From: Ilpo Järvinen <[email protected]>
The size of the staging area in FLASH for FPGA updates is dependent on the
size of the FPGA. Currently, the staging size is defined as a constant.
Larger FPGAs are coming soon and it will soon be necessary to support
different sizes for the staging area. Add a new staging_size member to the
csr_map structure to support a variable staging size.
The secure update driver does a sanity-check of the image size in
comparison to the size of the staging area in FLASH. Change the
staging size reference to a variable instead of a constant in order
to more readily support future, larger FPGAs.
Co-developed-by: Russ Weight <[email protected]>
Signed-off-by: Russ Weight <[email protected]>
Signed-off-by: Ilpo Järvinen <[email protected]>
Signed-off-by: Peter Colberg <[email protected]>
---
drivers/fpga/intel-m10-bmc-sec-update.c | 3 ++-
drivers/mfd/intel-m10-bmc-pmci.c | 1 +
drivers/mfd/intel-m10-bmc-spi.c | 1 +
include/linux/mfd/intel-m10-bmc.h | 1 +
4 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
index 89851b133709..7ac9f9f5af12 100644
--- a/drivers/fpga/intel-m10-bmc-sec-update.c
+++ b/drivers/fpga/intel-m10-bmc-sec-update.c
@@ -529,11 +529,12 @@ static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
const u8 *data, u32 size)
{
struct m10bmc_sec *sec = fwl->dd_handle;
+ const struct m10bmc_csr_map *csr_map = sec->m10bmc->info->csr_map;
u32 ret;
sec->cancel_request = false;
- if (!size || size > M10BMC_STAGING_SIZE)
+ if (!size || size > csr_map->staging_size)
return FW_UPLOAD_ERR_INVALID_SIZE;
if (sec->m10bmc->flash_bulk_ops)
diff --git a/drivers/mfd/intel-m10-bmc-pmci.c b/drivers/mfd/intel-m10-bmc-pmci.c
index 0392ef8b57d8..698c5933938b 100644
--- a/drivers/mfd/intel-m10-bmc-pmci.c
+++ b/drivers/mfd/intel-m10-bmc-pmci.c
@@ -370,6 +370,7 @@ static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
.pr_reh_addr = M10BMC_N6000_PR_REH_ADDR,
.pr_magic = M10BMC_N6000_PR_PROG_MAGIC,
.rsu_update_counter = M10BMC_N6000_STAGING_FLASH_COUNT,
+ .staging_size = M10BMC_STAGING_SIZE,
};
static const struct intel_m10bmc_platform_info m10bmc_pmci_n6000 = {
diff --git a/drivers/mfd/intel-m10-bmc-spi.c b/drivers/mfd/intel-m10-bmc-spi.c
index cbeb7de9e041..d64d28199df6 100644
--- a/drivers/mfd/intel-m10-bmc-spi.c
+++ b/drivers/mfd/intel-m10-bmc-spi.c
@@ -109,6 +109,7 @@ static const struct m10bmc_csr_map m10bmc_n3000_csr_map = {
.pr_reh_addr = M10BMC_N3000_PR_REH_ADDR,
.pr_magic = M10BMC_N3000_PR_PROG_MAGIC,
.rsu_update_counter = M10BMC_N3000_STAGING_FLASH_COUNT,
+ .staging_size = M10BMC_STAGING_SIZE,
};
static struct mfd_cell m10bmc_d5005_subdevs[] = {
diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
index ee66c9751003..988f1cd90032 100644
--- a/include/linux/mfd/intel-m10-bmc.h
+++ b/include/linux/mfd/intel-m10-bmc.h
@@ -205,6 +205,7 @@ struct m10bmc_csr_map {
unsigned int pr_reh_addr;
unsigned int pr_magic;
unsigned int rsu_update_counter;
+ unsigned int staging_size;
};
/**
--
2.44.0
On Thu, Mar 28, 2024 at 07:35:59PM -0400, Peter Colberg wrote:
> From: Ilpo Järvinen <[email protected]>
>
> The size of the staging area in FLASH for FPGA updates is dependent on the
> size of the FPGA. Currently, the staging size is defined as a constant.
