| Message ID | 20230814015700.70156-1-nathan@nathanrossi.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | soc: imx8m: Enable OCOTP clock for imx8mm before reading registers | expand |
On Mon, 14 Aug 2023 at 11:57, Nathan Rossi <nathan@nathanrossi.com> wrote: > > From: Nathan Rossi <nathan.rossi@digi.com> > > Commit 836fb30949d9 ("soc: imx8m: Enable OCOTP clock before reading the > register") added configuration to enable the OCOTP clock before > attempting to read from the associated registers. > > This same kexec issue is present with the imx8m SoCs that use the > imx8mm_soc_uid function (e.g. imx8mp). This requires the imx8mm_soc_uid > function to configure the OCOTP clock before accessing the associated > registers. This change implements the same clock enable functionality > that is present in the imx8mq_soc_revision function for the > imx8mm_soc_uid function. Gentle ping. Are there any comments or feedback on this change? Thanks, Nathan > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > --- > drivers/soc/imx/soc-imx8m.c | 10 ++++++++++ > 1 file changed, 10 insertions(+) > > diff --git a/drivers/soc/imx/soc-imx8m.c b/drivers/soc/imx/soc-imx8m.c > index 1dcd243df5..ec87d9d878 100644 > --- a/drivers/soc/imx/soc-imx8m.c > +++ b/drivers/soc/imx/soc-imx8m.c > @@ -100,6 +100,7 @@ static void __init imx8mm_soc_uid(void) > { > void __iomem *ocotp_base; > struct device_node *np; > + struct clk *clk; > u32 offset = of_machine_is_compatible("fsl,imx8mp") ? > IMX8MP_OCOTP_UID_OFFSET : 0; > > @@ -109,11 +110,20 @@ static void __init imx8mm_soc_uid(void) > > ocotp_base = of_iomap(np, 0); > WARN_ON(!ocotp_base); > + clk = of_clk_get_by_name(np, NULL); > + if (IS_ERR(clk)) { > + WARN_ON(IS_ERR(clk)); > + return; > + } > + > + clk_prepare_enable(clk); > > soc_uid = readl_relaxed(ocotp_base + OCOTP_UID_HIGH + offset); > soc_uid <<= 32; > soc_uid |= readl_relaxed(ocotp_base + OCOTP_UID_LOW + offset); > > + clk_disable_unprepare(clk); > + clk_put(clk); > iounmap(ocotp_base); > of_node_put(np); > } > --- > 2.40.1
Hi Nathan, On Sun, Aug 13, 2023 at 10:57 PM Nathan Rossi <nathan@nathanrossi.com> wrote: > > From: Nathan Rossi <nathan.rossi@digi.com> > > Commit 836fb30949d9 ("soc: imx8m: Enable OCOTP clock before reading the > register") added configuration to enable the OCOTP clock before > attempting to read from the associated registers. > > This same kexec issue is present with the imx8m SoCs that use the > imx8mm_soc_uid function (e.g. imx8mp). This requires the imx8mm_soc_uid > function to configure the OCOTP clock before accessing the associated > registers. This change implements the same clock enable functionality > that is present in the imx8mq_soc_revision function for the > imx8mm_soc_uid function. > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> Thanks for the fix: Reviewed-by: Fabio Estevam <festevam@gmail.com> Should it also have a Fixes tag?
On Mon, 18 Sept 2023 at 13:26, Fabio Estevam <festevam@gmail.com> wrote: > > Hi Nathan, > > On Sun, Aug 13, 2023 at 10:57 PM Nathan Rossi <nathan@nathanrossi.com> wrote: > > > > From: Nathan Rossi <nathan.rossi@digi.com> > > > > Commit 836fb30949d9 ("soc: imx8m: Enable OCOTP clock before reading the > > register") added configuration to enable the OCOTP clock before > > attempting to read from the associated registers. > > > > This same kexec issue is present with the imx8m SoCs that use the > > imx8mm_soc_uid function (e.g. imx8mp). This requires the imx8mm_soc_uid > > function to configure the OCOTP clock before accessing the associated > > registers. This change implements the same clock enable functionality > > that is present in the imx8mq_soc_revision function for the > > imx8mm_soc_uid function. > > > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > > Thanks for the fix: > > Reviewed-by: Fabio Estevam <festevam@gmail.com> > > Should it also have a Fixes tag? Potentially, though none of the previous commits (or their referenced commits) attempted to resolve the issue for the imx8mm_soc_uid function so I was not sure if a fixes would apply in this case or to which commit? Though with that said this change would only need to be applied to kernels that have applied the referenced commit, so that is probably the ideal target for the fixes. I can send a v2 with the Fixes included? Otherwise I have included the specific fixes commit information below. Fixes: 836fb30949d9 ("soc: imx8m: Enable OCOTP clock before reading the register") Thanks, Nathan
On Mon, Aug 14, 2023 at 01:57:00AM +0000, Nathan Rossi wrote: > From: Nathan Rossi <nathan.rossi@digi.com> > > Commit 836fb30949d9 ("soc: imx8m: Enable OCOTP clock before reading the > register") added configuration to enable the OCOTP clock before > attempting to read from the associated registers. > > This same kexec issue is present with the imx8m SoCs that use the > imx8mm_soc_uid function (e.g. imx8mp). This requires the imx8mm_soc_uid > function to configure the OCOTP clock before accessing the associated > registers. This change implements the same clock enable functionality > that is present in the imx8mq_soc_revision function for the > imx8mm_soc_uid function. > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> Applied, thanks!
diff --git a/drivers/soc/imx/soc-imx8m.c b/drivers/soc/imx/soc-imx8m.c index 1dcd243df5..ec87d9d878 100644 --- a/drivers/soc/imx/soc-imx8m.c +++ b/drivers/soc/imx/soc-imx8m.c @@ -100,6 +100,7 @@ static void __init imx8mm_soc_uid(void) { void __iomem *ocotp_base; struct device_node *np; + struct clk *clk; u32 offset = of_machine_is_compatible("fsl,imx8mp") ? IMX8MP_OCOTP_UID_OFFSET : 0; @@ -109,11 +110,20 @@ static void __init imx8mm_soc_uid(void) ocotp_base = of_iomap(np, 0); WARN_ON(!ocotp_base); + clk = of_clk_get_by_name(np, NULL); + if (IS_ERR(clk)) { + WARN_ON(IS_ERR(clk)); + return; + } + + clk_prepare_enable(clk); soc_uid = readl_relaxed(ocotp_base + OCOTP_UID_HIGH + offset); soc_uid <<= 32; soc_uid |= readl_relaxed(ocotp_base + OCOTP_UID_LOW + offset); + clk_disable_unprepare(clk); + clk_put(clk); iounmap(ocotp_base); of_node_put(np); }