WRITE_BUF 250#define ICE_MAX_I2C_DATA_SIZE FIELD_MAX(ICE_AQC_I2C_DATA_SIZE_M)
@@ -20,8 +21,6 @@* passed as I2C addr parameter. */ #define ICE_GNSS_UBX_WRITE_BYTES (ICE_MAX_I2C_WRITE_BYTES + 1) -#define ICE_MAX_UBX_READ_TRIES 255 -#define ICE_MAX_UBX_ACK_READ_TRIES 4095struct gnss_write_buf {struct list_head queue;
Signature · 6 lines
2.39.2 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [PATCH net-next v2 1/6] ice: do not busy-wait to read GNSS data
Quoted 143 lines
From: Michal Schmidt <[email protected]>
Sent: Wednesday, April 12, 2023 10:19 AMThe ice-gnss-<dev_name> kernel thread, which reads data from the u-blox GNSS module, keep a CPU core almost 100% busy. The main reason is that it busy-waits for data to become available.
A simple improvement would be to replace the "mdelay(10);" in
ice_gnss_read()with sleeping. A better fix is to not do any waiting directly in the function and just requeue this delayed work as needed. The advantage is that canceling the work fromice_gnss_exit()becomes immediate, rather than taking up to ~2.5 seconds (ICE_MAX_UBX_READ_TRIES
* 10 ms).This lowers the CPU usage of the ice-gnss-<dev_name> thread on my system from ~90 % to ~8 %.
I am not sure if the larger 0.1 s pause after inserting data into the gnss subsystem is really necessary, but I'm keeping that as it was.
Of course, ideally the driver would not have to poll at all, but I don't know if the E810 can watch for GNSS data availability over the i2c bus by itself and notify the driver.
Signed-off-by Michal Schmidt <[email protected]>---
drivers/net/ethernet/intel/ice/ice_gnss.c| 42 ++++++++++-------------drivers/net/ethernet/intel/ice/ice_gnss.h| 3 +-2 files changed, 20 insertions(+), 25 deletions(-)drivers/net/ethernet/intel/ice/ice_gnss.c +19
-23@@ -117,6 +117,7 @@ static void ice_gnss_read(struct kthread_work *work) { struct gnss_serial *gnss = container_of(work, struct gnss_serial, read_work.work); +unsigned long delay =ICE_GNSS_POLL_DATA_DELAY_TIME ;unsigned int i, bytes_read, data_len, count;struct ice_aqc_link_topo_addr link_topo;struct ice_pf *pf;@@ -136,11 +137,6 @@ static void ice_gnss_read(struct kthread_work *work)return ; hw = &pf->hw; - buf = (char *)get_zeroed_page (GFP_KERNEL ); -if (!buf) { - err = -ENOMEM ; - goto exit; - }memset (&link_topo,0 ,sizeof (struct ice_aqc_link_topo_addr )); link_topo.topo_params.index =ICE_E810T_GNSS_I2C_BUS ;@@ -151,25 +147,24 @@ static void ice_gnss_read(struct kthread_work *work) i2c_params =ICE_GNSS_UBX_DATA_LEN_WIDTH |ICE_AQC_I2C_USE_REPEATED_START ; -/* Read data length in a loop, when it's not 0 the data is ready */ -for (i =0 ; i <ICE_MAX_UBX_READ_TRIES ; i++) { - err =ice_aq_read_i2c (hw, link_topo,ICE_GNSS_UBX_I2C_BUS_ADDR , -cpu_to_le16 (ICE_GNSS_UBX_DATA_LEN_H ), - i2c_params, (u8 *)&data_len_b,NULL ); -if (err) - goto exit_buf; + err =ice_aq_read_i2c (hw, link_topo,ICE_GNSS_UBX_I2C_BUS_ADDR , +cpu_to_le16 (ICE_GNSS_UBX_DATA_LEN_H ), + i2c_params, (u8 *)&data_len_b,NULL ); +if (err) + goto requeue; - data_len =be16_to_cpu (data_len_b); -if (data_len !=0 && data_len !=U16_MAX ) -break ; + data_len =be16_to_cpu (data_len_b); +if (data_len ==0 || data_len ==U16_MAX ) + goto requeue; -mdelay (10 ); - } +/* The u-blox has data_len bytes for us to read */ data_len =min_t (typeof (data_len), data_len,PAGE_SIZE ); -if (!data_len) { + + buf = (char *)get_zeroed_page (GFP_KERNEL ); +if (!buf) { err = -ENOMEM ; - goto exit_buf; + goto requeue; }/* Read received data */ @@ -183,7 +178,7 @@ static void ice_gnss_read(struct kthread_work *work)cpu_to_le16 (ICE_GNSS_UBX_EMPTY_DATA ), bytes_read, &buf[i], NULL);if (err) - goto exit_buf; + goto free_buf; } count =gnss_insert_raw (pf->gnss_dev, buf, i);@@ -191,10 +186,11 @@ static void ice_gnss_read(struct kthread_work *work)dev_warn (ice_pf_to_dev (pf),"gnss_insert_raw ret=%d size=%d\n" , count, i); -exit_buf: + delay =ICE_GNSS_TIMER_DELAY_TIME ; +free_buf:free_page ((unsigned long )buf); -kthread_queue_delayed_work (gnss->kworker, &gnss->read_work, -ICE_GNSS_TIMER_DELAY_TIME ); +requeue: +kthread_queue_delayed_work (gnss->kworker, &gnss->read_work, delay);exit:
if (err)
dev_dbg(ice_pf_to_dev(pf), "GNSS failed to read err=%d\n",
err);drivers/net/ethernet/intel/ice/ice_gnss.h +1
-0@@ -5,6 +5,7 @@#define _ICE_GNSS_H_#define ICE_E810T_GNSS_I2C_BUS 0x2 +#define ICE_GNSS_POLL_DATA_DELAY_TIME (HZ / 100)/* poll every 10 ms */ #define ICE_GNSS_TIMER_DELAY_TIME (HZ / 10)/* 0.1 second per message */ #define ICE_GNSS_TTY_WRITE_BUF 250#define ICE_MAX_I2C_DATA_SIZE FIELD_MAX(ICE_AQC_I2C_DATA_SIZE_M)drivers/net/ethernet/intel/ice/ice_gnss.h +0
-3@@ -20,8 +21,6 @@* passed as I2C addr parameter. */ #define ICE_GNSS_UBX_WRITE_BYTES (ICE_MAX_I2C_WRITE_BYTES + 1) -#define ICE_MAX_UBX_READ_TRIES 255 -#define ICE_MAX_UBX_ACK_READ_TRIES 4095struct gnss_write_buf {struct list_head queue; --
Quoted 1 line
2.39.2
Looks good, thank you Michal!
