Re: [PATCH v2] mfd:rtsx: do retry when dma transfer error

[Lee Jones]

On Thu, 05 Jan 2017, åälinux wrote:

> > +		err = -EILSEQ;
> > Why are you using this error code?
> >
> 
> when kernel receive the error number is EILSEQ, the SD card requeset
> will be retried for some times, instead of dropping it directly.
> so we use EILSEQ and frequency reduction to ensure data can be 
> transmitted correctly as much as possible.

So this is function is SD card specific?

In which case, why isn't it in the SD card driver?

> steven feng
> Realsil Microelectronics CO. LTD.
> Mobileï181-6899-0403  Extï57594
> 
> å 2017å01æ04æ 20:12, Lee Jones åé:
> > On Wed, 04 Jan 2017, steven_feng@xxxxxxxxxxxxxx wrote:
> >
> >> From: steven_feng <steven_feng@xxxxxxxxxxxxxx>
> >>
> >> the request should be reissued when dma transfer error.
> >> for rts5227, the clock freq need to step reduce when error occurred.
> >>
> >> Signed-off-by: steven_feng <steven_feng@xxxxxxxxxxxxxx>
> >> ---
> >>  drivers/mfd/rtsx_pcr.c       | 16 ++++++++++++++--
> >>  include/linux/mfd/rtsx_pci.h |  4 ++++
> >>  2 files changed, 18 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/drivers/mfd/rtsx_pcr.c b/drivers/mfd/rtsx_pcr.c
> >> index 98029ee..63ee0ac 100644
> >> --- a/drivers/mfd/rtsx_pcr.c
> >> +++ b/drivers/mfd/rtsx_pcr.c
> >> @@ -30,6 +30,7 @@
> >>  #include <linux/platform_device.h>
> >>  #include <linux/mfd/core.h>
> >>  #include <linux/mfd/rtsx_pci.h>
> >> +#include <linux/mmc/card.h>
> >>  #include <asm/unaligned.h>
> >>  
> >>  #include "rtsx_pcr.h"
> >> @@ -452,8 +453,12 @@ int rtsx_pci_dma_transfer(struct rtsx_pcr *pcr, struct scatterlist *sglist,
> >>  	}
> >>  
> >>  	spin_lock_irqsave(&pcr->lock, flags);
> >> -	if (pcr->trans_result == TRANS_RESULT_FAIL)
> >> -		err = -EINVAL;
> >> +	if (pcr->trans_result == TRANS_RESULT_FAIL) {
> >> +		err = -EILSEQ;
> > Why are you using this error code?
> >
> >> +		if (pcr->dma_error_count < 8)
> > Please define the 8.
> >
> > RTSX_MAX_FREQ_REDUCTIONS or similar.
> >
> >> +			pcr->dma_error_count++;
> >> +	}
> >> +
> >>  	else if (pcr->trans_result == TRANS_NO_DEVICE)
> >>  		err = -ENODEV;
> >>  	spin_unlock_irqrestore(&pcr->lock, flags);
> >> @@ -659,6 +664,12 @@ int rtsx_pci_switch_clock(struct rtsx_pcr *pcr, unsigned int card_clock,
> >>  	if (err < 0)
> >>  		return err;
> >>  
> >> +	/* each time dma transfer error occurs, card clock decreased by 20MHz */
> > "Each"
> >
> >> +	if (card_clock == UHS_SDR104_MAX_DTR &&
> >> +		pcr->dma_error_count && PCI_PID(pcr) == RTS5227_DEVICE_ID)
> > Suggest placing "PCI_PID(pcr) == RTS5227_DEVICE_ID" on a new line.
> >
> >> +		card_clock = UHS_SDR104_MAX_DTR -
> >> +			pcr->dma_error_count * 20000000;
> > Suggest placing parantheis around:
> >
> > (UHS_SDR104_MAX_DTR - pcr->dma_error_count)
> >
> >>  	card_clock /= 1000000;
> >>  	pcr_dbg(pcr, "Switch card clock to %dMHz\n", card_clock);
> >>  
> >> @@ -894,6 +905,7 @@ static irqreturn_t rtsx_pci_isr(int irq, void *dev_id)
> >>  			pcr->card_removed |= SD_EXIST;
> >>  			pcr->card_inserted &= ~SD_EXIST;
> >>  		}
> >> +		pcr->dma_error_count = 0;
> >>  	}
> >>  
> >>  	if (int_reg & MS_INT) {
> >> diff --git a/include/linux/mfd/rtsx_pci.h b/include/linux/mfd/rtsx_pci.h
> >> index 7eb7cba..751c4042 100644
> >> --- a/include/linux/mfd/rtsx_pci.h
> >> +++ b/include/linux/mfd/rtsx_pci.h
> >> @@ -850,6 +850,8 @@
> >>  
> >>  #define rtsx_pci_init_cmd(pcr)		((pcr)->ci = 0)
> >>  
> >> +#define RTS5227_DEVICE_ID		0x5227
> >> +
> >>  struct rtsx_pcr;
> >>  
> >>  struct pcr_handle {
> >> @@ -957,6 +959,8 @@ struct rtsx_pcr {
> >>  
> >>  	int				num_slots;
> >>  	struct rtsx_slot		*slots;
> >> +
> >> +	u8				dma_error_count;
> >>  };
> >>  
> >>  #define CHK_PCI_PID(pcr, pid)		((pcr)->pci->device == (pid))
> 

> begin:vcard
> fn;quoted-printable:=E5=86=AF=E4=BC=9F
> n;quoted-printable:;=E5=86=AF=E4=BC=9F
> email;internet:steven_feng@xxxxxxxxxxxxxx
> tel;cell:18168990403
> version:2.1
> end:vcard
> 


-- 
Lee Jones
Linaro STMicroelectronics Landing Team Lead
Linaro.org â Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog