lkml.org 
[lkml]   [2022]   [Oct]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Date
SubjectRe: [PATCH v7 18/43] instrumented.h: add KMSAN support
without the revert, this include/linux/kmsan-checks.h patch no help.
busybox top, com.android.bluetooth cpu usage 12%.
my kernel config, config-6.1.0-rc1-android-x86_64+

CPU: 3.6% usr 12.1% sys 0.0% nic 84.1% idle 0.0% io 0.0% irq 0.0% sirq
Load average: 1.49 0.65 0.25 4/1336 4940
PID PPID USER STAT VSZ %VSZ CPU %CPU COMMAND
1381 926 1002 S 1265m 16.1 5 12.4 {droid.bluetooth}
com.android.bluetooth
919 1 0 S< 187m 2.4 2 2.2 /system/bin/surfaceflinger
2220 926 10327 S 1882m 24.0 7 1.1 com.baidu.input
3163 926 10175 S 2153m 27.5 0 0.0 {rojekti.clipper}
fi.rojekti.clipper
1250 926 1000 S< 2140m 27.3 0 0.0 system_server
2548 926 10125 S 2078m 26.5 4 0.0 {chiztech.swapps}
com.schiztech.swapps
2516 926 10217 S 2058m 26.3 2 0.0 {ogle.android.gm}
com.google.android.gm
2925 926 10318 S 1876m 24.0 1 0.0 {onelli.juicessh}
com.sonelli.juicessh
926 1 0 S 1780m 22.7 7 0.0 {main} zygote
3374 3369 0 S 1541m 19.7 2 0.0 {main} com.topjohnwu.magisk:root
3079 926 10021 S 1477m 18.9 0 0.0 {ndroid.systemui}
com.android.systemui

2022-10-20 5:36 GMT+08:00, Marco Elver <elver@google.com>:
> On Thu, Oct 20, 2022 at 04:07AM +0800, youling 257 wrote:
>> That is i did,i already test, remove "u64 __tmp…kmsan_unpoison_memory", no
>> help.
>> i only remove kmsan_copy_to_user, fix my issue.
>
> Ok - does only the below work (without the reverts)?
>
> diff --git a/include/linux/kmsan-checks.h b/include/linux/kmsan-checks.h
> index c4cae333deec..eb05caa8f523 100644
> --- a/include/linux/kmsan-checks.h
> +++ b/include/linux/kmsan-checks.h
> @@ -73,8 +73,8 @@ static inline void kmsan_unpoison_memory(const void
> *address, size_t size)
> static inline void kmsan_check_memory(const void *address, size_t size)
> {
> }
> -static inline void kmsan_copy_to_user(void __user *to, const void *from,
> - size_t to_copy, size_t left)
> +static __always_inline void kmsan_copy_to_user(void __user *to, const void
> *from,
> + size_t to_copy, size_t left)
> {
> }
>
>
> ... because when you say only removing kmsan_copy_to_user() (from
> instrument_put_user()) works, it really doesn't make any sense. The only
> explanation would be if the compiler inlining is broken.
>
[unhandled content-type:application/octet-stream]
\
 
 \ /
  Last update: 2022-10-20 07:53    [W:0.100 / U:0.084 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site