Some background about this.
On original qsdk ipq8064 based firmware there was a big separation from
boot partition and user partition. With boot partition we refer to
partition used to init the router (bootloader, spm firmware and other
internal stuff) With user partition we refer to linux partition and data
partition not used to init the router.
When someone had to write to these boot partition a special mode was
needed, to switch the nand driver to this special configuration.
Upstream version of the nandc driver totally dropped this and the result
is that if someone try to read data from these partition a CRC warning
is printed and if someone try to write that (if for example someone
wants to replace the bootloader) result is a broken system as the data
is badly written.
This series comes to fix this.
A user can declare offset and size of these special partition using the
qcom,boot-pages binding.
An initial implementation of this assumed that the boot-pages started
from the start of the nand but we discover that some device have backup
of these special partition and we can have situation where we have this
partition scheme
- APPSBL (require special mode)
- APPSBLENV (doesn't require special mode)
- ART
- APPSBLBK (back of APPSBL require special mode)
- APPSBLENVBK (back of APPSBLENV doesn't require special mode)
With this configuration we need to declare sparse boot page and we can't
assume boot-pages always starts from the start of the nand.
A user can use this form to declare sparse boot pages
qcom,boot-pages = <0x0 0x0c80000 0x0c80000 0x0500000>;
The driver internally will parse this array, convert it to nand pages
and check internally on every read/write if this special configuration
should used for that page or the normal one.
The reason for all of this is that qcom FOR SOME REASON, disable ECC for
spare data only for these boot partition and we need to reflect this
special configuration to mute these warning and to permit actually
writing to these pages.
v5:
- Rename boot-pages to boot-partitions
- Add additional check to parsing function
- Rename unprotect_spare_data to codeword_fixup
- Add additional info from Manivannan
- Add patch to remove holes in qcom_nand_host struct
v4:
- Fix wrong compatible set for boot-pages (ipq8074 instead of ipq806x)
v3:
- Fix typo in Docmunetation commit desription
- Add items description for uint32-matrix
v2:
- Add fixes from Krzysztof in Documentation
Ansuel Smith (3):
mtd: nand: raw: qcom_nandc: add support for unprotected spare data
pages
dt-bindings: mtd: qcom_nandc: document qcom,boot-partitions binding
mtd: nand: raw: qcom_nandc: reorder qcom_nand_host struct
.../devicetree/bindings/mtd/qcom,nandc.yaml | 26 +++
drivers/mtd/nand/raw/qcom_nandc.c | 179 +++++++++++++++++-
2 files changed, 198 insertions(+), 7 deletions(-)
--
2.36.1
Reorder qcom_nand_host to save holes in the struct.
Signed-off-by: Ansuel Smith <[email protected]>
---
drivers/mtd/nand/raw/qcom_nandc.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/drivers/mtd/nand/raw/qcom_nandc.c b/drivers/mtd/nand/raw/qcom_nandc.c
index 06ee9a836a3b..110f839c9e51 100644
--- a/drivers/mtd/nand/raw/qcom_nandc.c
+++ b/drivers/mtd/nand/raw/qcom_nandc.c
@@ -475,11 +475,13 @@ struct qcom_nand_host {
int cs;
int cw_size;
int cw_data;
- bool use_ecc;
- bool bch_enabled;
int ecc_bytes_hw;
int spare_bytes;
int bbm_size;
+
+ bool codeword_fixup;
+ bool use_ecc;
+ bool bch_enabled;
u8 status;
int last_command;
@@ -490,7 +492,6 @@ struct qcom_nand_host {
u32 clrflashstatus;
u32 clrreadstatus;
- bool codeword_fixup;
int nr_boot_partitions;
struct qcom_nand_boot_partition *boot_partitions;
};
--
2.36.1
On Wed, Jun 08, 2022 at 02:10:30AM +0200, Ansuel Smith wrote:
> Reorder qcom_nand_host to save holes in the struct.
>
> Signed-off-by: Ansuel Smith <[email protected]>
If this patch gets moved to 2/3, you could save few changes. Also, do the same
for other structs as well.
Thanks,
Mani
> ---
> drivers/mtd/nand/raw/qcom_nandc.c | 7 ++++---
> 1 file changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/mtd/nand/raw/qcom_nandc.c b/drivers/mtd/nand/raw/qcom_nandc.c
> index 06ee9a836a3b..110f839c9e51 100644
> --- a/drivers/mtd/nand/raw/qcom_nandc.c
> +++ b/drivers/mtd/nand/raw/qcom_nandc.c
> @@ -475,11 +475,13 @@ struct qcom_nand_host {
> int cs;
> int cw_size;
> int cw_data;
> - bool use_ecc;
> - bool bch_enabled;
> int ecc_bytes_hw;
> int spare_bytes;
> int bbm_size;
> +
> + bool codeword_fixup;
> + bool use_ecc;
> + bool bch_enabled;
> u8 status;
> int last_command;
>
> @@ -490,7 +492,6 @@ struct qcom_nand_host {
> u32 clrflashstatus;
> u32 clrreadstatus;
>
> - bool codeword_fixup;
> int nr_boot_partitions;
> struct qcom_nand_boot_partition *boot_partitions;
> };
> --
> 2.36.1
>
--
மணிவண்ணன் சதாசிவம்
On Thu, Jun 09, 2022 at 12:52:40PM +0530, Manivannan Sadhasivam wrote:
> On Wed, Jun 08, 2022 at 02:10:30AM +0200, Ansuel Smith wrote:
> > Reorder qcom_nand_host to save holes in the struct.