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [PATCH net-next v2 1/6] ice: do not busy-wait to read GNSS data
In reply to Michal SchmidtQuoted 14 lines · Michal Schmidt
The ice-gnss-<dev_name> kernel thread, which reads data from the u-blox GNSS module, keep a CPU core almost 100% busy. The main reason is that it busy-waits for data to become available.
A simple improvement would be to replace the "mdelay(10);" in
ice_gnss_read()with sleeping. A better fix is to not do any waiting directly in the function and just requeue this delayed work as needed. The advantage is that canceling the work fromice_gnss_exit()becomes immediate, rather than taking up to ~2.5 seconds (ICE_MAX_UBX_READ_TRIES
* 10 ms).This lowers the CPU usage of the ice-gnss-<dev_name> thread on my system from ~90 % to ~8 %.
Quoted 8 lines
I am not sure if the larger 0.1 s pause after inserting data into the gnss subsystem is really necessary, but I'm keeping that as it was.
Of course, ideally the driver would not have to poll at all, but I don't know if the E810 can watch for GNSS data availability over the i2c bus by itself and notify the driver.
Signed-off-by Michal Schmidt <[email protected]>
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [Intel-wired-lan] [PATCH net-next v2 1/6] ice: do not busy-wait to read GNSS data
In reply to Michal SchmidtQuoted 10 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Michal Schmidt
Sent: Wednesday, April 12, 2023 1:19 AM
To: [email protected]
Cc: Andrew Lunn <[email protected]>; [email protected]; Brandeburg, Jesse <[email protected]>; Kolacinski, Karol <[email protected]>; Nguyen, Anthony L <[email protected]>; Simon Horman <[email protected]>
Subject: [Intel-wired-lan] [PATCH net-next v2 1/6] ice: do not busy-wait to read GNSS dataThe ice-gnss-<dev_name> kernel thread, which reads data from the u-blox GNSS module, keep a CPU core almost 100% busy. The main reason is that it busy-waits for data to become available.
A simple improvement would be to replace the "mdelay(10);" in
ice_gnss_read() with sleeping. A better fix is to not do any waiting directly in the function and just requeue this delayed work as needed. The advantage is that canceling the work from ice_gnss_exit() becomes immediate, rather than taking up to ~2.5 seconds (ICE_MAX_UBX_READ_TRIES
* 10 ms).
Quoted 12 lines
This lowers the CPU usage of the ice-gnss-<dev_name> thread on my system from ~90 % to ~8 %.
I am not sure if the larger 0.1 s pause after inserting data into the gnss subsystem is really necessary, but I'm keeping that as it was.
Of course, ideally the driver would not have to poll at all, but I don't know if the E810 can watch for GNSS data availability over the i2c bus by itself and notify the driver.
Signed-off-by Michal Schmidt <[email protected]>---
drivers/net/ethernet/intel/ice/ice_gnss.c| 42 ++++++++++-------------drivers/net/ethernet/intel/ice/ice_gnss.h| 3 +-2 files changed, 20 insertions(+), 25 deletions(-)
PATCH v2 2/6 ice: increase the GNSS data polling interval to 20 ms
In reply to Michal SchmidtDouble the GNSS data polling interval from 10 ms to 20 ms. According to Karol Kolacinski from the Intel team, they have been planning to make this change.
--- drivers/net/ethernet/intel/ice/ice_gnss.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/intel/ice/ice_gnss.h +1 -1
@@ -5,7 +5,7 @@#define _ICE_GNSS_H_#define ICE_E810T_GNSS_I2C_BUS 0x2 -#define ICE_GNSS_POLL_DATA_DELAY_TIME (HZ / 100)/* poll every 10 ms */ +#define ICE_GNSS_POLL_DATA_DELAY_TIME (HZ / 50)/* poll every 20 ms */ #define ICE_GNSS_TIMER_DELAY_TIME (HZ / 10)/* 0.1 second per message */ #define ICE_GNSS_TTY_WRITE_BUF 250#define ICE_MAX_I2C_DATA_SIZE FIELD_MAX(ICE_AQC_I2C_DATA_SIZE_M)
Signature · 6 lines
2.39.2 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [PATCH net-next v2 2/6] ice: increase the GNSS data polling interval to 20 ms
Quoted 27 lines
From: Michal Schmidt <[email protected]>
Sent: Wednesday, April 12, 2023 10:19 AMDouble the GNSS data polling interval from 10 ms to 20 ms. According to Karol Kolacinski from the Intel team, they have been planning to make this change.
Signed-off-by Michal Schmidt <[email protected]>---
drivers/net/ethernet/intel/ice/ice_gnss.h| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)drivers/net/ethernet/intel/ice/ice_gnss.h +1
-1@@ -5,7 +5,7 @@#define _ICE_GNSS_H_#define ICE_E810T_GNSS_I2C_BUS 0x2 -#define ICE_GNSS_POLL_DATA_DELAY_TIME (HZ / 100)/* poll every 10 ms */ +#define ICE_GNSS_POLL_DATA_DELAY_TIME (HZ / 50)/* poll every 20 ms */ #define ICE_GNSS_TIMER_DELAY_TIME (HZ / 10)/* 0.1 second per message */ #define ICE_GNSS_TTY_WRITE_BUF 250#define ICE_MAX_I2C_DATA_SIZE FIELD_MAX(ICE_AQC_I2C_DATA_SIZE_M)
Quoted 2 lines
--
2.39.2
Looks good, thank you Michal!