> Larger FPGAs are coming soon and it will soon be necessary to support
Soon? When? You cannot add some feature without a user case. If you do
have a use case, put the patch in the same patchset.
Thanks,
Yilun
> different sizes for the staging area. Add a new staging_size member to the
> csr_map structure to support a variable staging size.
>
> The secure update driver does a sanity-check of the image size in
> comparison to the size of the staging area in FLASH. Change the
> staging size reference to a variable instead of a constant in order
> to more readily support future, larger FPGAs.
>
> Co-developed-by: Russ Weight <[email protected]>
> Signed-off-by: Russ Weight <[email protected]>
> Signed-off-by: Ilpo Järvinen <[email protected]>
> Signed-off-by: Peter Colberg <[email protected]>
> ---
> drivers/fpga/intel-m10-bmc-sec-update.c | 3 ++-
> drivers/mfd/intel-m10-bmc-pmci.c | 1 +
> drivers/mfd/intel-m10-bmc-spi.c | 1 +
> include/linux/mfd/intel-m10-bmc.h | 1 +
> 4 files changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
> index 89851b133709..7ac9f9f5af12 100644
> --- a/drivers/fpga/intel-m10-bmc-sec-update.c
> +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
> @@ -529,11 +529,12 @@ static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
> const u8 *data, u32 size)
> {
> struct m10bmc_sec *sec = fwl->dd_handle;
> + const struct m10bmc_csr_map *csr_map = sec->m10bmc->info->csr_map;
> u32 ret;
>
> sec->cancel_request = false;
>
> - if (!size || size > M10BMC_STAGING_SIZE)
> + if (!size || size > csr_map->staging_size)
> return FW_UPLOAD_ERR_INVALID_SIZE;
>
> if (sec->m10bmc->flash_bulk_ops)
> diff --git a/drivers/mfd/intel-m10-bmc-pmci.c b/drivers/mfd/intel-m10-bmc-pmci.c
> index 0392ef8b57d8..698c5933938b 100644
> --- a/drivers/mfd/intel-m10-bmc-pmci.c
> +++ b/drivers/mfd/intel-m10-bmc-pmci.c
> @@ -370,6 +370,7 @@ static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
> .pr_reh_addr = M10BMC_N6000_PR_REH_ADDR,
> .pr_magic = M10BMC_N6000_PR_PROG_MAGIC,
> .rsu_update_counter = M10BMC_N6000_STAGING_FLASH_COUNT,
> + .staging_size = M10BMC_STAGING_SIZE,
> };
>
> static const struct intel_m10bmc_platform_info m10bmc_pmci_n6000 = {
> diff --git a/drivers/mfd/intel-m10-bmc-spi.c b/drivers/mfd/intel-m10-bmc-spi.c
> index cbeb7de9e041..d64d28199df6 100644
> --- a/drivers/mfd/intel-m10-bmc-spi.c
> +++ b/drivers/mfd/intel-m10-bmc-spi.c
> @@ -109,6 +109,7 @@ static const struct m10bmc_csr_map m10bmc_n3000_csr_map = {
> .pr_reh_addr = M10BMC_N3000_PR_REH_ADDR,
> .pr_magic = M10BMC_N3000_PR_PROG_MAGIC,
> .rsu_update_counter = M10BMC_N3000_STAGING_FLASH_COUNT,
> + .staging_size = M10BMC_STAGING_SIZE,
> };
>
> static struct mfd_cell m10bmc_d5005_subdevs[] = {
> diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
> index ee66c9751003..988f1cd90032 100644
> --- a/include/linux/mfd/intel-m10-bmc.h
> +++ b/include/linux/mfd/intel-m10-bmc.h
> @@ -205,6 +205,7 @@ struct m10bmc_csr_map {
> unsigned int pr_reh_addr;
> unsigned int pr_magic;
> unsigned int rsu_update_counter;
> + unsigned int staging_size;
> };
>
> /**
> --
> 2.44.0
>
>
On Mon, Apr 01, 2024 at 05:46:29PM +0800, Xu Yilun wrote:
> On Thu, Mar 28, 2024 at 07:35:59PM -0400, Peter Colberg wrote:
> > From: Ilpo J?rvinen <[email protected]>
> >
> > The size of the staging area in FLASH for FPGA updates is dependent on the
> > size of the FPGA. Currently, the staging size is defined as a constant.
