| Message ID | 20210227135354.1696310-1-festevam@gmail.com |
|---|---|
| State | Rejected |
| Delegated to: | Stefano Babic |
| Headers | show |
| Series | Revert "mmc: fsl_esdhc_imx: remove the 1ms delay before sending command" | expand |
> -----Original Message----- > From: Fabio Estevam [mailto:festevam@gmail.com] > Sent: 2021年2月27日 21:54 > To: Peng Fan <peng.fan@nxp.com> > Cc: u-boot@lists.denx.de; lorusak@gmail.com; > andrey.zhizhikin@leica-geosystems.com; Ye Li <ye.li@nxp.com>; Bough Chen > <haibo.chen@nxp.com>; sbabic@denx.de; trini@konsulko.com; Fabio Estevam > <festevam@gmail.com> > Subject: [PATCH] Revert "mmc: fsl_esdhc_imx: remove the 1ms delay before > sending command" > > Removing the 1ms delay before sending command causes a regression on > imx8mq-evk where the SD card cannot be accessed. > > Since this 1ms delay has been present since the driver introduction in 2008, > keep it to avoid regressions. > > This reverts commit 9098682200e6cca4b776638a51200dafa16f50fb. > Hi Fabio, This 1ms delay exist in esdhc_send_cmd_common, which means every time we send a command, will delay 1ms, which will involve a huge delay overall for the whole u-boot cycle. For the issue you meet, I already find the root cause, and already reply in the mail loop: [EXT] Re: [PATCH 4/4] imx8mq_evk: Enable the USB3.0 host port So I think no need to involve this 1ms back again. Best regards Haibo Chen > Reported-by: Ye Li <ye.li@nxp.com> > Signed-off-by: Fabio Estevam <festevam@gmail.com> > --- > drivers/mmc/fsl_esdhc_imx.c | 7 +++++++ > 1 file changed, 7 insertions(+) > > diff --git a/drivers/mmc/fsl_esdhc_imx.c b/drivers/mmc/fsl_esdhc_imx.c index > e0e132698e30..a0a5c8960563 100644 > --- a/drivers/mmc/fsl_esdhc_imx.c > +++ b/drivers/mmc/fsl_esdhc_imx.c > @@ -463,6 +463,13 @@ static int esdhc_send_cmd_common(struct > fsl_esdhc_priv *priv, struct mmc *mmc, > while (esdhc_read32(®s->prsstat) & PRSSTAT_DLA) > ; > > + /* Wait at least 8 SD clock cycles before the next command */ > + /* > + * Note: This is way more than 8 cycles, but 1ms seems to > + * resolve timing issues with some cards > + */ > + udelay(1000); > + > /* Set up for a data transfer if we have one */ > if (data) { > err = esdhc_setup_data(priv, mmc, data); > -- > 2.25.1
diff --git a/drivers/mmc/fsl_esdhc_imx.c b/drivers/mmc/fsl_esdhc_imx.c index e0e132698e30..a0a5c8960563 100644 --- a/drivers/mmc/fsl_esdhc_imx.c +++ b/drivers/mmc/fsl_esdhc_imx.c @@ -463,6 +463,13 @@ static int esdhc_send_cmd_common(struct fsl_esdhc_priv *priv, struct mmc *mmc, while (esdhc_read32(®s->prsstat) & PRSSTAT_DLA) ; + /* Wait at least 8 SD clock cycles before the next command */ + /* + * Note: This is way more than 8 cycles, but 1ms seems to + * resolve timing issues with some cards + */ + udelay(1000); + /* Set up for a data transfer if we have one */ if (data) { err = esdhc_setup_data(priv, mmc, data);
Removing the 1ms delay before sending command causes a regression on imx8mq-evk where the SD card cannot be accessed. Since this 1ms delay has been present since the driver introduction in 2008, keep it to avoid regressions. This reverts commit 9098682200e6cca4b776638a51200dafa16f50fb. Reported-by: Ye Li <ye.li@nxp.com> Signed-off-by: Fabio Estevam <festevam@gmail.com> --- drivers/mmc/fsl_esdhc_imx.c | 7 +++++++ 1 file changed, 7 insertions(+)