Re: [PATCH net-next v2 2/6] ice: increase the GNSS data polling interval to 20 ms
In reply to Michal SchmidtQuoted 5 lines · Michal Schmidt
Double the GNSS data polling interval from 10 ms to 20 ms. According to Karol Kolacinski from the Intel team, they have been planning to make this change.
Signed-off-by Michal Schmidt <[email protected]>
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [Intel-wired-lan] [PATCH net-next v2 2/6] ice: increase the GNSS data polling interval to 20 ms
In reply to Michal SchmidtQuoted 7 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Michal Schmidt
Sent: Wednesday, April 12, 2023 1:19 AM
To: [email protected]
Cc: Andrew Lunn <[email protected]>; [email protected]; Brandeburg, Jesse <[email protected]>; Kolacinski, Karol <[email protected]>; Nguyen, Anthony L <[email protected]>; Simon Horman <[email protected]>
Subject: [Intel-wired-lan] [PATCH net-next v2 2/6] ice: increase the GNSS data polling interval to 20 ms
Quoted 8 lines
Double the GNSS data polling interval from 10 ms to 20 ms.
According to Karol Kolacinski from the Intel team, they have been planning to make this change.Signed-off-by Michal Schmidt <[email protected]>--- drivers/net/ethernet/intel/ice/ice_gnss.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
PATCH v2 3/6 ice: remove ice_ctl_q_info::sq_cmd_timeout
In reply to Michal Schmidtsq_cmd_timeout is initialized to ICE_CTL_Q_SQ_CMD_TIMEOUT and never changed, so just use the constant directly.
--- drivers/net/ethernet/intel/ice/ice_common.c | 2 +- drivers/net/ethernet/intel/ice/ice_controlq.c | 5 +---- drivers/net/ethernet/intel/ice/ice_controlq.h | 1 - 3 files changed, 2 insertions(+), 6 deletions(-)
drivers/net/ethernet/intel/ice/ice_common.c +1 -1
@@ -2000,7 +2000,7 @@ void ice_release_res(struct ice_hw *hw, enum ice_aq_res_ids res)/* there are some rare cases when trying to release the resource * results in an admin queue timeout, so handle them correctly */ -while ((status == -EIO ) && (total_delay < hw->adminq.sq_cmd_timeout)) { +while ((status == -EIO ) && (total_delay <ICE_CTL_Q_SQ_CMD_TIMEOUT )) {mdelay (1 ); status =ice_aq_release_res (hw, res,0 ,NULL ); total_delay++;
drivers/net/ethernet/intel/ice/ice_controlq.c +1 -4
@@ -637,9 +637,6 @@ static int ice_init_ctrlq(struct ice_hw *hw, enum ice_ctl_q q_type)return -EIO ; } -/* setup SQ command write back timeout */ - cq->sq_cmd_timeout =ICE_CTL_Q_SQ_CMD_TIMEOUT ; -/* allocate the ATQ */ ret_code =ice_init_sq (hw, cq);if (ret_code)
@@ -1066,7 +1063,7 @@ ice_sq_send_cmd(struct ice_hw *hw, struct ice_ctl_q_info *cq,udelay (ICE_CTL_Q_SQ_CMD_USEC ); total_delay++; - }while (total_delay < cq->sq_cmd_timeout); + }while (total_delay <ICE_CTL_Q_SQ_CMD_TIMEOUT );/* if ready, copy the desc back to temp */ if (ice_sq_done (hw, cq)) {
drivers/net/ethernet/intel/ice/ice_controlq.h +0 -1
@@ -87,7 +87,6 @@ struct ice_ctl_q_info {enum ice_ctl_q qtype;struct ice_ctl_q_ring rq;/* receive queue */ struct ice_ctl_q_ring sq;/* send queue */ -u32 sq_cmd_timeout;/* send queue cmd write back timeout */ u16 num_rq_entries;/* receive queue depth */ u16 num_sq_entries;/* send queue depth */ u16 rq_buf_size;/* receive queue buffer size */
Signature · 6 lines
2.39.2 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [PATCH net-next v2 3/6] ice: remove ice_ctl_q_info::sq_cmd_timeout
Quoted 71 lines
From: Michal Schmidt <[email protected]>
Sent: Wednesday, April 12, 2023 10:19 AM
sq_cmd_timeoutis initialized toICE_CTL_Q_SQ_CMD_TIMEOUTand never changed, so just use the constant directly.Suggested-by Simon Horman <[email protected]>Signed-off-by Michal Schmidt <[email protected]>---
drivers/net/ethernet/intel/ice/ice_common.c | 2 +- drivers/net/ethernet/intel/ice/ice_controlq.c | 5 +---- drivers/net/ethernet/intel/ice/ice_controlq.h | 1 - 3 files changed, 2 insertions(+), 6 deletions(-)drivers/net/ethernet/intel/ice/ice_common.c +1
-1@@ -2000,7 +2000,7 @@ void ice_release_res(struct ice_hw *hw, enum ice_aq_res_ids res)/* there are some rare cases when trying to release the resource * results in an admin queue timeout, so handle them correctly */ -while ((status == -EIO ) && (total_delay < hw->adminq.sq_cmd_timeout)) { +while ((status == -EIO ) && (total_delay <ICE_CTL_Q_SQ_CMD_TIMEOUT )) {mdelay(1); status = ice_aq_release_res(hw, res, 0, NULL); total_delay++;drivers/net/ethernet/intel/ice/ice_controlq.c +0
-3@@ -637,9 +637,6 @@ static int ice_init_ctrlq(struct ice_hw *hw, enum ice_ctl_q q_type)return -EIO ; } -/* setup SQ command write back timeout */ - cq->sq_cmd_timeout =ICE_CTL_Q_SQ_CMD_TIMEOUT ; -/* allocate the ATQ */ ret_code =ice_init_sq (hw, cq);if (
ret_code)drivers/net/ethernet/intel/ice/ice_controlq.c +1
-1@@ -1066,7 +1063,7 @@ ice_sq_send_cmd(struct ice_hw *hw, structice_ctl_q_info *cq,udelay (ICE_CTL_Q_SQ_CMD_USEC ); total_delay++; - }while (total_delay < cq->sq_cmd_timeout); + }while (total_delay <ICE_CTL_Q_SQ_CMD_TIMEOUT );/* if ready, copy the desc back to temp */ if (ice_sq_done(hw, cq)) {drivers/net/ethernet/intel/ice/ice_controlq.h +0
-2@@ -87,7 +87,6 @@ struct ice_ctl_q_info {enum ice_ctl_q qtype;struct ice_ctl_q_ring rq;/* receive queue */ struct ice_ctl_q_ring sq;/* send queue */ -u32 sq_cmd_timeout;/* send queue cmd write back timeout */ u16 num_rq_entries;/* receive queue depth */ u16 num_sq_entries;/* send queue depth */ u16 rq_buf_size;/* receive queue buffer size */ --
Quoted 1 line
2.39.2
Looks good, thank you Michal!
