pa-fazil
/
android_device_qcom_common
mirror of https://github.com/fazilsheik96/android_device_qcom_common.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
android_device_qcom_common/vendor/perf/init.aospa.perf.rc

100 lines
4.6 KiB
Plaintext
Raw Normal View History

common: perf: add more dm devices for read_ahead_kb tuning There are dozens of APEX modules now Change-Id: I1f489cfa39cc9d3a50135fc24b31cea6500d5e57 Signed-off-by: Juhyung Park <qkrwngud825@gmail.com>
2022-02-13 08:57:38 +00:00
# Copyright (C) 2022 Paranoid Android
common: init: Override readahead to 128KiB after post-boot. * Qualcomm decided to increase the read-ahead value to 512KiB [1]. * This value resulted in frequent cases where read-ahead would be waiting for new I/O requests and thus ending up being synchronous rather than asynchronous. * As result, the main goal of Qualcomm's patch [1] which is to increase throughput ends up in a regression where read throughput is significantly lower than it ever would be with the recommended [2] read-ahead value. * This patch uses our existing loop in ramdisk to set the read-ahead value to 128KiB. [1] https://source.codeaurora.org/quic/la/kernel/msm-4.19/commit/include/linux/mm.h?h=LA.UM.9.12.r1-09000-SMxx50.0&id=25d04e4ceb7c067907901842c223d62973d73a75 [2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/include/linux/mm.h?h=v4.19.160#n2397 Change-Id: I1980d904568fe4f1ce7b676452bcc44e4689f02f
2017-01-30 13:08:58 +00:00
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
common: perf: move all read_ahead_kb settings to init.aospa.perf.rc Change-Id: I496d15b2d398d68b59e019e2c917924f829df5c9 Signed-off-by: Juhyung Park <qkrwngud825@gmail.com>
2022-02-13 09:02:09 +00:00
on post-fs-data
common: perf: disable zram readahead for all devices Change-Id: I12a4ffb916c56df30577e75098dec10694c61e66 Signed-off-by: Juhyung Park <qkrwngud825@gmail.com>
2022-02-13 09:04:23 +00:00
# Disable zram readahead
write /proc/sys/vm/page-cluster 0
common: perf: move all read_ahead_kb settings to init.aospa.perf.rc Change-Id: I496d15b2d398d68b59e019e2c917924f829df5c9 Signed-off-by: Juhyung Park <qkrwngud825@gmail.com>
2022-02-13 09:02:09 +00:00
# Override readahead to 2MiB during boot
write /sys/block/dm-0/queue/read_ahead_kb 2048
write /sys/block/dm-1/queue/read_ahead_kb 2048
write /sys/block/dm-2/queue/read_ahead_kb 2048
write /sys/block/dm-3/queue/read_ahead_kb 2048
write /sys/block/dm-4/queue/read_ahead_kb 2048
write /sys/block/dm-5/queue/read_ahead_kb 2048
write /sys/block/dm-6/queue/read_ahead_kb 2048
write /sys/block/dm-7/queue/read_ahead_kb 2048
write /sys/block/dm-8/queue/read_ahead_kb 2048
write /sys/block/dm-9/queue/read_ahead_kb 2048
write /sys/block/mmcblk0/queue/read_ahead_kb 2048
write /sys/block/sda/queue/read_ahead_kb 2048
write /sys/block/sdb/queue/read_ahead_kb 2048
write /sys/block/sdc/queue/read_ahead_kb 2048
write /sys/block/sdd/queue/read_ahead_kb 2048
write /sys/block/sde/queue/read_ahead_kb 2048
write /sys/block/sdf/queue/read_ahead_kb 2048
common: init: Override readahead to 128KiB after post-boot. * Qualcomm decided to increase the read-ahead value to 512KiB [1]. * This value resulted in frequent cases where read-ahead would be waiting for new I/O requests and thus ending up being synchronous rather than asynchronous. * As result, the main goal of Qualcomm's patch [1] which is to increase throughput ends up in a regression where read throughput is significantly lower than it ever would be with the recommended [2] read-ahead value. * This patch uses our existing loop in ramdisk to set the read-ahead value to 128KiB. [1] https://source.codeaurora.org/quic/la/kernel/msm-4.19/commit/include/linux/mm.h?h=LA.UM.9.12.r1-09000-SMxx50.0&id=25d04e4ceb7c067907901842c223d62973d73a75 [2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/include/linux/mm.h?h=v4.19.160#n2397 Change-Id: I1980d904568fe4f1ce7b676452bcc44e4689f02f
2017-01-30 13:08:58 +00:00
on property:vendor.post_boot.parsed=1
common: perf: move all read_ahead_kb settings to init.aospa.perf.rc Change-Id: I496d15b2d398d68b59e019e2c917924f829df5c9 Signed-off-by: Juhyung Park <qkrwngud825@gmail.com>
2022-02-13 09:02:09 +00:00
# Reset readahead to 128KiB
common: init: Override readahead to 128KiB after post-boot. * Qualcomm decided to increase the read-ahead value to 512KiB [1]. * This value resulted in frequent cases where read-ahead would be waiting for new I/O requests and thus ending up being synchronous rather than asynchronous. * As result, the main goal of Qualcomm's patch [1] which is to increase throughput ends up in a regression where read throughput is significantly lower than it ever would be with the recommended [2] read-ahead value. * This patch uses our existing loop in ramdisk to set the read-ahead value to 128KiB. [1] https://source.codeaurora.org/quic/la/kernel/msm-4.19/commit/include/linux/mm.h?h=LA.UM.9.12.r1-09000-SMxx50.0&id=25d04e4ceb7c067907901842c223d62973d73a75 [2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/include/linux/mm.h?h=v4.19.160#n2397 Change-Id: I1980d904568fe4f1ce7b676452bcc44e4689f02f
