Skip to contentSkip to conversation

mmc: Set optimal I/O size when mmc_setip_queue

PATCH In review [email protected] 4 participants 13 messages Aug 18, 2023 – Aug 28, 2023

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mmc: Set optimal I/O size when mmc_setip_queue

From: Sharp Xia <[email protected]>

MMC does not set readahead and uses the default VM_READAHEAD_PAGES resulting in slower reading speed.
Use the max_req_size reported by host driver to set the optimal I/O size to improve performance.

---
 drivers/mmc/core/queue.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/core/queue.c +1 -0
index b396e3900717..fc83c4917360 100644
@@ -359,6 +359,7 @@ static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)
 		blk_queue_bounce_limit(mq->queue, BLK_BOUNCE_HIGH);
 	blk_queue_max_hw_sectors(mq->queue,
 		min(host->max_blk_count, host->max_req_size / 512));
+	blk_queue_io_opt(mq->queue, host->max_req_size);
 	if (host->can_dma_map_merge)
 		WARN(!blk_queue_can_use_dma_map_merging(mq->queue,
 							mmc_dev(host)),
Signature · 1 line
2.18.0
#1 lore raw
Quoted 7 lines · On Fri, 18 Aug 2023 at 04:45, <[email protected]> wrote:

From: Sharp Xia <[email protected]>

MMC does not set readahead and uses the default VM_READAHEAD_PAGES resulting in slower reading speed.
Use the max_req_size reported by host driver to set the optimal I/O size to improve performance.

This seems reasonable to me. However, it would be nice if you could share some performance numbers too - comparing before and after $subject patch.

Kind regards
Uffe

Quoted 21 lines
---
 drivers/mmc/core/queue.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/core/queue.c +1 -1
index b396e3900717..fc83c4917360 100644
@@ -359,6 +359,7 @@ static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)
                blk_queue_bounce_limit(mq->queue, BLK_BOUNCE_HIGH);
        blk_queue_max_hw_sectors(mq->queue,
                min(host->max_blk_count, host->max_req_size / 512));
+       blk_queue_io_opt(mq->queue, host->max_req_size);
        if (host->can_dma_map_merge)
                WARN(!blk_queue_can_use_dma_map_merging(mq->queue,
                                                        mmc_dev(host)),
--

2.18.0

#2 lore raw
Quoted 34 lines · Ulf Hansson

External email : Please do not click links or open attachments until you have verified the sender or the content.

Quoted 7 lines · On Fri, 18 Aug 2023 at 04:45, <[email protected]> wrote:

From: Sharp Xia <[email protected]>

MMC does not set readahead and uses the default VM_READAHEAD_PAGES resulting in slower reading speed.
Use the max_req_size reported by host driver to set the optimal I/O size to improve performance.

This seems reasonable to me. However, it would be nice if you could share some performance numbers too - comparing before and after $subject patch.

Kind regards
Uffe

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue

Quoted 11 lines
---
 drivers/mmc/core/queue.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/core/queue.c +0 -0
index b396e3900717..fc83c4917360 100644
@@ -359,6 +359,7 @@ static void mmc_setup_queue(struct mmc_queue

*mq, struct mmc_card *card)

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)

Quoted 3 lines
drivers/mmc/core/queue.c +0 -0
                blk_queue_bounce_limit(mq->queue, BLK_BOUNCE_HIGH);
        blk_queue_max_hw_sectors(mq->queue,
                min(host->max_blk_count, host->max_req_size /
Quoted 9 lines

512));

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)

Quoted 4 lines
drivers/mmc/core/queue.c +1 -0
+       blk_queue_io_opt(mq->queue, host->max_req_size);
        if (host->can_dma_map_merge)
                WARN(!blk_queue_can_use_dma_map_merging(mq->queue,
                                                        mmc_dev(hos

t)),

Quoted 3 lines

--
2.18.0

I test this patch on internal platform(kernel-5.15).

Before:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 37.124446 s, 59 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 128

After:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 28.956049 s, 76 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 1024

#11 lore raw
In reply to Ulf Hansson
Quoted 34 lines · Ulf Hansson

External email : Please do not click links or open attachments until you have verified the sender or the content.

