| Message ID | 1441097207-2162-1-git-send-email-clsee@altera.com |
|---|---|
| State | Superseded |
| Delegated to: | Marek Vasut |
| Headers | show |
On Tuesday, September 01, 2015 at 10:46:47 AM, Chin Liang See wrote: > Ensuring spi_calibration is run when there is a change of sclk > frequency. This will ensure the qspi flash access works for high > sclk frequency > > Signed-off-by: Chin Liang See <clsee@altera.com> > Cc: Dinh Nguyen <dinguyen@opensource.altera.com> > Cc: Marek Vasut <marex@denx.de> > Cc: Stefan Roese <sr@denx.de> > Cc: Vikas Manocha <vikas.manocha@st.com> > Cc: Jagannadh Teki <jteki@openedev.com> > Cc: Pavel Machek <pavel@denx.de> > --- > drivers/spi/cadence_qspi.c | 3 +++ > 1 files changed, 3 insertions(+), 0 deletions(-) > > diff --git a/drivers/spi/cadence_qspi.c b/drivers/spi/cadence_qspi.c > index 34a0f46..512bf2d 100644 > --- a/drivers/spi/cadence_qspi.c > +++ b/drivers/spi/cadence_qspi.c > @@ -128,6 +128,9 @@ static int cadence_spi_set_speed(struct udevice *bus, > uint hz) > > cadence_spi_write_speed(bus, hz); > > + /* to ensure spi_calibration is run when SCLK frequency change */ > + plat->max_hz = hz; > + This looks like a hack, doesn't this change just subvert the condition below to enforce the calibration ? > /* Calibration required for different SCLK speed or chip select */ > if (priv->qspi_calibrated_hz != plat->max_hz || > priv->qspi_calibrated_cs != spi_chip_select(bus)) { Best regards, Marek Vasut
On Tue, 2015-09-01 at 11:01 +0200, marex@denx.de wrote: > On Tuesday, September 01, 2015 at 10:46:47 AM, Chin Liang See wrote: > > Ensuring spi_calibration is run when there is a change of sclk > > frequency. This will ensure the qspi flash access works for high > > sclk frequency > > > > Signed-off-by: Chin Liang See <clsee@altera.com> > > Cc: Dinh Nguyen <dinguyen@opensource.altera.com> > > Cc: Marek Vasut <marex@denx.de> > > Cc: Stefan Roese <sr@denx.de> > > Cc: Vikas Manocha <vikas.manocha@st.com> > > Cc: Jagannadh Teki <jteki@openedev.com> > > Cc: Pavel Machek <pavel@denx.de> > > --- > > drivers/spi/cadence_qspi.c | 3 +++ > > 1 files changed, 3 insertions(+), 0 deletions(-) > > > > diff --git a/drivers/spi/cadence_qspi.c b/drivers/spi/cadence_qspi.c > > index 34a0f46..512bf2d 100644 > > --- a/drivers/spi/cadence_qspi.c > > +++ b/drivers/spi/cadence_qspi.c > > @@ -128,6 +128,9 @@ static int cadence_spi_set_speed(struct udevice *bus, > > uint hz) > > > > cadence_spi_write_speed(bus, hz); > > > > + /* to ensure spi_calibration is run when SCLK frequency change */ > > + plat->max_hz = hz; > > + > > This looks like a hack, doesn't this change just subvert the condition > below to enforce the calibration ? Nope. I checked through debugger where plat->max_hz is initialized once with the value from device tree. When you use sf probe to change frequency, the value is not reflected to plat->max_hz. Hence the calibration is not run when frequency change. That lead to the sf probe fail when running at 80MHz or higher. Note the priv->qspi_calibrated_hz is updated after calibration succeed. Thanks Chin Liang > > > /* Calibration required for different SCLK speed or chip select */ > > if (priv->qspi_calibrated_hz != plat->max_hz || > > priv->qspi_calibrated_cs != spi_chip_select(bus)) { > > Best regards, > Marek Vasut
