Re: [PATCH] mtd: rawnand: arasan: Fix clock rate in NV-DDR

From: Olga Kitaina
Date: Tue Nov 30 2021 - 15:08:38 EST


Hi Miquel,

On 30.11.2021 10:20, Miquel Raynal wrote:
> Hi Olga,
>
> okitain@xxxxxxxxx wrote on Mon, 29 Nov 2021 21:06:05 +0300:
>
>> Hi Miquel,
>>
>> On 29.11.2021 11:55, Miquel Raynal wrote:
>>> Hi Olga,
>>>
>>> Please add all the MTD maintainers in copy, as requested by
>>> get_maintainers.pl.
>>>
>>> okitain@xxxxxxxxx wrote on Sat, 27 Nov 2021 21:07:58
>>> +0300:
>>>
>>>> According to the Arasan NAND controller spec,
>>>> the flash clock rate for SDR must be <= 100 MHz,
>>>> while for NV-DDR it must be the same as the rate
>>>> of the CLK line for the mode.
>>>
>>> I completely missed that, where did you get the information?
>>
>> The "Data Interface Transitions" chapter of the spec contains timings for flash clock setup in NV-DDR
>> and NV-DDR2 modes. The "time period" of those clocks is equal to tCK in NV-DDR and tRC in NV-DDR2.
>>
>> The same chapter should have information about necessary steps to switch from NV-DDR to SDR,
>> which includes setting the flash clock to 100 MHz.
>>
>>
>> Just to make sure i'm not shooting myself in the foot: am I changing the right clock?
>> The documentation points out that we have to change flash_clk, which i thought was
>> nfc->controller_clk and set up by anand->clk, but it seems like it might actually be nfc->bus_clk.
>
> I believe I made a serious mistake, re-reading the code it feels like
> I'm changing the system's clock (which basically changes nothing in our
> case) instead of changing the NAND bus clock.
>
>> In that case, does setting nfc->controller_clk to 100 MHz by default make sense?
>> There isn't a hard limit on what the system clock might be (beyond a specific SoC),
>> but there are timing requirements for the flash clock, and so setting a specific
>> system clock frequency seems unnecessary for most devices.
>>
>
> Please create a two-patch series:
> 1- Setting the right clock in the current code base (inverting bus_clk
> and controller_clk where relevant, setting one to 100MHz and letting
> the other as it is)
> 2- Changing the default NV-DDR rate based on tCK (below patch).
>
> Do you have the necessary hardware for testing?

I'm sorry to say - I do not. The SoC this problem was initially noticed on can't run latest Linux,
and even if it did I have no way of acquiring an NV-DDR-capable flash.

Since Bootlin merged in NV-DDR support into the kernel, is it possible for you to test
the next iteration of this patch series on NV-DDR hardware as well?
Say, by purposefully preventing NV-DDR mode 5 from being chosen in anfc_setup_interface()?

>>>> The driver previously always set 100 MHz for NV-DDR, which
>>>> would result in incorrect behavior for NV-DDR modes 0-4.
>>>>
>>>> The appropriate clock rate can be calculated
>>>> from the NV-DDR timing parameters as 1/tCK, or for rates
>>>> measured in picoseconds, 10^12 / nand_nvddr_timings->tCK_min.
>>>>
>>>
>>> You need a couple of Fixes + Cc: stable tags here, otherwise the
>>> patch looks good to me.
>>>
>>
>> Will include in the next iteration of the patch, thank you.
>>
>>>> Signed-off-by: Olga Kitaina <okitain@xxxxxxxxx>
>>>> ---
>>>> drivers/mtd/nand/raw/arasan-nand-controller.c | 8 +++++++-
>>>> 1 file changed, 7 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/drivers/mtd/nand/raw/arasan-nand-controller.c b/drivers/mtd/nand/raw/arasan-nand-controller.c
>>>> index 53bd10738418..ed4ee9942441 100644
>>>> --- a/drivers/mtd/nand/raw/arasan-nand-controller.c
>>>> +++ b/drivers/mtd/nand/raw/arasan-nand-controller.c
>>>> @@ -18,6 +18,7 @@
>>>> #include <linux/gpio/consumer.h>
>>>> #include <linux/interrupt.h>
>>>> #include <linux/iopoll.h>
>>>> +#include <linux/math64.h>
>>>> #include <linux/module.h>
>>>> #include <linux/mtd/mtd.h>
>>>> #include <linux/mtd/partitions.h>
>>>> @@ -1043,7 +1044,12 @@ static int anfc_setup_interface(struct nand_chip *chip, int target,
>>>> DQS_BUFF_SEL_OUT(dqs_mode);
>>>> }
>>>>
>>>> - anand->clk = ANFC_XLNX_SDR_DFLT_CORE_CLK;
>>>> + if (nand_interface_is_sdr)
>>>> + anand->clk = ANFC_XLNX_SDR_DFLT_CORE_CLK;
>>>> + else
>>>> + /* ONFI timings are defined in picoseconds */
>>>> + anand->clk = div_u64((u64)NSEC_PER_SEC * 1000,
>>>> + conf->timings.nvddr.tCK_min);
>>>>
>>>> /*
>>>> * Due to a hardware bug in the ZynqMP SoC, SDR timing modes 0-1 work
>>>>
>>>> base-commit: f53d4c109a666bf1a4883b45d546fba079258717
>>>
>>>
>>> Thanks,
>>> Miquèl
>>>
>
>
> Thanks,
> Miquèl
>

Thanks,
Olga.