Quoted 7 lines · On Fri, 18 Aug 2023 at 04:45, <[email protected]> wrote:

From: Sharp Xia <[email protected]>

MMC does not set readahead and uses the default VM_READAHEAD_PAGES resulting in slower reading speed.
Use the max_req_size reported by host driver to set the optimal I/O size to improve performance.

This seems reasonable to me. However, it would be nice if you could share some performance numbers too - comparing before and after $subject patch.

Kind regards
Uffe

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue

Quoted 11 lines
---
 drivers/mmc/core/queue.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/core/queue.c +0 -0
index b396e3900717..fc83c4917360 100644
@@ -359,6 +359,7 @@ static void mmc_setup_queue(struct mmc_queue

*mq, struct mmc_card *card)

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)

Quoted 3 lines
drivers/mmc/core/queue.c +0 -0
                blk_queue_bounce_limit(mq->queue, BLK_BOUNCE_HIGH);
        blk_queue_max_hw_sectors(mq->queue,
                min(host->max_blk_count, host->max_req_size /
Quoted 9 lines

512));

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)

Quoted 4 lines
drivers/mmc/core/queue.c +1 -0
+       blk_queue_io_opt(mq->queue, host->max_req_size);
        if (host->can_dma_map_merge)
                WARN(!blk_queue_can_use_dma_map_merging(mq->queue,
                                                        mmc_dev(hos

t)),

Quoted 3 lines

--
2.18.0

I test this patch on internal platform(kernel-5.15).

Before:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 37.124446 s, 59 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 128

After:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 28.956049 s, 76 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 1024

#12 lore raw
In reply to Ulf Hansson
Quoted 34 lines · Ulf Hansson

External email : Please do not click links or open attachments until you have verified the sender or the content.

Quoted 7 lines · On Fri, 18 Aug 2023 at 04:45, <[email protected]> wrote:

From: Sharp Xia <[email protected]>

MMC does not set readahead and uses the default VM_READAHEAD_PAGES resulting in slower reading speed.
Use the max_req_size reported by host driver to set the optimal I/O size to improve performance.

This seems reasonable to me. However, it would be nice if you could share some performance numbers too - comparing before and after $subject patch.

Kind regards
Uffe

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue

Quoted 11 lines
---
 drivers/mmc/core/queue.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/core/queue.c +0 -0
index b396e3900717..fc83c4917360 100644
@@ -359,6 +359,7 @@ static void mmc_setup_queue(struct mmc_queue

*mq, struct mmc_card *card)

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)

Quoted 3 lines
drivers/mmc/core/queue.c +0 -0
                blk_queue_bounce_limit(mq->queue, BLK_BOUNCE_HIGH);
        blk_queue_max_hw_sectors(mq->queue,
                min(host->max_blk_count, host->max_req_size /
Quoted 9 lines

512));

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)

Quoted 4 lines
drivers/mmc/core/queue.c +1 -0
+       blk_queue_io_opt(mq->queue, host->max_req_size);
        if (host->can_dma_map_merge)
                WARN(!blk_queue_can_use_dma_map_merging(mq->queue,
                                                        mmc_dev(hos

t)),

Quoted 3 lines

--
2.18.0

I test this patch on internal platform(kernel-5.15).

Before:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 37.124446 s, 59 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 128

After:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 28.956049 s, 76 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 1024

#3 lore raw
In reply to Sharp Xia (夏宇彬)

Hi Sharp,

Quoted 40 lines · Sharp Xia (夏宇彬)
Quoted 39 lines · Ulf Hansson

External email : Please do not click links or open attachments until you have verified the sender or the content.

Quoted 7 lines · On Fri, 18 Aug 2023 at 04:45, <[email protected]> wrote:

From: Sharp Xia <[email protected]>

MMC does not set readahead and uses the default VM_READAHEAD_PAGES resulting in slower reading speed.
Use the max_req_size reported by host driver to set the optimal I/O size to improve performance.

This seems reasonable to me. However, it would be nice if you could share some performance numbers too - comparing before and after $subject patch.

