Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH] MIPS: Scan the DMI system information
References: <5959f904-5c46-30a7-7a4f-17f692aca320@loongson.cn>
To:     Jiaxun Yang <jiaxun.yang@flygoat.com>,
        Paul Burton <paulburton@kernel.org>,
        Ralf Baechle <ralf@linux-mips.org>,
        James Hogan <jhogan@kernel.org>,
        Jean Delvare <jdelvare@suse.com>,
        Huacai Chen <chenhc@lemote.com>
Cc:     linux-mips@vger.kernel.org, linux-kernel@vger.kernel.org,
        Yinglu Yang <yangyinglu@loongson.cn>,
        Xuefeng Li <lixuefeng@loongson.cn>
From:   Tiezhu Yang <yangtiezhu@loongson.cn>
Message-ID: <5c042bd8-40ad-e84f-588d-f3ee56f7216d@loongson.cn>
Date:   Thu, 7 Nov 2019 17:26:58 +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: <5959f904-5c46-30a7-7a4f-17f692aca320@loongson.cn>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Sorry to resend this email because the mail list server was denied
due to it is not plain text.

On 11/07/2019 11:42 AM, Jiaxun Yang wrote:
> 于 2019年11月7日 GMT+08:00 上午10:42:23, Tiezhu Yang<yangtiezhu@loongson.cn>  写到:
>> On 11/07/2019 08:35 AM, Jiaxun Yang wrote:
>>> 于 2019年11月7日 GMT+08:00 上午12:05:41, Tiezhu Yang
>> <yangtiezhu@loongson.cn>  写到:
>>>> Enable DMI scanning on the MIPS architecture, this setups DMI
>>>> identifiers
>>>> (dmi_system_id) for printing it out on task dumps and prepares DIMM
>>>> entry
>>>> information (dmi_memdev_info) from the SMBIOS table. With this
>> patch,
>>>> the
>>>> driver can easily match various of mainboards.
>>>>
>>>> In the SMBIOS reference specification, the table anchor string
>> "_SM_"
>>>> is
>>>> present in the address range 0xF0000 to 0xFFFFF on a 16-byte
>> boundary,
>>>> but there exists a special case for loongson platform, when call
>>>> function
>>>> dmi_early_remap, it should specify the start address to 0xFFFE000
>> due
>>>> to
>>>> it is reserved for SMBIOS and can be normally access in the BIOS.
>>>>
>>>> Co-developed-by: Yinglu Yang<yangyinglu@loongson.cn>
>>>> Signed-off-by: Yinglu Yang<yangyinglu@loongson.cn>
>>>> Signed-off-by: Tiezhu Yang<yangtiezhu@loongson.cn>
>>>> ---
>>>> arch/mips/Kconfig           | 12 ++++++++++++
>>>> arch/mips/include/asm/dmi.h | 43
>>>> +++++++++++++++++++++++++++++++++++++++++++
>>>> arch/mips/kernel/setup.c    |  2 ++
>>>> 3 files changed, 57 insertions(+)
>>>> create mode 100644 arch/mips/include/asm/dmi.h
>>>>
>>>> diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
>>>> index 7cb8947..0a67b18 100644
>>>> --- a/arch/mips/Kconfig
>>>> +++ b/arch/mips/Kconfig
>>>> @@ -2757,6 +2757,18 @@ config HW_PERF_EVENTS
>>>> 	  Enable hardware performance counter support for perf events. If
>>>> 	  disabled, perf events will use software events only.
>>>>
>>>> +# Mark as expert because too many people got it wrong.
>>>> +# The code disables itself when not needed.
>>>> +config DMI
>>>> +	default y
>>>> +	select DMI_SCAN_MACHINE_NON_EFI_FALLBACK
>>>> +	bool "Enable DMI scanning" if EXPERT
>>>> +	help
>>>> +	  Enabled scanning of DMI to identify machine quirks. Say Y
>>>> +	  here unless you have verified that your setup is not
>>>> +	  affected by entries in the DMI blacklist. Required by PNP
>>>> +	  BIOS code.
>>>> +
>>>> config SMP
>>>> 	bool "Multi-Processing support"
>>>> 	depends on SYS_SUPPORTS_SMP
>>>> diff --git a/arch/mips/include/asm/dmi.h
>> b/arch/mips/include/asm/dmi.h
>>>> new file mode 100644
>>>> index 0000000..1f3da37
>>>> --- /dev/null
>>>> +++ b/arch/mips/include/asm/dmi.h
>>>> @@ -0,0 +1,43 @@
>>>> +/* SPDX-License-Identifier: GPL-2.0 */
>>>> +#ifndef _ASM_MIPS_DMI_H
>>>> +#define _ASM_MIPS_DMI_H
>>>> +
>>>> +#define dmi_early_remap		mips_early_memremap
>>>> +#define dmi_early_unmap		mips_early_memunmap
>>>> +#define dmi_remap(_x, _l)	mips_memremap(_x, _l, MEMREMAP_WB)
>>>> +#define dmi_unmap(_x)		mips_memunmap(_x)
>>>> +
>>>> +#define dmi_alloc(l)		memblock_alloc_low(l, PAGE_SIZE)
>>>> +
>>>> +void __init *mips_early_memremap(resource_size_t phys_addr,
>> unsigned
>>>> long size)
>>>> +{
>>>> +#if defined(CONFIG_MACH_LOONGSON64)
>>>> +	if (phys_addr == 0xF0000)
>>>> +		phys_addr = 0xFFFE000;
>>>> +
>>>> +	return (void *)TO_CAC(phys_addr);
>>>> +#else
>>>> +	return NULL;
>>>> +#endif
>>>> +}
>>> Hi Tiezhu,
>>>
>>> It is really tricky to hijack dmi address here during remap.
>>> I think we should set the dmi table address at  dmi_scan.c by a marco
>> or something else rather than hijack it during remap.
>>
>> Hi Jiaxun,
>>
>> Thanks for your review. I agree with you, let me think about it and try
>> to
>> find a proper way, and then I will send a v2 patch.