Re: [PATCH net-next v2 3/6] ice: remove ice_ctl_q_info::sq_cmd_timeout
In reply to Michal SchmidtQuoted 5 lines · Michal Schmidt
sq_cmd_timeoutis initialized toICE_CTL_Q_SQ_CMD_TIMEOUTand never changed, so just use the constant directly.Suggested-by Simon Horman <[email protected]>Signed-off-by Michal Schmidt <[email protected]>
LGTM, thanks.
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [Intel-wired-lan] [PATCH net-next v2 3/6] ice: remove ice_ctl_q_info::sq_cmd_timeout
In reply to Michal SchmidtQuoted 7 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Michal Schmidt
Sent: Wednesday, April 12, 2023 1:19 AM
To: [email protected]
Cc: Andrew Lunn <[email protected]>; [email protected]; Brandeburg, Jesse <[email protected]>; Kolacinski, Karol <[email protected]>; Nguyen, Anthony L <[email protected]>; Simon Horman <[email protected]>
Subject: [Intel-wired-lan] [PATCH net-next v2 3/6] ice: removeice_ctl_q_info::sq_cmd_timeout
Quoted 9 lines
sq_cmd_timeoutis initialized toICE_CTL_Q_SQ_CMD_TIMEOUTand never changed, so just use the constant directly.Suggested-by Simon Horman <[email protected]>Signed-off-by Michal Schmidt <[email protected]>--- drivers/net/ethernet/intel/ice/ice_common.c | 2 +- drivers/net/ethernet/intel/ice/ice_controlq.c | 5 +---- drivers/net/ethernet/intel/ice/ice_controlq.h | 1 - 3 files changed, 2 insertions(+), 6 deletions(-)
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
PATCH v2 4/6 ice: sleep, don't busy-wait, for ICE_CTL_Q_SQ_CMD_TIMEOUT
In reply to Michal SchmidtThe driver polls for ice_sq_done() with a 100 µs period for up to 1 s and it uses udelay to do that.
Let's use usleep_range instead. We know sleeping is allowed here, because we're holding a mutex (cq->sq_lock). To preserve the total max waiting time, measure the timeout in jiffies.
ICE_CTL_Q_SQ_CMD_TIMEOUT is used also in ice_release_res(), but there the polling period is 1 ms (i.e. 10 times longer). Since the timeout was expressed in terms of the number of loops, the total timeout in this function is 10 s. I do not know if this is intentional. This patch keeps it.
The patch lowers the CPU usage of the ice-gnss-<dev_name> kernel thread on my system from ~8 % to less than 1 %.
I received a report of high CPU usage with ptp4l where the busy-waiting in ice_sq_send_cmd dominated the profile. This patch has been tested in that usecase too and it made a huge improvement there.