Kind regards
Uffe

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue

Quoted 11 lines
---
  drivers/mmc/core/queue.c | 1 +
  1 file changed, 1 insertion(+)
drivers/mmc/core/queue.c +0 -0
index b396e3900717..fc83c4917360 100644
@@ -359,6 +359,7 @@ static void mmc_setup_queue(struct mmc_queue

*mq, struct mmc_card *card)

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)

Quoted 3 lines
drivers/mmc/core/queue.c +0 -0
                 blk_queue_bounce_limit(mq->queue, BLK_BOUNCE_HIGH);
         blk_queue_max_hw_sectors(mq->queue,
                 min(host->max_blk_count, host->max_req_size /

512));

Quoted 4 lines
drivers/mmc/core/queue.c +1 -0
+       blk_queue_io_opt(mq->queue, host->max_req_size);
         if (host->can_dma_map_merge)
                 WARN(!blk_queue_can_use_dma_map_merging(mq->queue,
                                                         mmc_dev(hos
Quoted 6 lines
Quoted 4 lines

t)),

Quoted 3 lines

--
2.18.0

I test this patch on internal platform(kernel-5.15).

I patched this one and the test shows me a stable 11% performance drop.

Before:
echo 3 > proc/sys/vm/drop_caches && dd if=/data/1GB.img of=/dev/null

2048000+0 records in
2048000+0 records out
1048576000 bytes (0.9 G) copied, 3.912249 s, 256 M/s

After:
echo 3 > proc/sys/vm/drop_caches && dd if=/data/1GB.img of=/dev/null 2048000+0 records in
2048000+0 records out
1048576000 bytes (0.9 G) copied, 4.436271 s, 225 M/s

Quoted 19 lines

Before:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 37.124446 s, 59 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 128

After:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 28.956049 s, 76 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 1024

#4 lore raw
Quoted 67 lines · Shawn Lin

Hi Sharp,

Quoted 3 lines · Sharp Xia (夏宇彬)
Quoted 2 lines · Ulf Hansson

External email : Please do not click links or open attachments

until

Quoted 6 lines
Quoted 6 lines

you have verified the sender or the content.

Quoted 4 lines · On Fri, 18 Aug 2023 at 04:45, <[email protected]> wrote:

From: Sharp Xia <[email protected]>

MMC does not set readahead and uses the default

VM_READAHEAD_PAGES

Quoted 5 lines
Quoted 5 lines
Quoted 3 lines

resulting in slower reading speed.
Use the max_req_size reported by host driver to set the optimal I/O size to improve performance.

This seems reasonable to me. However, it would be nice if you

could

Quoted 12 lines
Quoted 12 lines

share some performance numbers too - comparing before and after $subject patch.

Kind regards
Uffe

Quoted 6 lines
---
  drivers/mmc/core/queue.c | 1 +
  1 file changed, 1 insertion(+)
Quoted 7 lines
Quoted 7 lines

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue

Quoted 5 lines
drivers/mmc/core/queue.c +0 -0
index b396e3900717..fc83c4917360 100644
@@ -359,6 +359,7 @@ static void mmc_setup_queue(struct mmc_queue

*mq, struct mmc_card *card)

Quoted 1 line
drivers/mmc/core/queue.c +0 -0
                 blk_queue_bounce_limit(mq->queue,

BLK_BOUNCE_HIGH);

Quoted 8 lines
Quoted 6 lines

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue · line 361

Quoted 2 lines
drivers/mmc/core/queue.c +0 -0
         blk_queue_max_hw_sectors(mq->queue,
                 min(host->max_blk_count, host->max_req_size /

512));

Quoted 3 lines
drivers/mmc/core/queue.c +1 -0
+       blk_queue_io_opt(mq->queue, host->max_req_size);
         if (host->can_dma_map_merge)
                 WARN(!blk_queue_can_use_dma_map_merging(mq-

queue,

Quoted 1 line
Quoted 1 line
drivers/mmc/core/queue.c +0 -0
                                                         mmc_dev(

hos

Quoted 6 lines
Quoted 4 lines

t)),

Quoted 3 lines

--
2.18.0

I test this patch on internal platform(kernel-5.15).

I patched this one and the test shows me a stable 11% performance drop.

