PD#SWPL-138314
Problem:
restart network will involve too much
problem, remove this flow, left only for txhd2
Solution:
remove network restart when resume
only txhd2 will use this function
Verify:
BE311
Change-Id: Ib8773adf20f0176b053fc22802426725d83d2ccc
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
PD#SWPL-138487
Problem:
add new flow for txhd2
Solution:
add new flow for txhd2
Verify:
BE311
Change-Id: Id8a839b8fdb53b58d2cbcb7651b58156ecb14d92
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
PD#SWPL-128836
Problem:
txhd2 resume eth0 down
Solution:
txhd2 use new pll
Verify:
be311
Change-Id: Iaf079569e3d25006aaa3dcd824658be383d35b9f
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
PD#SWPL-117003
Problem:
skip the new flow in common14
Solution:
common14 add new flow for ethernet,
which our phy not support yet.
Verify:
verify with T3
Change-Id: I297f6b83523d95410f3333c3e216d9dbadc644bb
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
PD#SWPL-114869
Problem:
wol may not work normally in some special case.
Solution:
complete 2 wol patch
1.come from https://scgit.amlogic.com/#/c/293893/
2.hold sometimes after resume
Verify:
QA_verify with oppen
PD#SWPL-114869
Change-Id: I7557b4946e8b3ac455f2c28e88a6c1563a0cb4ae
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
PD#SWPL-114869
Problem:
phy not work atfer resume sometimes
Solution:
restart ethernet after resume
Verify:
QA_verify with oppen
Change-Id: I88bac8f0c24f0e2890b24a1508e05e74ed667b5c
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
PD#SWPL-106998
Problem:
c3 mini defconfig built error
Solution:
Fixed c3 mini defconfig built error
Verify:
C3 aw402
Change-Id: Idbbe53d8775f82bfff880ac3ce65b57574cf15da
Signed-off-by: yao zhang1 <yao.zhang1@amlogic.com>
PD#SWPL-110094
Problem:
eth interface has not closed during shutdown
Solution:
add shutdown function
Verify:
T963D4-AY309#custom verify
Change-Id: I235bac19eb7771b9a92eaa143986bdf47d19bbcd
Signed-off-by: xiangyang.yan <xiangyang.yan@amlogic.com>
PD#SWPL-105841
Problem:
C3_5.15 buildroot use different config
not arm64 arm32
Solution:
add control for c3_5.15
Verify:
compile no error
Change-Id: Id9ba5a59e060aef522e6453075df6df7c3a31011
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
PD#SWPL-98490
Problem:
porting wol function
Solution:
porting wol function
Verify:
AR321-T965D4-REF#049
Change-Id: Iea34df3e69d2dc46488a7008eab3788ab5b58094
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
PD#SWPL-77313
Problem:
bringup eth
Solution:
bringup eth
Verify:
Verified on AP222_S905Y4#005
Signed-off-by: Zhuo Wang <zhuo.wang@amlogic.com>
Change-Id: I82947be4b46c14086effc804a1681ec9c1d99ecb
PD#SWPL-106998
Problem:
Boottime is too slow.
Solution:
Speed up boottime 50ms.
Verify:
C3 aw402
Change-Id: I48d8d319f933d89fa9c32c91db7139c51c47ce9d
Signed-off-by: yao zhang1 <yao.zhang1@amlogic.com>
PD#SWPL-163856
Problem:
not support Image.zstd build target
Solution:
add it
Change-Id: I95a4214fd19a3d6206f972ddc3f318de5b58cc4c
Signed-off-by: Tao Zeng <tao.zeng@amlogic.com>
PD#SWPL-57667
Problem:
ZSTD compress level is fixed
Solution:
add module parameters to control it. By default, set
compress level to 1
Verify:
t5d am311
Signed-off-by: Tao Zeng <tao.zeng@amlogic.com>
Change-Id: I4c80eb8cd591c8a2197c7781e8afe589f121ffd7
PD#SWPL-80202
Problem:
If enabled 32bit kasan on T5W with VMAP, kernel can't boot.