On Tue, 2015-09-01 at 04:13 -0500, Chin Liang See wrote: > On Tue, 2015-09-01 at 11:01 +0200, marex@denx.de wrote: > > On Tuesday, September 01, 2015 at 10:46:47 AM, Chin Liang See wrote: > > > Ensuring spi_calibration is run when there is a change of sclk > > > frequency. This will ensure the qspi flash access works for high > > > sclk frequency > > > > > > Signed-off-by: Chin Liang See <clsee@altera.com> > > > Cc: Dinh Nguyen <dinguyen@opensource.altera.com> > > > Cc: Marek Vasut <marex@denx.de> > > > Cc: Stefan Roese <sr@denx.de> > > > Cc: Vikas Manocha <vikas.manocha@st.com> > > > Cc: Jagannadh Teki <jteki@openedev.com> > > > Cc: Pavel Machek <pavel@denx.de> > > > --- > > > drivers/spi/cadence_qspi.c | 3 +++ > > > 1 files changed, 3 insertions(+), 0 deletions(-) > > > > > > diff --git a/drivers/spi/cadence_qspi.c b/drivers/spi/cadence_qspi.c > > > index 34a0f46..512bf2d 100644 > > > --- a/drivers/spi/cadence_qspi.c > > > +++ b/drivers/spi/cadence_qspi.c > > > @@ -128,6 +128,9 @@ static int cadence_spi_set_speed(struct udevice *bus, > > > uint hz) > > > > > > cadence_spi_write_speed(bus, hz); > > > > > > + /* to ensure spi_calibration is run when SCLK frequency change */ > > > + plat->max_hz = hz; > > > + > > > > This looks like a hack, doesn't this change just subvert the condition > > below to enforce the calibration ? > > Nope. I checked through debugger where plat->max_hz is initialized once > with the value from device tree. When you use sf probe to change > frequency, the value is not reflected to plat->max_hz. Hence the > calibration is not run when frequency change. That lead to the sf probe > fail when running at 80MHz or higher. Note the priv->qspi_calibrated_hz > is updated after calibration succeed. > > Thanks > Chin Liang Here are the useful info U-Boot 2015.10-rc2-06835-gf7d7156-dirty (Sep 01 2015 - 08:29:14 +0000) CPU: Altera SoCFPGA Platform FPGA: Altera Cyclone V, SE/A6 or SX/C6 or ST/D6, version 0x0 BOOT: SD/MMC External Transceiver (1.8V) Watchdog enabled I2C: ready DRAM: 1 GiB MMC: SOCFPGA DWMMC: 0 Using default environment In: serial Out: serial Err: serial Model: Altera SOCFPGA Cyclone V SoC Development Kit Net: Error: ethernet@ff702000 address not set. No ethernet found. Hit any key to stop autoboot: 0 => sf probe 0 1000000 SF: Detected N25Q512 with page size 256 Bytes, erase size 4 KiB, total 64 MiB => md ff705010 1 ff705010: 0000000f .... => sf probe 0 10000000 SF: Detected N25Q512 with page size 256 Bytes, erase size 4 KiB, total 64 MiB => md ff705010 1 ff705010: 0000000f .... => sf probe 0 20000000 SF: Detected N25Q512 with page size 256 Bytes, erase size 4 KiB, total 64 MiB => md ff705010 1 ff705010: 0000000d .... => sf probe 0 50000000 SF: Detected N25Q512 with page size 256 Bytes, erase size 4 KiB, total 64 MiB => md ff705010 1 ff705010: 00000007 .... => sf probe 0 80000000 SF: Detected N25Q512 with page size 256 Bytes, erase size 4 KiB, total 64 MiB => md ff705010 1 ff705010: 00000007 .... => sf probe 0 100000000 SF: Detected N25Q512 with page size 256 Bytes, erase size 4 KiB, total 64 MiB => md ff705010 1 ff705010: 00000007 .... Thanks Chin Liang > > > > > > /* Calibration required for different SCLK speed or chip select */ > > > if (priv->qspi_calibrated_hz != plat->max_hz || > > > priv->qspi_calibrated_cs != spi_chip_select(bus)) { > > > > Best regards, > > Marek Vasut >
On Tuesday, September 01, 2015 at 11:13:38 AM, Chin Liang See wrote: > On Tue, 2015-09-01 at 11:01 +0200, marex@denx.de wrote: > > On Tuesday, September 01, 2015 at 10:46:47 AM, Chin Liang See wrote: > > > Ensuring spi_calibration is run when there is a change of sclk > > > frequency. This will ensure the qspi flash access works for high > > > sclk frequency > > > > > > Signed-off-by: Chin Liang See <clsee@altera.com> > > > Cc: Dinh Nguyen <dinguyen@opensource.altera.com> > > > Cc: Marek Vasut <marex@denx.de> > > > Cc: Stefan Roese <sr@denx.de> > > > Cc: Vikas Manocha <vikas.manocha@st.com> > > > Cc: Jagannadh Teki <jteki@openedev.com> > > > Cc: Pavel Machek <pavel@denx.de> > > > --- > > > > > > drivers/spi/cadence_qspi.c | 3 +++ > > > 1 files changed, 3 insertions(+), 