Before:
echo 3 > proc/sys/vm/drop_caches && dd if=/data/1GB.img of=/dev/null

2048000+0 records in
2048000+0 records out
1048576000 bytes (0.9 G) copied, 3.912249 s, 256 M/s

Quoted 25 lines

After:
echo 3 > proc/sys/vm/drop_caches && dd if=/data/1GB.img of=/dev/null 2048000+0 records in
2048000+0 records out
1048576000 bytes (0.9 G) copied, 4.436271 s, 225 M/s

Quoted 19 lines

Before:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 37.124446 s, 59 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 128

After:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 28.956049 s, 76 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 1024

Hi Shawn,

What is your readahead value before and after applying this patch?

#5 lore raw
Quoted 3 lines · On 2023/8/25 16:39, [email protected] wrote:
Quoted 2 lines · Shawn Lin

Hi Sharp,

...

Quoted 6 lines
Quoted 2 lines
Quoted 2 lines

1024

Hi Shawn,

What is your readahead value before and after applying this patch?

The original readahead is 128, and after applying the patch is 1024

cat /d/mmc0/ios

clock:          200000000 Hz
actual clock:   200000000 Hz
vdd:            18 (3.0 ~ 3.1 V)
bus mode:       2 (push-pull)
chip select:    0 (don't care)
power mode:     2 (on)
bus width:      3 (8 bits)
timing spec:    10 (mmc HS400 enhanced strobe)
signal voltage: 1 (1.80 V)
driver type:    0 (driver type B)

The driver I used is sdhci-of-dwcmshc.c with a KLMBG2JETDB041 eMMC chip.

#9 lore raw
In reply to [email protected]
Quoted 94 lines · On Fri, Aug 25, 2023 at 7:43 PM <[email protected]> wrote:
Quoted 67 lines · Shawn Lin

Hi Sharp,

Quoted 3 lines · Sharp Xia (夏宇彬)
Quoted 2 lines · Ulf Hansson

External email : Please do not click links or open attachments

until

Quoted 6 lines
Quoted 6 lines

you have verified the sender or the content.

Quoted 4 lines · On Fri, 18 Aug 2023 at 04:45, <[email protected]> wrote:

From: Sharp Xia <[email protected]>

MMC does not set readahead and uses the default

VM_READAHEAD_PAGES

Quoted 5 lines
Quoted 5 lines
Quoted 3 lines

resulting in slower reading speed.
Use the max_req_size reported by host driver to set the optimal I/O size to improve performance.

This seems reasonable to me. However, it would be nice if you

could

Quoted 12 lines
Quoted 12 lines

share some performance numbers too - comparing before and after $subject patch.

Kind regards
Uffe

Quoted 6 lines
---
  drivers/mmc/core/queue.c | 1 +
  1 file changed, 1 insertion(+)
Quoted 7 lines
Quoted 7 lines

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue

Quoted 5 lines
drivers/mmc/core/queue.c +0 -0
index b396e3900717..fc83c4917360 100644
@@ -359,6 +359,7 @@ static void mmc_setup_queue(struct mmc_queue

*mq, struct mmc_card *card)

Quoted 1 line
drivers/mmc/core/queue.c +0 -0
                 blk_queue_bounce_limit(mq->queue,

BLK_BOUNCE_HIGH);

Quoted 8 lines
Quoted 6 lines

drivers/mmc/core/queue.c static void mmc_setup_queue(struct mmc_queue · line 361

Quoted 2 lines
drivers/mmc/core/queue.c +0 -0
         blk_queue_max_hw_sectors(mq->queue,
                 min(host->max_blk_count, host->max_req_size /

512));

Quoted 3 lines
drivers/mmc/core/queue.c +1 -0
+       blk_queue_io_opt(mq->queue, host->max_req_size);
         if (host->can_dma_map_merge)
                 WARN(!blk_queue_can_use_dma_map_merging(mq-

queue,

Quoted 1 line
Quoted 1 line
drivers/mmc/core/queue.c +0 -0
                                                         mmc_dev(

hos

Quoted 6 lines
Quoted 4 lines

t)),

Quoted 3 lines

--
2.18.0

I test this patch on internal platform(kernel-5.15).