--- drivers/net/ethernet/intel/ice/ice_common.c | 14 +++++++------- drivers/net/ethernet/intel/ice/ice_controlq.c | 9 +++++---- drivers/net/ethernet/intel/ice/ice_controlq.h | 2 +- 3 files changed, 13 insertions(+), 12 deletions(-)
drivers/net/ethernet/intel/ice/ice_common.c +7 -7
@@ -1992,19 +1992,19 @@ ice_acquire_res(struct ice_hw *hw, enum ice_aq_res_ids res,*/ void ice_release_res (struct ice_hw *hw,enum ice_aq_res_ids res) { -u32 total_delay =0 ; +unsigned long timeout;int status; - status =ice_aq_release_res (hw, res,0 ,NULL ); -/* there are some rare cases when trying to release the resource * results in an admin queue timeout, so handle them correctly */ -while ((status == -EIO ) && (total_delay <ICE_CTL_Q_SQ_CMD_TIMEOUT )) { -mdelay (1 ); + timeout = jiffies +10 *ICE_CTL_Q_SQ_CMD_TIMEOUT ; +do { status =ice_aq_release_res (hw, res,0 ,NULL ); - total_delay++; - } +if (status != -EIO ) +break ; +usleep_range (1000 ,2000 ); + }while (time_before (jiffies, timeout)); }/**
drivers/net/ethernet/intel/ice/ice_controlq.c +5 -4
@@ -964,7 +964,7 @@ ice_sq_send_cmd(struct ice_hw *hw, struct ice_ctl_q_info *cq,struct ice_aq_desc *desc_on_ring;bool cmd_completed = false;struct ice_sq_cd *details; -u32 total_delay =0 ; +unsigned long timeout;int status =0 ;u16 retval =0 ;u32 val =0 ;
@@ -1057,13 +1057,14 @@ ice_sq_send_cmd(struct ice_hw *hw, struct ice_ctl_q_info *cq, cq->sq.next_to_use =0 ;wr32 (hw, cq->sq.tail, cq->sq.next_to_use); + timeout = jiffies +ICE_CTL_Q_SQ_CMD_TIMEOUT ;do {if (ice_sq_done (hw, cq))break ; -udelay (ICE_CTL_Q_SQ_CMD_USEC ); - total_delay++; - }while (total_delay <ICE_CTL_Q_SQ_CMD_TIMEOUT ); +usleep_range (ICE_CTL_Q_SQ_CMD_USEC , +ICE_CTL_Q_SQ_CMD_USEC *3 /2 ); + }while (time_before (jiffies, timeout));/* if ready, copy the desc back to temp */ if (ice_sq_done (hw, cq)) {
drivers/net/ethernet/intel/ice/ice_controlq.h +1 -1
@@ -34,7 +34,7 @@ enum ice_ctl_q { };/* Control Queue timeout settings - max delay 1s */ -#define ICE_CTL_Q_SQ_CMD_TIMEOUT 10000/* Count 10000 times */ +#define ICE_CTL_Q_SQ_CMD_TIMEOUT HZ/* Wait max 1s */ #define ICE_CTL_Q_SQ_CMD_USEC 100/* Check every 100usec */ #define ICE_CTL_Q_ADMIN_INIT_TIMEOUT 10/* Count 10 times */ #define ICE_CTL_Q_ADMIN_INIT_MSEC 100/* Check every 100msec */
Signature · 6 lines
2.39.2 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [PATCH net-next v2 4/6] ice: sleep, don't busy-wait, for ICE_CTL_Q_SQ_CMD_TIMEOUT
Quoted 114 lines
From: Michal Schmidt <[email protected]>
Sent: Wednesday, April 12, 2023 10:19 AMThe driver polls for
ice_sq_done()with a 100 µs period for up to 1 s and it uses udelay to do that.Let's use
usleep_rangeinstead. We know sleeping is allowed here, because we're holding a mutex (cq->sq_lock). To preserve the total max waiting time, measure the timeout in jiffies.
ICE_CTL_Q_SQ_CMD_TIMEOUTis used also inice_release_res(), but there the polling period is 1 ms (i.e. 10 times longer). Since the timeout was expressed in terms of the number of loops, the total timeout in this function is 10 s. I do not know if this is intentional. This patch keeps it.The patch lowers the CPU usage of the ice-gnss-<dev_name> kernel thread on my system from ~8 % to less than 1 %.
I received a report of high CPU usage with ptp4l where the busy-waiting in
ice_sq_send_cmddominated the profile. This patch has been tested in that usecase too and it made a huge improvement there.Tested-by Brent Rowsell <[email protected]>Signed-off-by Michal Schmidt <[email protected]>---
drivers/net/ethernet/intel/ice/ice_common.c | 14 +++++++------- drivers/net/ethernet/intel/ice/ice_controlq.c | 9 +++++---- drivers/net/ethernet/intel/ice/ice_controlq.h | 2 +- 3 files changed, 13 insertions(+), 12 deletions(-)drivers/net/ethernet/intel/ice/ice_common.c +7
-7@@ -1992,19 +1992,19 @@ ice_acquire_res(struct ice_hw *hw, enumice_aq_res_ids res, */ void ice_release_res (struct ice_hw *hw,enum ice_aq_res_ids res) { -u32 total_delay =0 ; +unsigned long timeout;int status; - status =ice_aq_release_res (hw, res,0 ,NULL ); -/* there are some rare cases when trying to release the resource * results in an admin queue timeout, so handle them correctly */ -while ((status == -EIO ) && (total_delay <ICE_CTL_Q_SQ_CMD_TIMEOUT )) { -mdelay (1 ); + timeout = jiffies +10 *ICE_CTL_Q_SQ_CMD_TIMEOUT ; +do { status =ice_aq_release_res (hw, res,0 ,NULL ); - total_delay++; - } +if (status != -EIO ) +break ; +usleep_range (1000 ,2000 ); + }while (time_before (jiffies, timeout)); }/**
drivers/net/ethernet/intel/ice/ice_controlq.c +1
-1@@ -964,7 +964,7 @@ ice_sq_send_cmd(struct ice_hw *hw, structice_ctl_q_info *cq, struct ice_aq_desc *desc_on_ring;bool cmd_completed = false;struct ice_sq_cd *details; -u32 total_delay =0 ; +unsigned long timeout;int status =0 ;u16 retval =0 ;u32 val = 0;drivers/net/ethernet/intel/ice/ice_controlq.c +4
-3@@ -1057,13 +1057,14 @@ ice_sq_send_cmd(struct ice_hw *hw, struct ice_ctl_q_info *cq, cq->sq.next_to_use =0 ;wr32 (hw, cq->sq.tail, cq->sq.next_to_use); + timeout = jiffies +ICE_CTL_Q_SQ_CMD_TIMEOUT ;do {if (ice_sq_done (hw, cq))break ; -udelay (ICE_CTL_Q_SQ_CMD_USEC ); - total_delay++; - }while (total_delay <ICE_CTL_Q_SQ_CMD_TIMEOUT ); +usleep_range (ICE_CTL_Q_SQ_CMD_USEC , +ICE_CTL_Q_SQ_CMD_USEC *3 /2 ); + }while (time_before (jiffies, timeout));/* if ready, copy the desc back to temp */ if (ice_sq_done(hw, cq)) {drivers/net/ethernet/intel/ice/ice_controlq.h +1
-2@@ -34,7 +34,7 @@ enum ice_ctl_q { };/* Control Queue timeout settings - max delay 1s */ -#define ICE_CTL_Q_SQ_CMD_TIMEOUT 10000/* Count 10000 times */ +#define ICE_CTL_Q_SQ_CMD_TIMEOUT HZ/* Wait max 1s */ #define ICE_CTL_Q_SQ_CMD_USEC 100/* Check every 100usec */ #define ICE_CTL_Q_ADMIN_INIT_TIMEOUT 10/* Count 10 times */ #define ICE_CTL_Q_ADMIN_INIT_MSEC 100/* Check every 100msec */ --
Quoted 1 line
2.39.2
Looks good, thank you Michal!
