mmc: mtk-sd: Implement Host Software Queue for eMMC and SD Card

Add support for Host Software Queue (HSQ) and enable it when the
controller instance does not have Command Queue Engine HW support.

It was chosen to enable HSQ only for eMMC and SD/MicroSD cards
and not for SDIO as performance improvements are seen only for
the former.

Performance was measured with a SanDisk Extreme Ultra A2 MicroSD
card in a MediaTek MT8195T Acer Chromebook Spin 513 (CP513-2H),
by running FIO (bs=4k) on an ArchLinux userspace.

.... Summarizing ....
Random read:     +24.28% IOPS, +24.29% BW
Sequential read: +3.14%  IOPS, +3.49%  BW
Random RW (avg): +50.53% IOPS, +50.68% BW

Below, more data from the benchmarks.

Before:
 - Random read: IOPS=1643, BW=6574KiB/s
   bw (  KiB/s): min= 4578, max= 7440, per=99.95%, avg=6571.55, stdev=74.16, samples=953
   iops        : min= 1144, max= 1860, avg=1642.14, stdev=18.54, samples=953
   lat (msec)  : 100=0.01%, 250=0.12%, 500=0.38%, 750=97.89%, 1000=1.44%, 2000=0.16%
 - Sequential read: IOPS=19.1k, BW=74.4MiB/s
   bw (  KiB/s): min=12288, max=118483, per=100.00%, avg=76293.38, stdev=1971.42, samples=956
   iops        : min= 3072, max=29620, avg=19072.14, stdev=492.87, samples=956
   lat (msec)  : 4=0.01%, 10=0.01%, 20=0.21%, 50=23.95%, 100=75.67%, 250=0.05%, 500=0.03%, 750=0.08%
 - Random R/W: read: IOPS=282, BW=1129KiB/s (1156kB/s)  write: IOPS=284, BW=1136KiB/s
   read bw (  KiB/s): min=   31, max= 3496, per=100.00%, avg=1703.67, stdev=155.42, samples=630
   read iops        : min=    7, max=  873, avg=425.22, stdev=38.85, samples=630
   wri  bw (  KiB/s): min=   31, max= 3443, per=100.00%, avg=1674.27, stdev=164.23, samples=644
   wri  iops        : min=    7, max=  860, avg=417.87, stdev=41.03, samples=644
   lat (msec)   : 250=0.13%, 500=0.44%, 750=0.84%, 1000=22.29%, 2000=74.01%, >=2000=2.30%

After:
 - Random read: IOPS=2042, BW=8171KiB/s
   bw (  KiB/s): min= 4907, max= 9072, per=99.94%, avg=8166.80, stdev=93.77, samples=954
   iops        : min= 1226, max= 2268, avg=2040.78, stdev=23.41, samples=954
   lat (msec)   : 100=0.03%, 250=0.13%, 500=52.88%, 750=46.64%, 1000=0.32%
 - Sequential read: IOPS=19.7k, BW=77.0MiB/s
   bw (  KiB/s): min=67980, max=94248, per=100.00%, avg=78894.27, stdev=1475.07, samples=956
   iops        : min=16994, max=23562, avg=19722.45, stdev=368.76, samples=956
   lat (msec)   : 4=0.01%, 10=0.01%, 20=0.05%, 50=28.78%, 100=71.14%, 250=0.01%, 500=0.02%
 - Random R/W: read: IOPS=424, BW=1699KiB/s  write: IOPS=428, BW=1714KiB/s
   read bw (  KiB/s): min=  228, max= 2856, per=100.00%, avg=1796.60, stdev=112.59, samples=901
   read iops        : min=   54, max=  712, avg=447.81, stdev=28.21, samples=901
   wri  bw (  KiB/s): min=   28, max= 2904, per=100.00%, avg=1780.11, stdev=128.27, samples=916
   wri  iops        : min=    4, max=  724, avg=443.69, stdev=32.14, samples=916

Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240930090156.33537-1-angelogioacchino.delregno@collabora.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>

0 files changed, 0 insertions, 0 deletions