I patched this one and the test shows me a stable 11% performance drop.

Before:
echo 3 > proc/sys/vm/drop_caches && dd if=/data/1GB.img of=/dev/null

2048000+0 records in
2048000+0 records out
1048576000 bytes (0.9 G) copied, 3.912249 s, 256 M/s

Quoted 25 lines

After:
echo 3 > proc/sys/vm/drop_caches && dd if=/data/1GB.img of=/dev/null 2048000+0 records in
2048000+0 records out
1048576000 bytes (0.9 G) copied, 4.436271 s, 225 M/s

Quoted 19 lines

Before:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 37.124446 s, 59 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 128

After:
console:/ # echo 3 > /proc/sys/vm/drop_caches console:/ # dd if=/mnt/media_rw/8031-130D/super.img of=/dev/null 4485393+1 records in
4485393+1 records out
2296521564 bytes (2.1 G) copied, 28.956049 s, 76 M/s console:/ # cat /sys/block/mmcblk0/queue/read_ahead_kb 1024

Quoted 4 lines

Hi Shawn,

What is your readahead value before and after applying this patch?

Hi Sharp
Use "echo 1024 > sys/block/mmcblk0/queue/read_ahead_kb" instead of "blk_queue_io_opt(mq->queue, host->max_req_size);"?

#6 lore raw
In reply to Shawn Lin
Quoted 31 lines · Shawn Lin
Quoted 3 lines · On 2023/8/25 16:39, [email protected] wrote:
Quoted 2 lines · Shawn Lin

Hi Sharp,

...

Quoted 6 lines
Quoted 2 lines
Quoted 2 lines

1024

Hi Shawn,

What is your readahead value before and after applying this patch?

The original readahead is 128, and after applying the patch is 1024

cat /d/mmc0/ios

clock:          200000000 Hz
actual clock:   200000000 Hz
vdd:            18 (3.0 ~ 3.1 V)
bus mode:       2 (push-pull)
chip select:    0 (don't care)
power mode:     2 (on)
bus width:      3 (8 bits)
timing spec:    10 (mmc HS400 enhanced strobe)
signal voltage: 1 (1.80 V)
driver type:    0 (driver type B)
Quoted 2 lines

The driver I used is sdhci-of-dwcmshc.c with a KLMBG2JETDB041 eMMC chip.

I tested with RK3568 and sdhci-of-dwcmshc.c driver, the performance improved by 2~3%.

Before:

root@OpenWrt:/mnt/mmcblk0p3# time dd if=test.img of=/dev/null
2097152+0 records in
2097152+0 records out
real    0m 6.01s
user    0m 0.84s
sys     0m 2.89s
root@OpenWrt:/mnt/mmcblk0p3# cat /sys/block/mmcblk0/queue/read_ahead_kb
128

After:

root@OpenWrt:/mnt/mmcblk0p3# echo 3 > /proc/sys/vm/drop_caches
root@OpenWrt:/mnt/mmcblk0p3# time dd if=test.img of=/dev/null
2097152+0 records in
2097152+0 records out
real    0m 5.86s
user    0m 1.04s
sys     0m 3.18s
root@OpenWrt:/mnt/mmcblk0p3# cat /sys/block/mmcblk0/queue/read_ahead_kb
1024
root@OpenWrt:/sys/kernel/debug/mmc0# cat ios
clock:          200000000 Hz
actual clock:   200000000 Hz
vdd:            18 (3.0 ~ 3.1 V)
bus mode:       2 (push-pull)
chip select:    0 (don't care)
power mode:     2 (on)
bus width:      3 (8 bits)
timing spec:    9 (mmc HS200)
signal voltage: 1 (1.80 V)
driver type:    0 (driver type B)
#10 lore raw
In reply to Wenchao Chen
Quoted 5 lines · Wenchao Chen

Hi Sharp
Use "echo 1024 > sys/block/mmcblk0/queue/read_ahead_kb" instead of "blk_queue_io_opt(mq->queue, host->max_req_size);"?

Hi Wenchao,

User space does not know the max_req_size of each mmc host. And when the SD card is hot inserted,
it is complicated for the user space to modify this value.