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [PATCH net-next v2 4/6] ice: sleep, don't busy-wait, for ICE_CTL_Q_SQ_CMD_TIMEOUT
In reply to Michal SchmidtQuoted 16 lines · Michal Schmidt
The driver polls for
ice_sq_done()with a 100 µs period for up to 1 s and it uses udelay to do that.Let's use
usleep_rangeinstead. We know sleeping is allowed here, because we're holding a mutex (cq->sq_lock). To preserve the total max waiting time, measure the timeout in jiffies.
ICE_CTL_Q_SQ_CMD_TIMEOUTis used also inice_release_res(), but there the polling period is 1 ms (i.e. 10 times longer). Since the timeout was expressed in terms of the number of loops, the total timeout in this function is 10 s. I do not know if this is intentional. This patch keeps it.The patch lowers the CPU usage of the ice-gnss-<dev_name> kernel thread on my system from ~8 % to less than 1 %.
Quoted 6 lines
I received a report of high CPU usage with ptp4l where the busy-waiting in
ice_sq_send_cmddominated the profile. This patch has been tested in that usecase too and it made a huge improvement there.Tested-by Brent Rowsell <[email protected]>Signed-off-by Michal Schmidt <[email protected]>
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [Intel-wired-lan] [PATCH net-next v2 4/6] ice: sleep, don't busy-wait, for ICE_CTL_Q_SQ_CMD_TIMEOUT
In reply to Michal SchmidtQuoted 15 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Michal Schmidt
Sent: Wednesday, April 12, 2023 1:19 AM
To: [email protected]
Cc: Brent Rowsell <[email protected]>; Andrew Lunn <[email protected]>; [email protected]; Brandeburg, Jesse <[email protected]>; Kolacinski, Karol <[email protected]>; Nguyen, Anthony L <[email protected]>; Simon Horman <[email protected]>
Subject: [Intel-wired-lan] [PATCH net-next v2 4/6] ice: sleep, don't busy-wait, forICE_CTL_Q_SQ_CMD_TIMEOUTThe driver polls for
ice_sq_done()with a 100 µs period for up to 1 s and it uses udelay to do that.Let's use
usleep_rangeinstead. We know sleeping is allowed here, because we're holding a mutex (cq->sq_lock). To preserve the total max waiting time, measure the timeout in jiffies.
ICE_CTL_Q_SQ_CMD_TIMEOUTis used also inice_release_res(), but there the polling period is 1 ms (i.e. 10 times longer). Since the timeout was expressed in terms of the number of loops, the total timeout in this function is 10 s. I do not know if this is intentional. This patch keeps it.The patch lowers the CPU usage of the ice-gnss-<dev_name> kernel thread on my system from ~8 % to less than 1 %.
Quoted 9 lines
I received a report of high CPU usage with ptp4l where the busy-waiting in
ice_sq_send_cmddominated the profile. This patch has been tested in that usecase too and it made a huge improvement there.Tested-by Brent Rowsell <[email protected]>Signed-off-by Michal Schmidt <[email protected]>---
drivers/net/ethernet/intel/ice/ice_common.c| 14 +++++++-------drivers/net/ethernet/intel/ice/ice_controlq.c| 9 +++++----drivers/net/ethernet/intel/ice/ice_controlq.h| 2 +-3 files changed, 13 insertions(+), 12 deletions(-)
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
PATCH v2 5/6 ice: remove unused buffer copy code in ice_sq_send_cmd_retry()
In reply to Michal SchmidtThe 'buf_cpy'-related code in ice_sq_send_cmd_retry() looks broken. 'buf' is nowhere copied into 'buf_cpy'.
The reason this does not cause problems is that all commands for which 'is_cmd_for_retry' is true go with a NULL buf.
Let's remove 'buf_cpy'. Add a WARN_ON in case the assumption no longer holds in the future.
