A64 SID controller has some issues when readind data, To exampine the
problem I've done the following steps.
When reading the whole nvmem memory in one chunk the returned bytes
are valid:
dd if=/sys/bus/nvmem/devices/sunxi-sid0/nvmem 2>/dev/null | hexdump -C
00000000 ba 00 c0 92 20 46 10 84 00 45 34 50 0e 04 26 48 |.... F...E4P..&H|
00000010 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00000030 00 00 00 00 87 07 8d 07 8e 07 00 00 00 00 00 00 |................|
00000040 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00000100
When bs is set to 4 bytes the data is no longer valid:
dd if=/sys/bus/nvmem/devices/sunxi-sid0/nvmem bs=4 2>/dev/null | hexdump -C
00000000 ba 00 c0 92 20 46 10 84 00 45 34 50 0e 04 26 48 |.... F...E4P..&H|
00000010 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00000030 00 00 00 00 87 00 00 00 8e 00 00 00 00 00 00 00 |................|
00000040 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00000100
You can see that only the data at 0x34 and 0x38 is corrupted. It appears
that A64 needs minimun 8 bytes block size;
dd if=/sys/bus/nvmem/devices/sunxi-sid0/nvmem bs=8 2>/dev/null | hexdump -C
00000000 ba 00 c0 92 20 46 10 84 00 45 34 50 0e 04 26 48 |.... F...E4P..&H|
00000010 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00000030 00 00 00 00 87 07 8d 07 8e 07 00 00 00 00 00 00 |................|
00000040 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00000100
In the driver stride is set to 4 and word_size to 1. When you're using
nvmem as thermal calibration data in the dts you write something like this:
sid: eeprom@1c14000 {
compatible = "allwinner,sun50i-a64-sid";
.....
thermal_calibration: calib@34 {
reg = <0x34 0x08>;
};
};
This will return incorrect data.
One way to fix this is to set stride/word_size to 8, but this will be
inconvenient for the dts.
Other is to enable reading data via register access. After the fix:
dd if=/sys/bus/nvmem/devices/sunxi-sid0/nvmem bs=4 2>/dev/null | hexdump -C
00000000 ba 00 c0 92 20 46 10 84 00 45 34 50 0e 04 26 48 |.... F...E4P..&H|
00000010 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00000030 00 00 00 00 87 07 8d 07 8e 07 00 00 00 00 00 00 |................|
00000040 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00000100
Stefan Mavrodiev (1):
nvmem: sunxi_sid: fix A64 SID controller support
drivers/nvmem/sunxi_sid.c | 1 +
1 file changed, 1 insertion(+)
--
2.17.1
Like in H3, A64 SID controller doesn't return correct data
when using direct access. It appears that on A64, SID needs
8 bytes of word_size.
Workaround is to enable read by registers.
Signed-off-by: Stefan Mavrodiev <[email protected]>
---
drivers/nvmem/sunxi_sid.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/nvmem/sunxi_sid.c b/drivers/nvmem/sunxi_sid.c
index a079a80ddf2c..e26ef1bbf198 100644
--- a/drivers/nvmem/sunxi_sid.c
+++ b/drivers/nvmem/sunxi_sid.c
@@ -186,6 +186,7 @@ static const struct sunxi_sid_cfg sun8i_h3_cfg = {
static const struct sunxi_sid_cfg sun50i_a64_cfg = {
.value_offset = 0x200,
.size = 0x100,
+ .need_register_readout = true,
};
static const struct sunxi_sid_cfg sun50i_h6_cfg = {
--
2.17.1
On Wed, Jul 31, 2019 at 3:15 PM Stefan Mavrodiev <[email protected]> wrote:
>
> Like in H3, A64 SID controller doesn't return correct data
> when using direct access. It appears that on A64, SID needs
> 8 bytes of word_size.
>
> Workaround is to enable read by registers.
>
> Signed-off-by: Stefan Mavrodiev <[email protected]>
Acked-by: Chen-Yu Tsai <[email protected]>
And for single patches, you don't need to write a separate cover letter.
Just put whatever you need to add after the "---" separator.
> ---
> drivers/nvmem/sunxi_sid.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/nvmem/sunxi_sid.c b/drivers/nvmem/sunxi_sid.c
> index a079a80ddf2c..e26ef1bbf198 100644
> --- a/drivers/nvmem/sunxi_sid.c
> +++ b/drivers/nvmem/sunxi_sid.c
> @@ -186,6 +186,7 @@ static const struct sunxi_sid_cfg sun8i_h3_cfg = {
> static const struct sunxi_sid_cfg sun50i_a64_cfg = {
> .value_offset = 0x200,
> .size = 0x100,
> + .need_register_readout = true,
> };
>
> static const struct sunxi_sid_cfg sun50i_h6_cfg = {
> --
> 2.17.1
>
On Wed, Jul 31, 2019 at 1:43 AM Chen-Yu Tsai <[email protected]> wrote:
>
> On Wed, Jul 31, 2019 at 3:15 PM Stefan Mavrodiev <[email protected]> wrote:
> >
> > Like in H3, A64 SID controller doesn't return correct data
> > when using direct access. It appears that on A64, SID needs
> > 8 bytes of word_size.
> >
> > Workaround is to enable read by registers.
I came up with identical patch while adding A64 support into
sun8i-thermal driver, so:
> >
> > Signed-off-by: Stefan Mavrodiev <[email protected]>
>
> Acked-by: Chen-Yu Tsai <[email protected]>
Tested-by: Vasily Khoruzhick <[email protected]>
>
> And for single patches, you don't need to write a separate cover letter.
> Just put whatever you need to add after the "---" separator.
>
> > ---
> > drivers/nvmem/sunxi_sid.c | 1 +
> > 1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/nvmem/sunxi_sid.c b/drivers/nvmem/sunxi_sid.c
> > index a079a80ddf2c..e26ef1bbf198 100644
> > --- a/drivers/nvmem/sunxi_sid.c
> > +++ b/drivers/nvmem/sunxi_sid.c
> > @@ -186,6 +186,7 @@ static const struct sunxi_sid_cfg sun8i_h3_cfg = {
> > static const struct sunxi_sid_cfg sun50i_a64_cfg = {
> > .value_offset = 0x200,
> > .size = 0x100,
> > + .need_register_readout = true,
> > };
> >
> > static const struct sunxi_sid_cfg sun50i_h6_cfg = {
> > --
> > 2.17.1
> >
>
> --
> You received this message because you are subscribed to the Google Groups "linux-sunxi" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> To view this discussion on the web, visit https://groups.google.com/d/msgid/linux-sunxi/CAGb2v64tzMypnB5Ho2A-gWPk2yYsmH9tNn%2BOKfb51c%2Bd6pK%3Dkw%40mail.gmail.com.
On 31/07/2019 08:14, Stefan Mavrodiev wrote:
> Like in H3, A64 SID controller doesn't return correct data
> when using direct access. It appears that on A64, SID needs
> 8 bytes of word_size.
>
> Workaround is to enable read by registers.
>
> Signed-off-by: Stefan Mavrodiev <[email protected]>
Applied Thanks,
srini