#7 lore raw
In reply to [email protected]

Hi Sharp

Quoted 3 lines · On 2023/8/27 0:26, [email protected] wrote:
Quoted 2 lines · Shawn Lin

After more testing, most of my platforms which runs at HS400/HS200 mode shows nearly no differences with the readahead ranging from 128 to 1024. Yet just a board shows a performance drop now. Highly suspect it's eMMC chip depends. I would recommand leave it to the BSP guys to decide which
readahead value is best for their usage.

Quoted 36 lines

I tested with RK3568 and sdhci-of-dwcmshc.c driver, the performance improved by 2~3%.

Before:

root@OpenWrt:/mnt/mmcblk0p3# time dd if=test.img of=/dev/null
2097152+0 records in
2097152+0 records out
real    0m 6.01s
user    0m 0.84s
sys     0m 2.89s
root@OpenWrt:/mnt/mmcblk0p3# cat /sys/block/mmcblk0/queue/read_ahead_kb
128

After:

root@OpenWrt:/mnt/mmcblk0p3# echo 3 > /proc/sys/vm/drop_caches
root@OpenWrt:/mnt/mmcblk0p3# time dd if=test.img of=/dev/null
2097152+0 records in
2097152+0 records out
real    0m 5.86s
user    0m 1.04s
sys     0m 3.18s
root@OpenWrt:/mnt/mmcblk0p3# cat /sys/block/mmcblk0/queue/read_ahead_kb
1024
root@OpenWrt:/sys/kernel/debug/mmc0# cat ios
clock:          200000000 Hz
actual clock:   200000000 Hz
vdd:            18 (3.0 ~ 3.1 V)
bus mode:       2 (push-pull)
chip select:    0 (don't care)
power mode:     2 (on)
bus width:      3 (8 bits)
timing spec:    9 (mmc HS200)
signal voltage: 1 (1.80 V)
driver type:    0 (driver type B)
#8 lore raw
Quoted 8 lines · Shawn Lin

Hi Sharp

Quoted 3 lines · On 2023/8/27 0:26, [email protected] wrote:
Quoted 2 lines · Shawn Lin
Quoted 5 lines

After more testing, most of my platforms which runs at HS400/HS200 mode shows nearly no differences with the readahead ranging from 128 to 1024. Yet just a board shows a performance drop now. Highly suspect it's eMMC chip depends. I would recommand leave it to the BSP guys to decide which readahead value is best for their usage.

That's a very good point. The SD/eMMC card certainly behaves differently, depending on the request-size.

Another thing we could consider doing, could be to combine the information about the request-size from the mmc host, with some relevant information from the registers in the card (not sure exactly what though).

Quoted 25 lines
Quoted 25 lines

I tested with RK3568 and sdhci-of-dwcmshc.c driver, the performance improved by 2~3%.

Before:

root@OpenWrt:/mnt/mmcblk0p3# time dd if=test.img of=/dev/null
2097152+0 records in
2097152+0 records out
real    0m 6.01s
user    0m 0.84s
sys     0m 2.89s
root@OpenWrt:/mnt/mmcblk0p3# cat /sys/block/mmcblk0/queue/read_ahead_kb
128

After:

root@OpenWrt:/mnt/mmcblk0p3# echo 3 > /proc/sys/vm/drop_caches
root@OpenWrt:/mnt/mmcblk0p3# time dd if=test.img of=/dev/null
2097152+0 records in
2097152+0 records out
real    0m 5.86s
user    0m 1.04s
sys     0m 3.18s
root@OpenWrt:/mnt/mmcblk0p3# cat /sys/block/mmcblk0/queue/read_ahead_kb
1024
Quoted 12 lines
Quoted 12 lines
root@OpenWrt:/sys/kernel/debug/mmc0# cat ios
clock:          200000000 Hz
actual clock:   200000000 Hz
vdd:            18 (3.0 ~ 3.1 V)
bus mode:       2 (push-pull)
chip select:    0 (don't care)
power mode:     2 (on)
bus width:      3 (8 bits)
timing spec:    9 (mmc HS200)
signal voltage: 1 (1.80 V)
driver type:    0 (driver type B)

Thanks for testing and sharing the data, both of you!

Kind regards
Uffe