Re: [PATCH V4 6/7] mmc: sdhci: Set error state for mmc driver

[Adrian Hunter]

On 12/03/2022 20:15, Sajida Bhanu (Temp) (QUIC) wrote:
> Hi,
> 
> Thanks for the review.
> 
> Please find the inline comments.
> 
> Thanks,
> Sajida
>> -----Original Message-----
>> From: Adrian Hunter <adrian.hunter@xxxxxxxxx>
>> Sent: Tuesday, March 8, 2022 3:36 PM
>> To: Sajida Bhanu (Temp) (QUIC) <quic_c_sbhanu@xxxxxxxxxxx>; Asutosh
>> Das (asd) <asutoshd@xxxxxxxxxxx>; ulf.hansson@xxxxxxxxxx;
>> agross@xxxxxxxxxx; bjorn.andersson@xxxxxxxxxx; linux-
>> mmc@xxxxxxxxxxxxxxx; linux-arm-msm@xxxxxxxxxxxxxxx; linux-
>> kernel@xxxxxxxxxxxxxxx
>> Cc: Veerabhadrarao Badiganti (QUIC) <quic_vbadigan@xxxxxxxxxxx>; Ram
>> Prakash Gupta (QUIC) <quic_rampraka@xxxxxxxxxxx>; Pradeep Pragallapati
>> (QUIC) <quic_pragalla@xxxxxxxxxxx>; Sarthak Garg (QUIC)
>> <quic_sartgarg@xxxxxxxxxxx>; Nitin Rawat (QUIC)
>> <quic_nitirawa@xxxxxxxxxxx>; Sayali Lokhande (QUIC)
>> <quic_sayalil@xxxxxxxxxxx>; Sayali Lokhande <sayalil@xxxxxxxxxxxxxx>; Bao
>> D . Nguyen <nguyenb@xxxxxxxxxxxxxx>
>> Subject: Re: [PATCH V4 6/7] mmc: sdhci: Set error state for mmc driver
>>
>> On 2.3.2022 15.03, Shaik Sajida Bhanu wrote:
>>> Set error state if any errors observed in eMMC and SD card driver level.
>>>
>>> Signed-off-by: Liangliang Lu <luliang@xxxxxxxxxxxxxx>
>>> Signed-off-by: Sayali Lokhande <sayalil@xxxxxxxxxxxxxx>
>>> Signed-off-by: Bao D. Nguyen <nguyenb@xxxxxxxxxxxxxx>
>>> Signed-off-by: Ram Prakash Gupta <quic_rampraka@xxxxxxxxxxx>
>>> Signed-off-by: Shaik Sajida Bhanu <quic_c_sbhanu@xxxxxxxxxxx>
>>> ---
>>>  drivers/mmc/host/sdhci.c | 22 ++++++++++++++++++++++
>>>  1 file changed, 22 insertions(+)
>>>
>>> diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c index
>>> 741fb06..4ba3797 100644
>>> --- a/drivers/mmc/host/sdhci.c
>>> +++ b/drivers/mmc/host/sdhci.c
>>> @@ -224,6 +224,8 @@ void sdhci_reset(struct sdhci_host *host, u8 mask)
>>>  		if (timedout) {
>>>  			pr_err("%s: Reset 0x%x never completed.\n",
>>>  				mmc_hostname(host->mmc), (int)mask);
>>> +			if (host->mmc && !host->mmc->err_state)
>>> +				mmc_debugfs_err_stats_enable(host-
>>> mmc);
>>
>> I really do not understand why this cannot be another err_stats entry such as
>> MMC_ERR_DRIVER and then as below?
>>
>> 			sdhci_err_stats_inc(host, DRIVER);
>>
>>
> Hi,
> 
> err_stats having all the error types.
> err_state is not the error type, if err_state set means some errors occurred in driver and we will go and check err_stats to know exact error.


In sdhci.c use only sdhci_err_stats_inc() adding err_stats entries as needed.
Seems like there are controller timeouts (say MMC_ERR_CTRL_TIMEOUT) and
unexpected interrupts (say MMC_ERR_UNEXPECTED_IRQ)

In drivers/mmc/core/debugfs.c calculate err_state as needed e.g.

static int mmc_err_state_get(void *data, u64 *val)
{
	struct mmc_host *host = data;

	*val = host->err_stats[MMC_ERR_REQ_TIMEOUT] ||
	       host->err_stats[MMC_ERR_CTRL_TIMEOUT] ||
	       host->err_stats[MMC_ERR_UNEXPECTED_IRQ] ||
	       <etc>

	return 0;
}