Hi Jiaxun,

It seems that there is no absolutely better way to handle this case.

1. use conditional compilation in drivers/firmware/dmi_scan.c:

#if defined(CONFIG_MACH_LOONGSON64)

p = dmi_early_remap(0xFFFE000, 0x10000);

#else

p = dmi_early_remap(0xF0000, 0x10000);

#endif


This will influence the common code.

2. use callback function in arch/mips/include/asm/dmi.h:

struct plat_dmi_ops {

         void (*early_memremap)(void);

         void (*memremap)(void);

};

extern struct plat_dmi_ops *dmi_ops;

void __init *mips_early_memremap(resource_size_t phys_addr, unsigned long size)

{

         dmi_ops->early_memremap();

}

void *mips_memremap(resource_size_t offset, size_t size, unsigned long flags)

{

         dmi_ops->memremap();

}


we can implement the callback function in various of MIPS platforms,
like this:

struct plat_dmi_ops loongson3_dmi_ops = {

         .early_memremap = loongson3_early_memremap,

         .memremap = loongson3_memremap,

};

register_dmi_ops(&loongson3_dmi_ops);

#ifdef CONFIG_DMI

void __init *loongson3_early_memremap(resource_size_t phys_addr, unsigned long size)

{

         if (phys_addr == 0xF0000)

                 phys_addr = 0xfffe000;

         return (void *)TO_CAC(phys_addr);

}

void *loongson3_memremap(resource_size_t offset, size_t size, unsigned long flags)

{

         return (void *)TO_CAC(phys_addr);

}

#else

void __init __iomem *loongson3_early_memremap(u64 phys_addr, unsigned long size)

{

         return NULL;

}

void __init __iomem *loongson3_memremap(u64 phys_addr, unsigned long size)

{

         return NULL;

}

#endif


This will not influence the common code.

What do you think?


Hi Jean,

Could you give some suggestions?

>>> Btw: Probably we should set DMI default y when MACH_LOONGSON64 is
>> set?
>>
>> CONFIG_DMI is set to y by default, I think there is no need to select
>> DMI when
>> set CONFIG_MACH_LOONGSON64.
> Sorry, I meant only default to y when MACH_LOONGSON64 is set or even depends on MACH_LOONGSON64. As Loongson is the only known MIPS platform that supports DMI. Default y unconditionally may lead to regression on other MIPS platforms.

OK, I prefer to select DMI when set MACH_LOONGSON64 instead of
DMI depends on MACH_LOONGSON64, I will modify the Kconfig file.

Thanks,

Tiezhu Yang

>> Thanks,
>>
>> Tiezhu Yang
>>
>>> Thanks.
>>>