> > Larger FPGAs are coming soon and it will soon be necessary to support
>
> Soon? When? You cannot add some feature without a user case. If you do
> have a use case, put the patch in the same patchset.
There may never be an up-streamed use-case. This is a very small
change intended to make it easier for a third-party vendor to
build a card that requires a larger staging area in FLASH. They
would have to add a new "struct m10bmc_csr_map", but they
wouldn't have to refactor this code as part of the change
This change does not introduce an unused function or variable.
It is more of a clean-up, making the code more flexible.
Can it not be taken as is?
- Russ
>
> Thanks,
> Yilun
>
> > different sizes for the staging area. Add a new staging_size member to the
> > csr_map structure to support a variable staging size.
> >
> > The secure update driver does a sanity-check of the image size in
> > comparison to the size of the staging area in FLASH. Change the
> > staging size reference to a variable instead of a constant in order
> > to more readily support future, larger FPGAs.
> >
> > Co-developed-by: Russ Weight <[email protected]>
> > Signed-off-by: Russ Weight <[email protected]>
> > Signed-off-by: Ilpo J?rvinen <[email protected]>
> > Signed-off-by: Peter Colberg <[email protected]>
> > ---
> > drivers/fpga/intel-m10-bmc-sec-update.c | 3 ++-
> > drivers/mfd/intel-m10-bmc-pmci.c | 1 +
> > drivers/mfd/intel-m10-bmc-spi.c | 1 +
> > include/linux/mfd/intel-m10-bmc.h | 1 +
> > 4 files changed, 5 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
> > index 89851b133709..7ac9f9f5af12 100644
> > --- a/drivers/fpga/intel-m10-bmc-sec-update.c
> > +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
> > @@ -529,11 +529,12 @@ static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
> > const u8 *data, u32 size)
> > {
> > struct m10bmc_sec *sec = fwl->dd_handle;
> > + const struct m10bmc_csr_map *csr_map = sec->m10bmc->info->csr_map;
> > u32 ret;
> >
> > sec->cancel_request = false;
> >
> > - if (!size || size > M10BMC_STAGING_SIZE)
> > + if (!size || size > csr_map->staging_size)
> > return FW_UPLOAD_ERR_INVALID_SIZE;
> >
> > if (sec->m10bmc->flash_bulk_ops)
> > diff --git a/drivers/mfd/intel-m10-bmc-pmci.c b/drivers/mfd/intel-m10-bmc-pmci.c
> > index 0392ef8b57d8..698c5933938b 100644
> > --- a/drivers/mfd/intel-m10-bmc-pmci.c
> > +++ b/drivers/mfd/intel-m10-bmc-pmci.c
> > @@ -370,6 +370,7 @@ static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
> > .pr_reh_addr = M10BMC_N6000_PR_REH_ADDR,
> > .pr_magic = M10BMC_N6000_PR_PROG_MAGIC,
> > .rsu_update_counter = M10BMC_N6000_STAGING_FLASH_COUNT,
> > + .staging_size = M10BMC_STAGING_SIZE,
> > };
> >
> > static const struct intel_m10bmc_platform_info m10bmc_pmci_n6000 = {