> >
> > Signed-off-by: Ansuel Smith <[email protected]>
>
> If this patch gets moved to 2/3, you could save few changes. Also, do the same
> for other structs as well.
>
> Thanks,
> Mani
>
Since 2/3 already had lots of changes didn't want to put a struct
reorder in it since it does touch also other values. Tell me if I should
squash the 2 commit.
> > ---
> > drivers/mtd/nand/raw/qcom_nandc.c | 7 ++++---
> > 1 file changed, 4 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/mtd/nand/raw/qcom_nandc.c b/drivers/mtd/nand/raw/qcom_nandc.c
> > index 06ee9a836a3b..110f839c9e51 100644
> > --- a/drivers/mtd/nand/raw/qcom_nandc.c
> > +++ b/drivers/mtd/nand/raw/qcom_nandc.c
> > @@ -475,11 +475,13 @@ struct qcom_nand_host {
> > int cs;
> > int cw_size;
> > int cw_data;
> > - bool use_ecc;
> > - bool bch_enabled;
> > int ecc_bytes_hw;
> > int spare_bytes;
> > int bbm_size;
> > +
> > + bool codeword_fixup;
> > + bool use_ecc;
> > + bool bch_enabled;
> > u8 status;
> > int last_command;
> >
> > @@ -490,7 +492,6 @@ struct qcom_nand_host {
> > u32 clrflashstatus;
> > u32 clrreadstatus;
> >
> > - bool codeword_fixup;
> > int nr_boot_partitions;
> > struct qcom_nand_boot_partition *boot_partitions;
> > };
> > --
> > 2.36.1
> >
>
> --
> மணிவண்ணன் சதாசிவம்
--
Ansuel
On Thu, Jun 09, 2022 at 12:26:53PM +0200, Ansuel Smith wrote:
> On Thu, Jun 09, 2022 at 12:52:40PM +0530, Manivannan Sadhasivam wrote:
> > On Wed, Jun 08, 2022 at 02:10:30AM +0200, Ansuel Smith wrote:
> > > Reorder qcom_nand_host to save holes in the struct.
> > >
> > > Signed-off-by: Ansuel Smith <[email protected]>
> >
> > If this patch gets moved to 2/3, you could save few changes. Also, do the same
> > for other structs as well.
> >
> > Thanks,
> > Mani
> >
>
> Since 2/3 already had lots of changes didn't want to put a struct
> reorder in it since it does touch also other values. Tell me if I should
> squash the 2 commit.
>
Not squashing but I mean to move this patch before the driver change (1/3), so
that "codeword_fixup" can be added in the correct place.
Also move the respective Kdoc comments.
Thanks,
Mani
> > > ---
> > > drivers/mtd/nand/raw/qcom_nandc.c | 7 ++++---
> > > 1 file changed, 4 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/drivers/mtd/nand/raw/qcom_nandc.c b/drivers/mtd/nand/raw/qcom_nandc.c
> > > index 06ee9a836a3b..110f839c9e51 100644
> > > --- a/drivers/mtd/nand/raw/qcom_nandc.c
> > > +++ b/drivers/mtd/nand/raw/qcom_nandc.c
> > > @@ -475,11 +475,13 @@ struct qcom_nand_host {
> > > int cs;
> > > int cw_size;
> > > int cw_data;
> > > - bool use_ecc;
> > > - bool bch_enabled;
> > > int ecc_bytes_hw;
> > > int spare_bytes;
> > > int bbm_size;
> > > +
> > > + bool codeword_fixup;
> > > + bool use_ecc;
> > > + bool bch_enabled;
> > > u8 status;
> > > int last_command;
> > >
> > > @@ -490,7 +492,6 @@ struct qcom_nand_host {
> > > u32 clrflashstatus;
> > > u32 clrreadstatus;
> > >
> > > - bool codeword_fixup;
> > > int nr_boot_partitions;
> > > struct qcom_nand_boot_partition *boot_partitions;
> > > };
> > > --
> > > 2.36.1
> > >
> >
> > --
> > மணிவண்ணன் சதாசிவம்
>
> --
> Ansuel
--
மணிவண்ணன் சதாசிவம்