0 deletions(-) > > > > > > diff --git a/drivers/spi/cadence_qspi.c b/drivers/spi/cadence_qspi.c > > > index 34a0f46..512bf2d 100644 > > > --- a/drivers/spi/cadence_qspi.c > > > +++ b/drivers/spi/cadence_qspi.c > > > @@ -128,6 +128,9 @@ static int cadence_spi_set_speed(struct udevice > > > *bus, uint hz) > > > > > > cadence_spi_write_speed(bus, hz); > > > > > > + /* to ensure spi_calibration is run when SCLK frequency change */ > > > + plat->max_hz = hz; > > > + > > > > This looks like a hack, doesn't this change just subvert the condition > > below to enforce the calibration ? > > Nope. I checked through debugger where plat->max_hz is initialized once > with the value from device tree. When you use sf probe to change > frequency, the value is not reflected to plat->max_hz. Hence the > calibration is not run when frequency change. That lead to the sf probe > fail when running at 80MHz or higher. Note the priv->qspi_calibrated_hz > is updated after calibration succeed. Isn't the purpose of ->max_hz to reflect the maximum frequency of the bus? I don't think this should thus ever be updated, in particular not by the user. I also think that setting ->max_hz once from DT is the correct behavior. Best regards, Marek Vasut
On Tue, 2015-09-01 at 11:19 +0200, marex@denx.de wrote: > On Tuesday, September 01, 2015 at 11:13:38 AM, Chin Liang See wrote: > > On Tue, 2015-09-01 at 11:01 +0200, marex@denx.de wrote: > > > On Tuesday, September 01, 2015 at 10:46:47 AM, Chin Liang See wrote: > > > > Ensuring spi_calibration is run when there is a change of sclk > > > > frequency. This will ensure the qspi flash access works for high > > > > sclk frequency > > > > > > > > Signed-off-by: Chin Liang See <clsee@altera.com> > > > > Cc: Dinh Nguyen <dinguyen@opensource.altera.com> > > > > Cc: Marek Vasut <marex@denx.de> > > > > Cc: Stefan Roese <sr@denx.de> > > > > Cc: Vikas Manocha <vikas.manocha@st.com> > > > > Cc: Jagannadh Teki <jteki@openedev.com> > > > > Cc: Pavel Machek <pavel@denx.de> > > > > --- > > > > > > > > drivers/spi/cadence_qspi.c | 3 +++ > > > > 1 files changed, 3 insertions(+), 0 deletions(-) > > > > > > > > diff --git a/drivers/spi/cadence_qspi.c b/drivers/spi/cadence_qspi.c > > > > index 34a0f46..512bf2d 100644 > > > > --- a/drivers/spi/cadence_qspi.c > > > > +++ b/drivers/spi/cadence_qspi.c > > > > @@ -128,6 +128,9 @@ static int cadence_spi_set_speed(struct udevice > > > > *bus, uint hz) > > > > > > > > cadence_spi_write_speed(bus, hz); > > > > > > > > + /* to ensure spi_calibration is run when SCLK frequency change */ > > > > + plat->max_hz = hz; > > > > + > > > > > > This looks like a hack, doesn't this change just subvert the condition > > > below to enforce the calibration ? > > > > Nope. I checked through debugger where plat->max_hz is initialized once > > with the value from device tree. When you use sf probe to change > > frequency, the value is not reflected to plat->max_hz. Hence the > > calibration is not run when frequency change. That lead to the sf probe > > fail when running at 80MHz or higher. Note the priv->qspi_calibrated_hz > > is updated after calibration succeed. > > Isn't the purpose of ->max_hz to reflect the maximum frequency of the bus? > I don't think this should thus ever be updated, in particular not by the > user. I also think that setting ->max_hz once from DT is the correct behavior. > Oh, in this case, there are more bug in the code. I saw there is mismatch of the usage of this variable. As example, the line "priv->qspi_calibrated_hz = plat->max_hz;" is wrong as it would be always static. Besides that, there is no check against the requested speed not to exceed the max_hz. Let me fix this all again. Thanks Chin Liang > Best regards, > Marek Vasut
