Subject: Re: kernel hangs when kprobe memcpy
To:     "Masami Hiramatsu (Google)" <mhiramat@kernel.org>
References: <d179086d-78d8-d0e3-e113-9072cffa55f4@loongson.cn>
 <19666c03-4bf6-7aac-3f1d-cd31ab7de2d5@loongson.cn>
 <20230112233629.fafdbbe07dddf364f8078df6@kernel.org>
 <d0484b6e-c8a3-65c8-2157-0da95c17b061@loongson.cn>
 <20230114143859.7ccc45c1c5d9ce302113ab0a@kernel.org>
 <b20cfc27-0e46-7e3b-e4f1-2e185ea516ab@loongson.cn>
 <20230116154123.a0ff8d6deaff3fe87b48461b@kernel.org>
Cc:     "Naveen N. Rao" <naveen.n.rao@linux.ibm.com>,
        Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>,
        "David S. Miller" <davem@davemloft.net>,
        linux-trace-kernel@vger.kernel.org, linux-kernel@vger.kernel.org
From:   Tiezhu Yang <yangtiezhu@loongson.cn>
Message-ID: <357a5386-f588-f3c6-96de-42fcbb36a535@loongson.cn>
Date:   Tue, 31 Jan 2023 11:38:41 +0800
User-Agent: Mozilla/5.0 (X11; Linux mips64; rv:45.0) Gecko/20100101
 Thunderbird/45.4.0
MIME-Version: 1.0
In-Reply-To: <20230116154123.a0ff8d6deaff3fe87b48461b@kernel.org>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 01/16/2023 02:41 PM, Masami Hiramatsu (Google) wrote:
> Hi Tiezhu,
>
> On Sat, 14 Jan 2023 14:53:21 +0800
> Tiezhu Yang <yangtiezhu@loongson.cn> wrote:

...

>>> If you saw that without any change, please report it. At least
>>> memcpy is already marked as noinstr.
>>
>> The current upstream mainline kernel has no problem, because it includes
>> commit e3a9e681adb7 ("x86/entry: Fixup bad_iret vs noinstr"), memcpy is
>> already marked as noinstr. But for the kernel without the above commit,
>> like kernel 5.7, it has problem.
>
> I've confirmed that kernel 5.4.228 (the latest stable tree) did not have
> this issue (it already rejects the memcpy).

I just tested the stable kernel 5.4.230 on x86_64 fedora 36,
without any change, kernel hangs when execute cmd:
modprobe kprobe_example symbol="memcpy"

https://cdn.kernel.org/pub/linux/kernel/v5.x/linux-5.4.230.tar.xz

Am I missing something? Since 5.4 is a stable tree, should we do 
something to fix it?

Thanks,
Tiezhu