--- drivers/net/ethernet/intel/ice/ice_common.c | 13 ++----------- 1 file changed, 2 insertions(+), 11 deletions(-)
drivers/net/ethernet/intel/ice/ice_common.c +2 -11
@@ -1619,7 +1619,6 @@ ice_sq_send_cmd_retry(struct ice_hw *hw, struct ice_ctl_q_info *cq, {struct ice_aq_desc desc_cpy;bool is_cmd_for_retry; -u8 *buf_cpy =NULL ;u8 idx =0 ;u16 opcode;int status;
@@ -1629,11 +1628,8 @@ ice_sq_send_cmd_retry(struct ice_hw *hw, struct ice_ctl_q_info *cq,memset (&desc_cpy,0 ,sizeof (desc_cpy));if (is_cmd_for_retry) { -if (buf) { - buf_cpy =kzalloc (buf_size,GFP_KERNEL ); -if (!buf_cpy) -return -ENOMEM ; - } +/* All retryable cmds are direct, without buf. */ +WARN_ON (buf);memcpy (&desc_cpy, desc,sizeof (desc_cpy)); }
@@ -1645,17 +1641,12 @@ ice_sq_send_cmd_retry(struct ice_hw *hw, struct ice_ctl_q_info *cq, hw->adminq.sq_last_status != ICE_AQ_RC_EBUSY) break; -if (buf_cpy) -memcpy (buf, buf_cpy, buf_size); -memcpy (desc, &desc_cpy,sizeof (desc_cpy));mdelay (ICE_SQ_SEND_DELAY_TIME_MS ); }while (++idx <ICE_SQ_SEND_MAX_EXECUTE ); -kfree (buf_cpy); -return status; }
Signature · 6 lines
2.39.2 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [PATCH net-next v2 5/6] ice: remove unused buffer copy code in ice_sq_send_cmd_retry()
Quoted 66 lines
From: Michal Schmidt <[email protected]>
Sent: Wednesday, April 12, 2023 10:19 AMThe 'buf_cpy'-related code in
ice_sq_send_cmd_retry()looks broken. 'buf' is nowhere copied into 'buf_cpy'.The reason this does not cause problems is that all commands for which '
is_cmd_for_retry' is true go with a NULL buf.Let's remove '
buf_cpy'. Add aWARN_ONin case the assumption no longer holds in the future.Signed-off-by Michal Schmidt <[email protected]>---
drivers/net/ethernet/intel/ice/ice_common.c| 13 ++-----------1 file changed, 2 insertions(+), 11 deletions(-)drivers/net/ethernet/intel/ice/ice_common.c +0
-1@@ -1619,7 +1619,6 @@ ice_sq_send_cmd_retry(struct ice_hw *hw, struct ice_ctl_q_info *cq, {struct ice_aq_desc desc_cpy;bool is_cmd_for_retry; -u8 *buf_cpy =NULL ;u8 idx =0 ;u16 opcode;int status;drivers/net/ethernet/intel/ice/ice_common.c +2
-5@@ -1629,11 +1628,8 @@ ice_sq_send_cmd_retry(struct ice_hw *hw, struct ice_ctl_q_info *cq,memset (&desc_cpy,0 ,sizeof (desc_cpy));if (is_cmd_for_retry) { -if (buf) { - buf_cpy =kzalloc (buf_size,GFP_KERNEL ); -if (!buf_cpy) -return -ENOMEM ; - } +/* All retryable cmds are direct, without buf. */ +WARN_ON (buf);memcpy (&desc_cpy, desc,sizeof (desc_cpy));}drivers/net/ethernet/intel/ice/ice_common.c +0
-6@@ -1645,17 +1641,12 @@ ice_sq_send_cmd_retry(struct ice_hw *hw, struct ice_ctl_q_info *cq, hw->adminq.sq_last_status != ICE_AQ_RC_EBUSY) break; -if (buf_cpy) -memcpy (buf, buf_cpy, buf_size); -memcpy (desc, &desc_cpy,sizeof (desc_cpy));mdelay (ICE_SQ_SEND_DELAY_TIME_MS ); }while (++idx <ICE_SQ_SEND_MAX_EXECUTE ); -kfree (buf_cpy); -return status; } --
Quoted 1 line
2.39.2
Looks good, thank you Michal!
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [PATCH net-next v2 5/6] ice: remove unused buffer copy code in ice_sq_send_cmd_retry()
In reply to Michal SchmidtQuoted 10 lines · Michal Schmidt
The 'buf_cpy'-related code in
ice_sq_send_cmd_retry()looks broken. 'buf' is nowhere copied into 'buf_cpy'.The reason this does not cause problems is that all commands for which '
is_cmd_for_retry' is true go with a NULL buf.Let's remove '
buf_cpy'. Add aWARN_ONin case the assumption no longer holds in the future.Signed-off-by Michal Schmidt <[email protected]>
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [Intel-wired-lan] [PATCH net-next v2 5/6] ice: remove unused buffer copy code in ice_sq_send_cmd_retry()
In reply to Michal SchmidtQuoted 10 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Michal Schmidt
Sent: Wednesday, April 12, 2023 1:19 AM
To: [email protected]
Cc: Andrew Lunn <[email protected]>; [email protected]; Brandeburg, Jesse <[email protected]>; Kolacinski, Karol <[email protected]>; Nguyen, Anthony L <[email protected]>; Simon Horman <[email protected]>
Subject: [Intel-wired-lan] [PATCH net-next v2 5/6] ice: remove unused buffer copy code inice_sq_send_cmd_retry()The 'buf_cpy'-related code in
ice_sq_send_cmd_retry()looks broken. 'buf' is nowhere copied into 'buf_cpy'.
Quoted 9 lines
The reason this does not cause problems is that all commands for which '
is_cmd_for_retry' is true go with a NULL buf.Let's remove '
buf_cpy'. Add aWARN_ONin case the assumption no longer holds in the future.Signed-off-by Michal Schmidt <[email protected]>--- drivers/net/ethernet/intel/ice/ice_common.c | 13 ++----------- 1 file changed, 2 insertions(+), 11 deletions(-)
PATCH v2 6/6 ice: sleep, don't busy-wait, in the SQ send retry loop
In reply to Michal Schmidt10 ms is a lot of time to spend busy-waiting. Sleeping is clearly allowed here, because we have just returned from ice_sq_send_cmd(), which takes a mutex.
On kernels with HZ=100, this msleep may be twice as long, but I don't think it matters.
I did not actually observe any retries happening here.
--- drivers/net/ethernet/intel/ice/ice_common.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/intel/ice/ice_common.c +1 -1
@@ -1643,7 +1643,7 @@ ice_sq_send_cmd_retry(struct ice_hw *hw, struct ice_ctl_q_info *cq,memcpy (desc, &desc_cpy,sizeof (desc_cpy)); -mdelay (ICE_SQ_SEND_DELAY_TIME_MS ); +msleep (ICE_SQ_SEND_DELAY_TIME_MS ); }while (++idx <ICE_SQ_SEND_MAX_EXECUTE );
Signature · 6 lines
2.39.2 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [PATCH net-next v2 6/6] ice: sleep, don't busy-wait, in the SQ send retry loop
Quoted 32 lines
From: Michal Schmidt <[email protected]>
Sent: Wednesday, April 12, 2023 10:19 AM10 ms is a lot of time to spend busy-waiting. Sleeping is clearly allowed here, because we have just returned from
ice_sq_send_cmd(), which takes a mutex.On kernels with HZ=100, this msleep may be twice as long, but I don't think it matters.