2017-01-30 13:08:58 +00:00
write /sys/block/dm-0/queue/read_ahead_kb 128
write /sys/block/dm-1/queue/read_ahead_kb 128
write /sys/block/dm-2/queue/read_ahead_kb 128
write /sys/block/dm-3/queue/read_ahead_kb 128
write /sys/block/dm-4/queue/read_ahead_kb 128
write /sys/block/dm-5/queue/read_ahead_kb 128
write /sys/block/dm-6/queue/read_ahead_kb 128
write /sys/block/dm-7/queue/read_ahead_kb 128
write /sys/block/dm-8/queue/read_ahead_kb 128
write /sys/block/dm-9/queue/read_ahead_kb 128
common: perf: add more dm devices for read_ahead_kb tuning There are dozens of APEX modules now Change-Id: I1f489cfa39cc9d3a50135fc24b31cea6500d5e57 Signed-off-by: Juhyung Park <qkrwngud825@gmail.com>
2022-02-13 08:57:38 +00:00
write /sys/block/dm-10/queue/read_ahead_kb 128
write /sys/block/dm-11/queue/read_ahead_kb 128
write /sys/block/dm-12/queue/read_ahead_kb 128
write /sys/block/dm-13/queue/read_ahead_kb 128
write /sys/block/dm-14/queue/read_ahead_kb 128
write /sys/block/dm-15/queue/read_ahead_kb 128
write /sys/block/dm-16/queue/read_ahead_kb 128
write /sys/block/dm-17/queue/read_ahead_kb 128
write /sys/block/dm-18/queue/read_ahead_kb 128
write /sys/block/dm-19/queue/read_ahead_kb 128
write /sys/block/dm-20/queue/read_ahead_kb 128
write /sys/block/dm-21/queue/read_ahead_kb 128
write /sys/block/dm-22/queue/read_ahead_kb 128
write /sys/block/dm-23/queue/read_ahead_kb 128
write /sys/block/dm-24/queue/read_ahead_kb 128
write /sys/block/dm-25/queue/read_ahead_kb 128
write /sys/block/dm-26/queue/read_ahead_kb 128
write /sys/block/dm-27/queue/read_ahead_kb 128
write /sys/block/dm-28/queue/read_ahead_kb 128
write /sys/block/dm-29/queue/read_ahead_kb 128
write /sys/block/dm-30/queue/read_ahead_kb 128
write /sys/block/dm-31/queue/read_ahead_kb 128
write /sys/block/dm-32/queue/read_ahead_kb 128
write /sys/block/dm-33/queue/read_ahead_kb 128
write /sys/block/dm-34/queue/read_ahead_kb 128
write /sys/block/dm-35/queue/read_ahead_kb 128
write /sys/block/dm-36/queue/read_ahead_kb 128
write /sys/block/dm-37/queue/read_ahead_kb 128
write /sys/block/dm-38/queue/read_ahead_kb 128
write /sys/block/dm-39/queue/read_ahead_kb 128
write /sys/block/dm-40/queue/read_ahead_kb 128
write /sys/block/dm-41/queue/read_ahead_kb 128
write /sys/block/dm-42/queue/read_ahead_kb 128
write /sys/block/dm-43/queue/read_ahead_kb 128
write /sys/block/dm-44/queue/read_ahead_kb 128
write /sys/block/dm-45/queue/read_ahead_kb 128
write /sys/block/dm-46/queue/read_ahead_kb 128
write /sys/block/dm-47/queue/read_ahead_kb 128
write /sys/block/dm-48/queue/read_ahead_kb 128
write /sys/block/dm-49/queue/read_ahead_kb 128
common: init: Override readahead to 128KiB after post-boot. * Qualcomm decided to increase the read-ahead value to 512KiB [1]. * This value resulted in frequent cases where read-ahead would be waiting for new I/O requests and thus ending up being synchronous rather than asynchronous. * As result, the main goal of Qualcomm's patch [1] which is to increase throughput ends up in a regression where read throughput is significantly lower than it ever would be with the recommended [2] read-ahead value. * This patch uses our existing loop in ramdisk to set the read-ahead value to 128KiB. [1] https://source.codeaurora.org/quic/la/kernel/msm-4.19/commit/include/linux/mm.h?h=LA.UM.9.12.r1-09000-SMxx50.0&id=25d04e4ceb7c067907901842c223d62973d73a75 [2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/include/linux/mm.h?h=v4.19.160#n2397 Change-Id: I1980d904568fe4f1ce7b676452bcc44e4689f02f
2017-01-30 13:08:58 +00:00
write /sys/block/mmcblk0/queue/read_ahead_kb 128
write /sys/block/sda/queue/read_ahead_kb 128
write /sys/block/sdb/queue/read_ahead_kb 128
write /sys/block/sdc/queue/read_ahead_kb 128
write /sys/block/sdd/queue/read_ahead_kb 128
write /sys/block/sde/queue/read_ahead_kb 128
write /sys/block/sdf/queue/read_ahead_kb 128
vendor: perf: Setup foreground CPU Sets for LAHAINA. * Other SoCs are welcome here. Please test thoroughly before , especially power usage. Change-Id: I66b29040701fb20aab6c8db370ff7810ce80bdb0
2022-02-14 04:32:14 +00:00
on property:ro.board.platform=lahaina
# Setup foreground cpuset for display composer.
write /dev/cpuset/foreground/cpus 0-6