Received-SPF: pass (google.com: domain of linux-kernel+bounces-19140-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Message-ID: <2e7ad0c9-ec5a-a65a-899d-a575b6201510@quicinc.com>
Date: Mon, 8 Jan 2024 13:02:25 +0530
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.13.0
Subject: Re: [PATCH 1/7] [RFC PATCH 1/7] osdump: Add one new os minidump
 module
To: ruipeng qi <ruipengqi7@gmail.com>,
        Krzysztof Kozlowski
	<krzysztof.kozlowski@linaro.org>
CC: <catalin.marinas@arm.com>, <will@kernel.org>, <bhe@redhat.com>,
        <vgoyal@redhat.com>, <dyoung@redhat.com>,
        <linux-kernel@vger.kernel.org>, <kexec@lists.infradead.org>,
        <zalbassam@google.com>, <dianders@chromium.org>,
        <mark.rutland@arm.com>, <lecopzer.chen@mediatek.com>, <maz@kernel.org>,
        <arnd@arndb.de>, <yury.norov@gmail.com>, <brauner@kernel.org>,
        <mcgrof@kernel.org>, <maninder1.s@samsung.com>,
        <michael.christie@oracle.com>, <samitolvanen@google.com>,
        <linux-arm-kernel@lists.infradead.org>, <qiruipeng@lixiang.com>
References: <20231221130930.32634-1-ruipengqi7@gmail.com>
 <e9aa311f-6f13-410e-bb35-20320275668c@linaro.org>
 <CADHLONGLZaZT4Ot56AnxOf=ROrTWvge5=NUx5uqKqJhpJibfOQ@mail.gmail.com>
 <6fe05804-da74-45cd-a6f1-df4644671d99@linaro.org>
 <CADHLONE2YnBJGJUp2-vn_XEpc7+quTMWF7y1zpV0ygeL2VpQxA@mail.gmail.com>
 <CADHLONHHZAeubePPfZbG1N0Nv3ZJM5OWaJWqchBUpiN5Q27UdQ@mail.gmail.com>
Content-Language: en-US
From: Mukesh Ojha <quic_mojha@quicinc.com>
In-Reply-To: <CADHLONHHZAeubePPfZbG1N0Nv3ZJM5OWaJWqchBUpiN5Q27UdQ@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 8bit


On 12/23/2023 9:00 PM, ruipeng qi wrote:
> On Sat, Dec 23, 2023 at 11:13 PM ruipeng qi <ruipengqi7@gmail.com> wrote:
> 
>>> I don't see a problem why minidump cannot do it.
>> Qcom minidump allows one kernel client to register its region to dump, and then
>> driver will create and maintain one elf head for it.
>> But osdump will dump thousands of non-continuous regions. If you register one
>> by one, and add an elf head for each, it will cause heavy overhead. So it store
>> data in binary, and reassemble a standard elf format file offline for debug.
>>
> Osdump will only dump useful data, which may have been allocated before but
> freed now. In a word, useful data are changing all the time. I think this is not
> suitable for Qcom minidump's framework。


Hey @ruipeng,

I like the idea, However, as i already said here

https://lore.kernel.org/lkml/520e377d-e990-c185-4a20-07806873e506@quicinc.com/

This looks more close to pstore ramoops based on Ram backend, as your 
entire thing is based on RAM content to be preserved across boot and
that way it aligns to pstore ramoops concept.

Also most of the stuff like compression is already available in pstore.
So, let me know if you think that way, will be happy to collaborate and 
test this.

-Mukesh

> 
> Best Regards
> Ruipeng
>