Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.35 as permitted sender) client-ip=23.128.96.35;
Message-ID: <162d90ab-e538-402e-90f1-304183bf6e76@gmail.com>
Date:   Sun, 8 Oct 2023 21:35:58 +0800
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: Fwd: kernel bug when performing heavy IO operations
To:     Matthew Wilcox <willy@infradead.org>
Cc:     Bagas Sanjaya <bagasdotme@gmail.com>, Chris Mason <clm@fb.com>,
        Josef Bacik <josef@toxicpanda.com>,
        David Sterba <dsterba@suse.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Linux btrfs <linux-btrfs@vger.kernel.org>,
        Linux Filesystem Development <linux-fsdevel@vger.kernel.org>
References: <f847bc14-8f53-0547-9082-bb3d1df9ae96@gmail.com>
 <ZOrG5698LPKTp5xM@casper.infradead.org>
 <7d8b4679-5cd5-4ba1-9996-1a239f7cb1c5@gmail.com>
 <ZOs5j93aAmZhrA/G@casper.infradead.org>
 <b290c417-de1b-4af8-9f5e-133abb79580d@gmail.com>
 <ZRPXlfGFWiCNZ6sh@casper.infradead.org>
Content-Language: en-US
From:   dianlujitao@gmail.com
In-Reply-To: <ZRPXlfGFWiCNZ6sh@casper.infradead.org>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

The problem does not occur with 6.1.56 lts kernel, not that old as you 
expected.

在 2023/9/27 15:19, Matthew Wilcox 写道:
> On Wed, Sep 27, 2023 at 01:36:52PM +0800, dianlujitao@gmail.com wrote:
>> Hello, I got some logs with 6.5.4 kernel from the official linux package of
>> Arch, no zen patches this time. Full dmesg is uploaded to
>> https://fars.ee/F1yM and below is a small snippet for your convenience, from
>> which PG_offline is no longer set:
>>
>> [177850.039441] BUG: Bad page map in process ld.lld pte:8000000edacc4025
>> pmd:147f96067
>> [177850.039454] page:000000007415dd6c refcount:22 mapcount:-237
>> mapping:00000000b0c37ca6 index:0x1075 pfn:0xedacc4
> It still looks like memory corruption to me.  If you go back to an older
> kernel (say 5.10 or 5.15) does the problem go away?  It's not really
> dispositive either way, since a newer kernel might drive the hardware
> closer to the edge, but it might give some clue.