Signed-off-by: NJeffle Xu <jefflexu@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

For the platform with pass-though NVME storage device.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

When c-state and mwait is passthroughed to the guest, enabling intel
idle driver can exploit them and benefit the performance greatly.
And potentially it can boost other features like haltpolling.

The user should remove idle=* from kernel cmdline to put it to work.
Signed-off-by: NYihao Wu <wuyihao@linux.alibaba.com>
Acked-by: NCaspar Zhang <caspar@linux.alibaba.com>

Guest halt polling can reduce the latency of waking a task on an idle vCPU.
The user should remove idle=* from kernel cmdline to put it to work.
Signed-off-by: NYihao Wu <wuyihao@linux.alibaba.com>
Acked-by: NMichael Wang <yun.wang@linux.alibaba.com>

The PM timer is availiable in case of ACPI is unusable or blacklisted.
So let's enable it as a fallback timer when APIC timer goes wrong.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Reviewed-by: NDust Li <dust.li@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

 arm64 alrealy enabled CCGROUP_BPF

Signed-off-by: dust.li <Dust Li dust.li@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NCaspar Zhang <caspar@linux.alibaba.com>

Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

This patch enables Hygon CPU support by default.
Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

Make configs look more pretty by regenerating them with `make -s
ARCH=arm64 menuconfig'. Note that current configs are made
against ck-4.19.91 and with dts8 (gcc-8.3.1).
Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

This is useful for getting block layer debugging information in debugfs,
also our io hang check feature depends on it.
Signed-off-by: NXiaoguang Wang <xiaoguang.wang@linux.alibaba.com>
Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

The deferred page init can help to improve memory initialization at boot
time.

We encountered one memory bandwidth regression in ck-release-17(v4.19.81).
Now, the fix patch for this issue has been backported in v4.19.84. so
enable it again.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Change THP policy from 'madvise' to 'always' since we've found it
benefits a lot in most scenarios, e.g. pmbench.
For defrag configuration, it will be initialized to 'madvise'.
Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>
Acked-by: NCaspar Zhang <caspar@linux.alibaba.com>

Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

This reverts commit 72b9f44eb5c19325fd850c489a0968c99d4c7dbc.

This configure caused lmbench benchmark regression more than
30%, so revert it.

Test by command `bw_pipe -M 10240 -P 16` on 32c64g VM.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

cloud-init can use qemu_fw_cfg module to configure user password. It can
reduce 3~5s boot time.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>

Signed-off-by: NWANG Siyuan <Siyuan.Wang@amd.com>
Reviewed-by: NCaspar Zhang <caspar@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

To support intel icelake, now enable RAS features, dax pmem and icelake
related configs.
Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Signed-off-by: NJiufei Xue <jiufei.xue@linux.alibaba.com>
Reviewed-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Deferred struct_page init can same more boot time, especially for large
memory instance. In my local test with 2C312GB VM, it decreases from
2.389s to 0.138s.
Please note that this has a potential performance impact on processes
running early in the lifetime of the system until these kthreads finish
the initialisation.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

To compile as modules can make the driver init parallel with systemd
startup, which can save about 40ms startup time.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

PS/2 mouse device probing costs more than 600ms after mouse driver
was initialized.
To compile it as modules can make the probing parallel with
init(systemd) startup, which can save about 600ms startup time.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

Enable this configuration to bypass such error reports:

    WARNING: drivers/powercap/intel_rapl_common.o(.text+0x10fb): Section mismatch in reference from the function rapl_add_package() to the variable .init.rodata:rapl_ids
    The function rapl_add_package() references
    the variable __initconst rapl_ids.
    This is often because rapl_add_package lacks a __initconst
    annotation or the annotation of rapl_ids is wrong.

    FATAL: modpost: Section mismatches detected.
    Set CONFIG_SECTION_MISMATCH_WARN_ONLY=y to allow them.
    make[1]: *** [__modpost] Error 1
Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

Turn on several configs to support icelake.
Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>
Reviewed-by: NXunlei Pang <xlpang@linux.alibaba.com>

Add CONFIG_DEV_DAX_KMEM and make some related configs as Y.
Signed-off-by: NYang Shi <yang.shi@linux.alibaba.com>
Reviewed-by: NGavin Shan <shan.gavin@linux.alibaba.com>

Signed-off-by: NCaspar Zhang <caspar@linux.alibaba.com>

The in-kernel torture test includes 2 kinds of stress test:
1. locking stress test;
   (for more info can refer to
    https://www.kernel.org/doc/Documentation/locking/locktorture.txt)
2. RCU stress test.
   (for more info can refer to
    https://www.kernel.org/doc/Documentation/RCU/torture.txt)
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Reviewed-by: NJoseph Qi <joseph.qi@linux.alibaba.com>

When in a KVM image, the user wants to use vHost_vSocks
the resource is already taken by virito_vSockets, so
set them as m to enable this use case.
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>