> > diff --git a/drivers/mfd/intel-m10-bmc-spi.c b/drivers/mfd/intel-m10-bmc-spi.c
> > index cbeb7de9e041..d64d28199df6 100644
> > --- a/drivers/mfd/intel-m10-bmc-spi.c
> > +++ b/drivers/mfd/intel-m10-bmc-spi.c
> > @@ -109,6 +109,7 @@ static const struct m10bmc_csr_map m10bmc_n3000_csr_map = {
> > .pr_reh_addr = M10BMC_N3000_PR_REH_ADDR,
> > .pr_magic = M10BMC_N3000_PR_PROG_MAGIC,
> > .rsu_update_counter = M10BMC_N3000_STAGING_FLASH_COUNT,
> > + .staging_size = M10BMC_STAGING_SIZE,
> > };
> >
> > static struct mfd_cell m10bmc_d5005_subdevs[] = {
> > diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
> > index ee66c9751003..988f1cd90032 100644
> > --- a/include/linux/mfd/intel-m10-bmc.h
> > +++ b/include/linux/mfd/intel-m10-bmc.h
> > @@ -205,6 +205,7 @@ struct m10bmc_csr_map {
> > unsigned int pr_reh_addr;
> > unsigned int pr_magic;
> > unsigned int rsu_update_counter;
> > + unsigned int staging_size;
> > };
> >
> > /**
> > --
> > 2.44.0
> >
> >
On Mon, Apr 01, 2024 at 10:09:05AM -0700, Russ Weight wrote:
> On Mon, Apr 01, 2024 at 05:46:29PM +0800, Xu Yilun wrote:
> > On Thu, Mar 28, 2024 at 07:35:59PM -0400, Peter Colberg wrote:
> > > From: Ilpo J?rvinen <[email protected]>
> > >
> > > The size of the staging area in FLASH for FPGA updates is dependent on the
> > > size of the FPGA. Currently, the staging size is defined as a constant.
> > > Larger FPGAs are coming soon and it will soon be necessary to support
> >
> > Soon? When? You cannot add some feature without a user case. If you do
> > have a use case, put the patch in the same patchset.
>
> There may never be an up-streamed use-case. This is a very small
> change intended to make it easier for a third-party vendor to
> build a card that requires a larger staging area in FLASH. They
> would have to add a new "struct m10bmc_csr_map", but they
> wouldn't have to refactor this code as part of the change
>
> This change does not introduce an unused function or variable.
> It is more of a clean-up, making the code more flexible.
>
> Can it not be taken as is?
Would it be acceptable to just change the commit message to something
like:
Do not hardwire the staging size in the secure update driver. Move
the staging size to the m10bmc_csr_map structure to make the size
assignment more flexible.
>
> - Russ
>
> >
> > Thanks,
> > Yilun
> >
> > > different sizes for the staging area. Add a new staging_size member to the
> > > csr_map structure to support a variable staging size.
> > >
> > > The secure update driver does a sanity-check of the image size in
> > > comparison to the size of the staging area in FLASH. Change the
> > > staging size reference to a variable instead of a constant in order
> > > to more readily support future, larger FPGAs.