This case can't happen on T5D.
Solution:
Rootcause is bss cleared but real data keeps in cache, not flushed
into DDR, and physical kasan shadow page soon be mapped to shadow
memory while in shadow virtual address we see random data, which
caused kasan check failed in vmap fixup function and crashed. So
just add a dcache flush function call after bss cleared.
Verify:
T5W
Change-Id: Id72d26046a2155ee8fea35c3bf8c7aa4fe52e7a4
Signed-off-by: Jianxiong Pan <jianxiong.pan@amlogic.com>
PD#SWPL-145324
Problem:
There are too many trace event inodes
Solution:
Add white list for trace event, this can save about 10MB memory
Verify:
t5m
Change-Id: I290020fc93e6114650313328ec8757720bfc8b1d
Signed-off-by: Tao Zeng <tao.zeng@amlogic.com>
PD#SWPL-139378
Problem:
MID_ORDER_GFP may cost a lot of CPU time in memor compaction when allocate
big system heap.
Solution:
change to HIGH_ORDER_GFP can prevent this cost.
Change order 4 to 3 let system heap can allocate more contiguous memory.
Verify:
ohm
Change-Id: I54799ad624fa2db3d63311209371654e35b2f5c3
Signed-off-by: Hanjie Lin <hanjie.lin@amlogic.com>
PD#SWPL-101774
Problem:
save_stack_trace can't simplify all callers'
use of stackdepot, because the value of
"__irqentry_text_start" and "__irqentry_text_end"
is same.
Solution:
"#define __exception_irq_entry __irq_entry" to
make the section ".irqentry.text" is not empty.
Verify:
THXD2
Change-Id: Ib99f366955a051cb439ee9f86dfe2c667288d2d0
Signed-off-by: qiankun.wang <qiankun.wang@amlogic.com>
PD#SWPL-101774
Problem:
save_stack_trace now record user calls
Solution:
add kernel_or_module_addr judgement before save stack
Verify:
t5w 32bit
Change-Id: Ie029f6082aa1be33924018ec655267b426054aae
Signed-off-by: qiankun.wang <qiankun.wang@amlogic.com>
PD#SWPL-132402
Problem:
read no map memory failed when arm build
Solution:
check memory map when arm build
Verify:
local
Change-Id: Ie9a0359168e3ed620a8bb2a5b7a3ef3e85d983bc
Signed-off-by: qinglin.li <qinglin.li@amlogic.com>
PD#SWPL-128513
Problem:
enable kasan on arm32 will crash
Solution:
disable AMLOGIC_VMAP when kasan enabled
increase modules region form 16M to 32M
Verify:
TXHD2
Change-Id: Ia909a2f3522f5d994a6b74f2e853f10d0a7ac6ca
Signed-off-by: song.han <song.han@amlogic.com>
PD#SWPL-124928
Problem:
Android P no LMKD.
Solution:
add statistics for free file cma and more.
Verify:
local.
Change-Id: If31f74666d97b9e42627e7d7fff693b52e0fbc2e
Signed-off-by: Jianxiong Pan <jianxiong.pan@amlogic.com>
PD#SWPL-114632
Problem:
Kernel size is too big
Solution:
Remove some CONFIGs
Verify:
c3 pass
Change-Id: I259aa858e541cd18400c94abea5b2bb447720fbf
Signed-off-by: qiankun.wang <qiankun.wang@amlogic.com>
PD#SWPL-95400
Problem:
If we pin a file in android, the whole data of this file
will be loaded to DDR, but most of these data may not be
used, this caused memory waste.
Solution:
delay mark mlocked flags for unevictable pages until
it has been faulted.
