Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Subject: Re: [PATCH v12 4/5] LoongArch: Mark some assembler symbols as
 non-kprobe-able
To:     Tiezhu Yang <yangtiezhu@loongson.cn>,
        Huacai Chen <chenhuacai@kernel.org>
Cc:     WANG Xuerui <kernel@xen0n.name>,
        Masami Hiramatsu <mhiramat@kernel.org>,
        loongarch@lists.linux.dev, linux-kernel@vger.kernel.org
References: <1674007261-9198-1-git-send-email-yangtiezhu@loongson.cn>
 <1674007261-9198-5-git-send-email-yangtiezhu@loongson.cn>
 <CAAhV-H7Dt5MhaMU1=D9HxPFR+xjjoQG6RguaYMydy=v_jvrtiA@mail.gmail.com>
 <48f0508f-3908-c6ca-c8ba-7c12dd6b3f11@loongson.cn>
From:   Jinyang He <hejinyang@loongson.cn>
Message-ID: <7f518dec-fd72-a73c-3f23-6372a83d65c2@loongson.cn>
Date:   Wed, 18 Jan 2023 14:05:50 +0800
User-Agent: Mozilla/5.0 (X11; Linux loongarch64; rv:68.0) Gecko/20100101
 Thunderbird/68.7.0
MIME-Version: 1.0
In-Reply-To: <48f0508f-3908-c6ca-c8ba-7c12dd6b3f11@loongson.cn>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Content-Language: en-US
Precedence: bulk


On 2023-01-18 12:23, Tiezhu Yang wrote:
>
>
> On 01/18/2023 12:14 PM, Huacai Chen wrote:
>> If memcpy should be blacklisted, then what about memset and memmove?
>
> According to the test results, there are no problems to probe
> memset and memmove, so no need to blacklist them for now,
> blacklist memcpy is because it may cause recursive exceptions,
> there is a detailed discussion in the following link:
>
> https://lore.kernel.org/lkml/20230114143859.7ccc45c1c5d9ce302113ab0a@kernel.org/ 
>

Hi, Tiezhu,

I cannot reproduce the results when kprobe memcpy. Could you please give
some details. Emm, I just replace "kernel_clone" with "memcpy" in
kprobe_example.c.

And for your call trace,

  handler_pre()
    pr_info()
      printk()
       _printk()
         vprintk()
           vprintk_store()
             memcpy()

I think when we should skip this time kprobe which triggered in
handler_{pre, post}. That means this time kprobe will not call
handler_{pre, post} agian, and not cause recursion. I remember
your codes had done this skip action. So, that's so strange if
recursion in handler_{pre, post}.


Thanks,

Jinyang


>
> Thanks,
> Tiezhu
>
>>
>> Huacai
>>
>> On Wed, Jan 18, 2023 at 10:01 AM Tiezhu Yang <yangtiezhu@loongson.cn> 
>> wrote:
>>>
>>> Some assembler symbols are not kprobe safe, such as handle_syscall
>>> (used as syscall exception handler), *memcpy* (may cause recursive
>>> exceptions), they can not be instrumented, just blacklist them for
>>> kprobing.
>>>
>>> Here is a related problem and discussion:
>>> Link: 
>>> https://lore.kernel.org/lkml/20230114143859.7ccc45c1c5d9ce302113ab0a@kernel.org/
>>>
>>> Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
>>> ---
>>>  arch/loongarch/include/asm/asm.h | 10 ++++++++++
>>>  arch/loongarch/kernel/entry.S    |  1 +
>>>  arch/loongarch/lib/memcpy.S      |  3 +++
>>>  3 files changed, 14 insertions(+)
>>>
>>> diff --git a/arch/loongarch/include/asm/asm.h 
>>> b/arch/loongarch/include/asm/asm.h
>>> index 40eea6a..f591b32 100644
>>> --- a/arch/loongarch/include/asm/asm.h
>>> +++ b/arch/loongarch/include/asm/asm.h
>>> @@ -188,4 +188,14 @@
>>>  #define PTRLOG         3
>>>  #endif
>>>
>>> +/* Annotate a function as being unsuitable for kprobes. */
>>> +#ifdef CONFIG_KPROBES
>>> +#define _ASM_NOKPROBE(name)                            \
>>> +       .pushsection "_kprobe_blacklist", "aw";         \
>>> +       .quad   name;                                   \
>>> +       .popsection
>>> +#else
>>> +#define _ASM_NOKPROBE(name)
>>> +#endif
>>> +
>>>  #endif /* __ASM_ASM_H */
>>> diff --git a/arch/loongarch/kernel/entry.S 
>>> b/arch/loongarch/kernel/entry.S
>>> index d53b631..55e23b1 100644
>>> --- a/arch/loongarch/kernel/entry.S
>>> +++ b/arch/loongarch/kernel/entry.S
>>> @@ -67,6 +67,7 @@ SYM_FUNC_START(handle_syscall)
>>>
>>>         RESTORE_ALL_AND_RET
>>>  SYM_FUNC_END(handle_syscall)
>>> +_ASM_NOKPROBE(handle_syscall)
>>>
>>>  SYM_CODE_START(ret_from_fork)
>>>         bl      schedule_tail           # a0 = struct task_struct *prev
>>> diff --git a/arch/loongarch/lib/memcpy.S b/arch/loongarch/lib/memcpy.S
>>> index 7c07d59..3b7e1de 100644
>>> --- a/arch/loongarch/lib/memcpy.S
>>> +++ b/arch/loongarch/lib/memcpy.S
>>> @@ -17,6 +17,7 @@ SYM_FUNC_START(memcpy)
>>>         ALTERNATIVE     "b __memcpy_generic", \
>>>                         "b __memcpy_fast", CPU_FEATURE_UAL
>>>  SYM_FUNC_END(memcpy)
>>> +_ASM_NOKPROBE(memcpy)
>>>
>>>  EXPORT_SYMBOL(memcpy)
>>>
>>> @@ -41,6 +42,7 @@ SYM_FUNC_START(__memcpy_generic)
>>>  2:     move    a0, a3
>>>         jr      ra
>>>  SYM_FUNC_END(__memcpy_generic)
>>> +_ASM_NOKPROBE(__memcpy_generic)
>>>
>>>  /*
>>>   * void *__memcpy_fast(void *dst, const void *src, size_t n)
>>> @@ -93,3 +95,4 @@ SYM_FUNC_START(__memcpy_fast)
>>>  3:     move    a0, a3
>>>         jr      ra
>>>  SYM_FUNC_END(__memcpy_fast)
>>> +_ASM_NOKPROBE(__memcpy_fast)
>>> -- 
>>> 2.1.0
>>>
>