Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] mips: kdump: Crash kernel should be able to see old
 memories
To:     Jiaxun Yang <jiaxun.yang@flygoat.com>,
        Thomas Bogendoerfer <tsbogend@alpha.franken.de>
References: <1618829807-12522-1-git-send-email-tangyouling@loongson.cn>
 <bd5b70cf-cfdf-0a4a-dfb5-08e030864138@flygoat.com>
Cc:     linux-mips@vger.kernel.org, linux-kernel@vger.kernel.org,
        Huacai Chen <chenhuacai@kernel.org>
From:   Youling Tang <tangyouling@loongson.cn>
Message-ID: <a04a736e-4f64-0491-165c-2544bb178d73@loongson.cn>
Date:   Tue, 20 Apr 2021 13:22:05 +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: <bd5b70cf-cfdf-0a4a-dfb5-08e030864138@flygoat.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi, Jiaxun

On 04/20/2021 09:11 AM, Jiaxun Yang wrote:
>
> 在 2021/4/19 18:56, Youling Tang 写道:
>> From: Huacai Chen <chenhc@lemote.com>
>>
>> kexec-tools use mem=X@Y to pass usable memories to crash kernel, but in
>> commit a94e4f24ec836c8984f83959 ("MIPS: init: Drop boot_mem_map") all
>> BIOS passed memories are removed by early_parse_mem(). I think this is
>> reasonable for a normal kernel but not for a crash kernel, because a
>> crash kernel should be able to see all old memories, even though it is
>> not supposed to use them.
>>
>> Fixes: a94e4f24ec836c8984f83959 ("MIPS: init: Drop boot_mem_map")
>> Signed-off-by: Huacai Chen <chenhuacai@kernel.org>
>> Signed-off-by: Youling Tang <tangyouling@loongson.cn>
>> ---
>>   arch/mips/kernel/setup.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/arch/mips/kernel/setup.c b/arch/mips/kernel/setup.c
>> index b86e241..ac90d3b 100644
>> --- a/arch/mips/kernel/setup.c
>> +++ b/arch/mips/kernel/setup.c
>> @@ -351,8 +351,10 @@ static int __init early_parse_mem(char *p)
>>        */
>>       if (usermem == 0) {
>>           usermem = 1;
>> +#ifndef CONFIG_CRASH_DUMP
>
> Why depend on a config instead of a runtime variable?
>
If not depend on config, we can determine whether the command line contains
the "elfcorehdr=" parameter, because the "mem=" and "elfcorhdr=" parameters
are automatically added in kexec-tools. So if there is an "elfcorehdr="
parameter in the command line, it means that the currently running kernel
is a capture kernel, and the memblock_remove() operation is not called.

The revised patch is as follows:
         if (usermem == 0) {
                 usermem = 1;
-               memblock_remove(memblock_start_of_DRAM(),
-                       memblock_end_of_DRAM() - memblock_start_of_DRAM());
+               if (!strstr(boot_command_line, "elfcorehdr")) {
+                       memblock_remove(memblock_start_of_DRAM(),
+                               memblock_end_of_DRAM() - 
memblock_start_of_DRAM());
+               }

Do you think it is feasible?
> Btw as you are fixing my commit please keep me CCed.
Sorry, I will add your CCed.

Thanks,
Youling
>
> Thanks.
>
>
> - Jiaxun
>