On Tuesday, September 01, 2015 at 11:30:59 AM, Chin Liang See wrote: > On Tue, 2015-09-01 at 11:19 +0200, marex@denx.de wrote: > > On Tuesday, September 01, 2015 at 11:13:38 AM, Chin Liang See wrote: > > > On Tue, 2015-09-01 at 11:01 +0200, marex@denx.de wrote: > > > > On Tuesday, September 01, 2015 at 10:46:47 AM, Chin Liang See wrote: > > > > > Ensuring spi_calibration is run when there is a change of sclk > > > > > frequency. This will ensure the qspi flash access works for high > > > > > sclk frequency > > > > > > > > > > Signed-off-by: Chin Liang See <clsee@altera.com> > > > > > Cc: Dinh Nguyen <dinguyen@opensource.altera.com> > > > > > Cc: Marek Vasut <marex@denx.de> > > > > > Cc: Stefan Roese <sr@denx.de> > > > > > Cc: Vikas Manocha <vikas.manocha@st.com> > > > > > Cc: Jagannadh Teki <jteki@openedev.com> > > > > > Cc: Pavel Machek <pavel@denx.de> > > > > > --- > > > > > > > > > > drivers/spi/cadence_qspi.c | 3 +++ > > > > > 1 files changed, 3 insertions(+), 0 deletions(-) > > > > > > > > > > diff --git a/drivers/spi/cadence_qspi.c > > > > > b/drivers/spi/cadence_qspi.c index 34a0f46..512bf2d 100644 > > > > > --- a/drivers/spi/cadence_qspi.c > > > > > +++ b/drivers/spi/cadence_qspi.c > > > > > @@ -128,6 +128,9 @@ static int cadence_spi_set_speed(struct udevice > > > > > *bus, uint hz) > > > > > > > > > > cadence_spi_write_speed(bus, hz); > > > > > > > > > > + /* to ensure spi_calibration is run when SCLK frequency change */ > > > > > + plat->max_hz = hz; > > > > > + > > > > > > > > This looks like a hack, doesn't this change just subvert the > > > > condition below to enforce the calibration ? > > > > > > Nope. I checked through debugger where plat->max_hz is initialized once > > > with the value from device tree. When you use sf probe to change > > > frequency, the value is not reflected to plat->max_hz. Hence the > > > calibration is not run when frequency change. That lead to the sf probe > > > fail when running at 80MHz or higher. Note the priv->qspi_calibrated_hz > > > is updated after calibration succeed. > > > > Isn't the purpose of ->max_hz to reflect the maximum frequency of the > > bus? I don't think this should thus ever be updated, in particular not > > by the user. I also think that setting ->max_hz once from DT is the > > correct behavior. > > Oh, in this case, there are more bug in the code. I saw there is > mismatch of the usage of this variable. As example, the line > "priv->qspi_calibrated_hz = plat->max_hz;" is wrong as it would be > always static. Besides that, there is no check against the requested > speed not to exceed the max_hz. Let me fix this all again. THanks Best regards, Marek Vasut
diff --git a/drivers/spi/cadence_qspi.c b/drivers/spi/cadence_qspi.c index 34a0f46..512bf2d 100644 --- a/drivers/spi/cadence_qspi.c +++ b/drivers/spi/cadence_qspi.c @@ -128,6 +128,9 @@ static int cadence_spi_set_speed(struct udevice *bus, uint hz) cadence_spi_write_speed(bus, hz); + /* to ensure spi_calibration is run when SCLK frequency change */ + plat->max_hz = hz; + /* Calibration required for different SCLK speed or chip select */ if (priv->qspi_calibrated_hz != plat->max_hz || priv->qspi_calibrated_cs != spi_chip_select(bus)) {
Ensuring spi_calibration is run when there is a change of sclk frequency. This will ensure the qspi flash access works for high sclk frequency Signed-off-by: Chin Liang See <clsee@altera.com> Cc: Dinh Nguyen <dinguyen@opensource.altera.com> Cc: Marek Vasut <marex@denx.de> Cc: Stefan Roese <sr@denx.de> Cc: Vikas Manocha <vikas.manocha@st.com> Cc: Jagannadh Teki <jteki@openedev.com> Cc: Pavel Machek <pavel@denx.de> --- drivers/spi/cadence_qspi.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)