linux-kernel - Re: [PATCH] mmc: sdhci-iproc: Add ACPI bindings for the rpi4

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Mon, 11 Jan 2021 15:43:54 -0600
From:   Jeremy Linton <jeremy.linton@....com>
To:     Stefan Wahren <stefan.wahren@...e.com>, linux-mmc@...r.kernel.org,
        Nicolas Saenz Julienne <nsaenzjulienne@...e.de>
Cc:     ulf.hansson@...aro.org, f.fainelli@...il.com,
        sbranden@...adcom.com, rjui@...adcom.com, adrian.hunter@...el.com,
        linux-kernel@...r.kernel.org,
        bcm-kernel-feedback-list@...adcom.com,
        linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH] mmc: sdhci-iproc: Add ACPI bindings for the rpi4

Hi,

On 1/11/21 6:23 AM, Stefan Wahren wrote:
> Hi,
> 
> Am 11.01.21 um 04:39 schrieb Jeremy Linton:
>> Hi,
>>
>> On 1/9/21 5:07 AM, Stefan Wahren wrote:
>>> Hi Jeremy,
>>>
>>> +add Nicolas
>>>
>>> Am 08.01.21 um 22:13 schrieb Jeremy Linton:
>>>> The rpi4 has a Arasan controller it carries over
>>>> from the rpi3, and a newer eMMC2 controller.
>>>> Because of a couple "quirks" it seems wiser to bind
>>>> these controllers to the same driver that DT is using
>>>> on this platform rather than the generic sdhci_acpi
>>>> driver with PNP0D40.
>>>>
>>>> So, we use BCM2847 for the older Arasan and BRCME88C
>>>> for the newer eMMC2.
>>>>
>>>> With this change linux is capable of utilizing the
>>>> SD card slot, and the wifi on this platform
>>>> with linux.
>>>>
>>>> Signed-off-by: Jeremy Linton <jeremy.linton@....com>
>>>> ---
>>>>    drivers/mmc/host/sdhci-iproc.c | 14 ++++++++++++++
>>>>    1 file changed, 14 insertions(+)
>>>>
>>>> diff --git a/drivers/mmc/host/sdhci-iproc.c
>>>> b/drivers/mmc/host/sdhci-iproc.c
>>>> index c9434b461aab..f79d97b41805 100644
>>>> --- a/drivers/mmc/host/sdhci-iproc.c
>>>> +++ b/drivers/mmc/host/sdhci-iproc.c
>>>> @@ -250,6 +250,14 @@ static const struct sdhci_pltfm_data
>>>> sdhci_bcm2835_pltfm_data = {
>>>>        .ops = &sdhci_iproc_32only_ops,
>>>>    };
>>>>    +static const struct sdhci_pltfm_data sdhci_bcm_arasan_data = {
>>>> +    .quirks = SDHCI_QUIRK_BROKEN_CARD_DETECTION |
>>>> +          SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK |
>>>> +          SDHCI_QUIRK_NO_HISPD_BIT,
>>>> +    .quirks2 = SDHCI_QUIRK2_PRESET_VALUE_BROKEN,
>>>> +    .ops = &sdhci_iproc_32only_ops,
>>>> +};
>>
>> First, thanks for taking a look at this!
>>
>>
>>> Why do we need an almost exact copy of bcm2835_data which works fine for
>>> all Raspberry Pi boards?
>>
>> The short answer to the remainder of this email is that i'm trying to
>> continue supporting existing OSs (windows) using the ACPI tables on
>> the rpi3/rpi4 while adding rpi4+Linux support.
>>
>> An even shorter answer is they don't work because ACPI doesn't provide
>> the same clock/attributes/etc controls that exist with DT.
>>
>> So, what happened here is that I got this controller "working" with
>> the generic PNP0D40 sdhci_acpi driver. I managed this only by
>> controlling the sdhci_caps/masks in the firmware. In theory this
>> minimizes the amount of work needed on the other OS which are booting
>> on the same ACPI tables (*bsds). They should only need to quirk the
>> bcm/arasan specific functionality, rather than some of the quirking
>> which change the caps behavior. But because we don't know which if any
>> of the older rpi/arasan quirks are still needed the safest solution is
>> to use the _iproc driver and just drop the quirk flags known to be
>> worked around by the firmware caps override.
> 
> okay, thanks for the explanation. I was also confused by bcm_arasan,
> because there is already an Arasan specific sdhci driver. But now it's
> clear to me.
> 
> Could you please add a short comment (above sdhci_bcm_arasan_data) why
> we cannot use bcm2835_data?

Sure.

> 
> Btw the subject isn't complete. The patch is also related to the rpi3.

Only via the historical ACPI tables. There isn't any attempt to get 
ACPI+Linux working on the rpi3. Its to far away from BSA. For starters 
it doesn't have a GIC. So while one could bind this driver on the rpi3, 
that would require being able to boot Linux on the rpi3 in ACPI mode.

Thanks again!



> 
> Best regards
> Stefan
> 
>