I did not actually observe any retries happening here.Signed-off-by Michal Schmidt <[email protected]>---
drivers/net/ethernet/intel/ice/ice_common.c| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)drivers/net/ethernet/intel/ice/ice_common.c +1
-2@@ -1643,7 +1643,7 @@ ice_sq_send_cmd_retry(struct ice_hw *hw, struct ice_ctl_q_info *cq,memcpy (desc, &desc_cpy,sizeof (desc_cpy)); -mdelay (ICE_SQ_SEND_DELAY_TIME_MS ); +msleep (ICE_SQ_SEND_DELAY_TIME_MS ); }while (++idx <ICE_SQ_SEND_MAX_EXECUTE ); --
Quoted 1 line
2.39.2
Looks good, thank you Michal!
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [PATCH net-next v2 6/6] ice: sleep, don't busy-wait, in the SQ send retry loop
In reply to Michal SchmidtQuoted 9 lines · Michal Schmidt
10 ms is a lot of time to spend busy-waiting. Sleeping is clearly allowed here, because we have just returned from
ice_sq_send_cmd(), which takes a mutex.On kernels with HZ=100, this msleep may be twice as long, but I don't think it matters.
I did not actually observe any retries happening here.Signed-off-by Michal Schmidt <[email protected]>
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
RE: [Intel-wired-lan] [PATCH net-next v2 6/6] ice: sleep, don't busy-wait, in the SQ send retry loop
In reply to Michal SchmidtQuoted 9 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Michal Schmidt
Sent: Wednesday, April 12, 2023 1:19 AM
To: [email protected]
Cc: Andrew Lunn <[email protected]>; [email protected]; Brandeburg, Jesse <[email protected]>; Kolacinski, Karol <[email protected]>; Nguyen, Anthony L <[email protected]>; Simon Horman <[email protected]>
Subject: [Intel-wired-lan] [PATCH net-next v2 6/6] ice: sleep, don't busy-wait, in the SQ send retry loop10 ms is a lot of time to spend busy-waiting. Sleeping is clearly allowed here, because we have just returned from
ice_sq_send_cmd(), which takes a mutex.
Quoted 8 lines
On kernels with HZ=100, this msleep may be twice as long, but I don't think it matters. I did not actually observe any retries happening here.
Signed-off-by Michal Schmidt <[email protected]>--- drivers/net/ethernet/intel/ice/ice_common.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Quoted 26 lines
From: Michal Schmidt <[email protected]>
Sent: Wednesday, April 12, 2023 10:19 AMThis series lowers the CPU usage of the ice driver when using its provided
/dev/gnss*.v2:
- Changed subject of patch 1. Requested by Andrew Lunn.
- Added patch 2 to change the polling interval as recommended by Intel.
- Added patch 3 to removesq_cmd_timeoutas suggested by Simon Horman.Michal Schmidt (6):
ice: do not busy-wait to read GNSS data
ice: increase the GNSS data polling interval to 20 ms
ice: removeice_ctl_q_info::sq_cmd_timeout
ice: sleep, don't busy-wait, forICE_CTL_Q_SQ_CMD_TIMEOUT
ice: remove unused buffer copy code inice_sq_send_cmd_retry()
ice: sleep, don't busy-wait, in the SQ send retry loopdrivers/net/ethernet/intel/ice/ice_common.c | 29 +++++-------- drivers/net/ethernet/intel/ice/ice_controlq.c | 12 +++--- drivers/net/ethernet/intel/ice/ice_controlq.h | 3 +- drivers/net/ethernet/intel/ice/ice_gnss.c | 42 +++++++++---------- drivers/net/ethernet/intel/ice/ice_gnss.h | 3 +- 5 files changed, 36 insertions(+), 53 deletions(-)
Quoted 2 lines
--
2.39.2
In general I couldn't find any issues with the series. Thank you all for suggestions and your work on this.
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Quoted 22 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Michal Schmidt
Sent: Wednesday, April 12, 2023 1:19 AM
To: [email protected]
Cc: Andrew Lunn <[email protected]>; [email protected]; Brandeburg, Jesse <[email protected]>; Kolacinski, Karol <[email protected]>; Nguyen, Anthony L <[email protected]>; Simon Horman <[email protected]>
Subject: [Intel-wired-lan] [PATCH net-next v2 0/6] ice: lower CPU usage with GNSSThis series lowers the CPU usage of the ice driver when using its provided
/dev/gnss*.v2:
- Changed subject of patch 1. Requested by Andrew Lunn.
- Added patch 2 to change the polling interval as recommended by Intel.
- Added patch 3 to removesq_cmd_timeoutas suggested by Simon Horman.Michal Schmidt (6):
ice: do not busy-wait to read GNSS data
ice: increase the GNSS data polling interval to 20 ms
ice: removeice_ctl_q_info::sq_cmd_timeout
ice: sleep, don't busy-wait, forICE_CTL_Q_SQ_CMD_TIMEOUT
ice: remove unused buffer copy code inice_sq_send_cmd_retry()
ice: sleep, don't busy-wait, in the SQ send retry loop
Quoted 6 lines
drivers/net/ethernet/intel/ice/ice_common.c | 29 +++++-------- drivers/net/ethernet/intel/ice/ice_controlq.c | 12 +++--- drivers/net/ethernet/intel/ice/ice_controlq.h | 3 +- drivers/net/ethernet/intel/ice/ice_gnss.c | 42 +++++++++---------- drivers/net/ethernet/intel/ice/ice_gnss.h | 3 +- 5 files changed, 36 insertions(+), 53 deletions(-)
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan