Re: [RESEND RFC 3/3] mmc: sdhci: Add more debug info in case of data error

[Ritesh Harjani]

Hi Adrian,

Thanks for reviewing.

On 1/19/2017 5:24 PM, Adrian Hunter wrote:
> On 10/01/17 08:41, Ritesh Harjani wrote:
> > print error log message and dump sdhc registers for debugging
> > purpose in case of data errors (except when tuning commands
> > generate CRC/timeout/end bit errors).
>
> It is a bit ugly and probably only useful during driver development, so I am
> not at all enthusiastic about this.
Ok, since this patch was anyway for debugging, I will drop this patch 
for now. I will revisit in case if the need really arises, since this 
could be useful in debugging data txfr error and avoid re-test cycle to 
get sdhci register dumps.



Also,
Do you think we should have this below change added? Could you please 
give your thoughts on this. Jeremy and I feel that even below info can 
be helpful along with sdhci_dumpregs -
He requested me to take below diff as another patch in my next revision 
of this series which he has submitted here -

https://patchwork.kernel.org/patch/9442449/


diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c
index 71654b9..5911f98 100644
--- a/drivers/mmc/host/sdhci.c
+++ b/drivers/mmc/host/sdhci.c
@@ -47,6 +47,27 @@  static void sdhci_finish_data(struct sdhci_host *);

 static void sdhci_enable_preset_value(struct sdhci_host *host, bool 
enable);

+static void sdhci_dump_rpm_info(struct sdhci_host *host)
+{
+	struct mmc_host *mmc = host->mmc;
+
+	pr_info("%s: 
rpmstatus[pltfm](runtime-suspend:usage_count:disable_depth)(%d:%d:%d)\n",
+		mmc_hostname(mmc), mmc->parent->power.runtime_status,
+		atomic_read(&mmc->parent->power.usage_count),
+		mmc->parent->power.disable_depth);
+}
+
+
+static void sdhci_dump_state(struct sdhci_host *host)
+{
+	struct mmc_host *mmc = host->mmc;
+
+	pr_info("%s: clk: %d claimer: %s pwr: %d\n",
+		mmc_hostname(mmc), host->clock,
+		mmc->claimer->comm, host->pwr);
+		sdhci_dump_rpm_info(host);
+}
+
 static void sdhci_dumpregs(struct sdhci_host *host)
 {
 	pr_err(DRIVER_NAME ": =========== REGISTER DUMP (%s)===========\n",
@@ -100,6 +121,10 @@  static void sdhci_dumpregs(struct sdhci_host *host)
 			       readl(host->ioaddr + SDHCI_ADMA_ADDRESS));
 	}

+	if (host->ops->dump_vendor_regs)
+		host->ops->dump_vendor_regs(host);
+
+	sdhci_dump_state(host);
 	pr_err(DRIVER_NAME ": ===========================================\n");
 }

Regards
Ritesh

> > Signed-off-by: Ritesh Harjani <riteshh@xxxxxxxxxxxxxx>
> > Signed-off-by: Asutosh Das <asutoshd@xxxxxxxxxxxxxx>
> > ---
> >  drivers/mmc/host/sdhci.c | 21 +++++++++++++++++++--
> >  1 file changed, 19 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c
> > index 73a8918..2e51e49 100644
> > --- a/drivers/mmc/host/sdhci.c
> > +++ b/drivers/mmc/host/sdhci.c
> > @@ -2616,9 +2616,26 @@ static void sdhci_data_irq(struct sdhci_host *host, u32 intmask)
> >  			host->ops->adma_workaround(host, intmask);
> >  	}
> >
> > -	if (host->data->error)
> > +	if (host->data->error) {
> > +		bool pr_msg = true;
> > +
> > +		if (intmask & (SDHCI_INT_DATA_CRC | SDHCI_INT_DATA_TIMEOUT |
> > +		    SDHCI_INT_DATA_END_BIT)) {
> > +			command = SDHCI_GET_CMD(sdhci_readw(host,
> > +							    SDHCI_COMMAND));
> > +			if (command == MMC_SEND_TUNING_BLOCK_HS200 ||
> > +			    command == MMC_SEND_TUNING_BLOCK)
> > +				pr_msg = false;
> > +		}
> > +		if (pr_msg) {
> > +			pr_err("%s: data txfr (0x%08x) error: %d\n",
> > +			       mmc_hostname(host->mmc), intmask,
> > +			       host->data->error);
> > +			sdhci_dumpregs(host);
> > +		}
> > +
> >  		sdhci_finish_data(host);
> > -	else {
> > +	} else {
> >  		if (intmask & (SDHCI_INT_DATA_AVAIL | SDHCI_INT_SPACE_AVAIL))
> >  			sdhci_transfer_pio(host);

--
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project