You can use following command to disable this function:
echo 0 > /proc/sys/vm/shrink_unevictable
Verify:
sc2_ah212
Change-Id: I05bb7654cc228604fe692429efbe40af713ca5c3
Signed-off-by: Jianxiong Pan <jianxiong.pan@amlogic.com>
PD#SWPL-94445
Problem:
play video system crash.
Solution:
disable FAULT_FLAG_SPECULATIVE.
Verify:
sc2_ah212.
Change-Id: Ib056ffe1f48e68ae8e7b848c23db8218710fbbb7
Signed-off-by: Jianxiong Pan <jianxiong.pan@amlogic.com>
PD#SWPL-92714
Problem:
llmv can't guarantee frame pointer register is saved in leaf function
calls. But x29 is dedicated used by vmap stack function. This caused
stack mistake and system hung.
Solution:
Add -mno-omit-leaf-frame-pointer to force x29 used as dedicated frame
pointer
Verify:
t5d am301
Change-Id: I688d61894bdbaaf8f968d1b7689f533d3d62cb88
Signed-off-by: Jianxiong Pan <jianxiong.pan@amlogic.com>
PD#SWPL-73422
Problem:
When driver/kernel call kmalloc with large size, memory may waste
if size is not equal to 2^n. For example, driver call kmalloc with
size 129KB, kmalloc will allocate a 256KB memory block to caller.
Then 127kb memory will be wasted if this caller do not free it.
Solution:
Free tail of slab memory if size is not match to 2^n. This change
can save about 900KB memory after boot, and more than 100KB during
runtime.
mm: reset page order to slab size. [1/1]
gki: modify the slabtrace to meet gki request. [1/2]
mm: fix slab info stat error. [2/2]
Verify:
local.
Change-Id: If0b5aa61f123c657ddd6f49b42b0a60220a25910
Signed-off-by: Jianxiong Pan <jianxiong.pan@amlogic.com>
PD#SWPL-72254
Problem:
No mem debug.
Solution:
porting it from 5.4.
Verify:
local.
Signed-off-by: Jianxiong Pan <jianxiong.pan@amlogic.com>
Change-Id: Ibdae855520510910e8f9a9dba9a469e0b46cc512
PD#SWPL-70022
PD#SWPL-131211
Problem:
No cma optimization on 5.15
Solution:
1, kernel only allow zram/anon pages use cma, new solution all
all movable pages can use cma;
2, add __GFP_NO_CMAto tell buddy system which cases can't use
cma;
3, keep high reference/active file cache pages when cma allocating;
4, hook real cma allocate/release interface to aml_cma.c, which can
use boost mode when allocate large amount of cma;
5, count cma isolated pages to avoid deadloop congestion wait in kswapd
or compaction case;
6, keep cma-unsuitable pages not compacte to cma area;
7, fix SWPL-131211
Verify:
local
Signed-off-by: Tao Zeng <tao.zeng@amlogic.com>
Change-Id: Ib82b898a596eba62e1aa1cb13719eab41f03ae58
PD#SWPL-70022
Problem:
No slab trace on 5.15
Solution:
Porting it from 5.4
Verify:
local
Signed-off-by: Tao Zeng <tao.zeng@amlogic.com>
Change-Id: I8604e6d144ba212e33db9d2a842b128c791fc2bd
PD#SWPL-70022
Problem:
No pagetrace function on 5.15 kernel
Solution:
porting it from 5.4
Verify:
local
Signed-off-by: Tao Zeng <tao.zeng@amlogic.com>
Change-Id: I69cc8b4f4fa4c323204f86e5c76613cd3d6087df
PD#SWPL-121806
Problem:
arm OPTPROBES will cause kprobe handler unexpected behavior
memory atrace need kprobe at handle_pte_fault()
Solution:
disable HAVE_OPTPROBES and noinline handle_pte_fault
Verify:
sc2
Change-Id: I4efbec5aeeb7b6fe56f6fbd0d413bae123549c5a
Signed-off-by: Hanjie Lin <hanjie.lin@amlogic.com>