> > >
> > > Co-developed-by: Russ Weight <[email protected]>
> > > Signed-off-by: Russ Weight <[email protected]>
> > > Signed-off-by: Ilpo J?rvinen <[email protected]>
> > > Signed-off-by: Peter Colberg <[email protected]>
> > > ---
> > > drivers/fpga/intel-m10-bmc-sec-update.c | 3 ++-
> > > drivers/mfd/intel-m10-bmc-pmci.c | 1 +
> > > drivers/mfd/intel-m10-bmc-spi.c | 1 +
> > > include/linux/mfd/intel-m10-bmc.h | 1 +
> > > 4 files changed, 5 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
> > > index 89851b133709..7ac9f9f5af12 100644
> > > --- a/drivers/fpga/intel-m10-bmc-sec-update.c
> > > +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
> > > @@ -529,11 +529,12 @@ static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
> > > const u8 *data, u32 size)
> > > {
> > > struct m10bmc_sec *sec = fwl->dd_handle;
> > > + const struct m10bmc_csr_map *csr_map = sec->m10bmc->info->csr_map;
> > > u32 ret;
> > >
> > > sec->cancel_request = false;
> > >
> > > - if (!size || size > M10BMC_STAGING_SIZE)
> > > + if (!size || size > csr_map->staging_size)
> > > return FW_UPLOAD_ERR_INVALID_SIZE;
> > >
> > > if (sec->m10bmc->flash_bulk_ops)
> > > diff --git a/drivers/mfd/intel-m10-bmc-pmci.c b/drivers/mfd/intel-m10-bmc-pmci.c
> > > index 0392ef8b57d8..698c5933938b 100644
> > > --- a/drivers/mfd/intel-m10-bmc-pmci.c
> > > +++ b/drivers/mfd/intel-m10-bmc-pmci.c
> > > @@ -370,6 +370,7 @@ static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
> > > .pr_reh_addr = M10BMC_N6000_PR_REH_ADDR,
> > > .pr_magic = M10BMC_N6000_PR_PROG_MAGIC,
> > > .rsu_update_counter = M10BMC_N6000_STAGING_FLASH_COUNT,
> > > + .staging_size = M10BMC_STAGING_SIZE,
> > > };
> > >
> > > static const struct intel_m10bmc_platform_info m10bmc_pmci_n6000 = {
> > > diff --git a/drivers/mfd/intel-m10-bmc-spi.c b/drivers/mfd/intel-m10-bmc-spi.c
> > > index cbeb7de9e041..d64d28199df6 100644
> > > --- a/drivers/mfd/intel-m10-bmc-spi.c
> > > +++ b/drivers/mfd/intel-m10-bmc-spi.c
> > > @@ -109,6 +109,7 @@ static const struct m10bmc_csr_map m10bmc_n3000_csr_map = {
> > > .pr_reh_addr = M10BMC_N3000_PR_REH_ADDR,
> > > .pr_magic = M10BMC_N3000_PR_PROG_MAGIC,
> > > .rsu_update_counter = M10BMC_N3000_STAGING_FLASH_COUNT,
> > > + .staging_size = M10BMC_STAGING_SIZE,
> > > };
> > >
> > > static struct mfd_cell m10bmc_d5005_subdevs[] = {
> > > diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
> > > index ee66c9751003..988f1cd90032 100644
> > > --- a/include/linux/mfd/intel-m10-bmc.h
> > > +++ b/include/linux/mfd/intel-m10-bmc.h
> > > @@ -205,6 +205,7 @@ struct m10bmc_csr_map {
> > > unsigned int pr_reh_addr;
> > > unsigned int pr_magic;
> > > unsigned int rsu_update_counter;
> > > + unsigned int staging_size;
> > > };
> > >
> > > /**
> > > --
> > > 2.44.0
> > >
> > >
On Mon, Apr 01, 2024 at 10:19:47AM -0700, Russ Weight wrote:
>
> On Mon, Apr 01, 2024 at 10:09:05AM -0700, Russ Weight wrote:
> > On Mon, Apr 01, 2024 at 05:46:29PM +0800, Xu Yilun wrote:
> > > On Thu, Mar 28, 2024 at 07:35:59PM -0400, Peter Colberg wrote:
> > > > From: Ilpo Järvinen <[email protected]>
> > > >
> > > > The size of the staging area in FLASH for FPGA updates is dependent on the
> > > > size of the FPGA. Currently, the staging size is defined as a constant.
> > > > Larger FPGAs are coming soon and it will soon be necessary to support
> > >
> > > Soon? When? You cannot add some feature without a user case. If you do
> > > have a use case, put the patch in the same patchset.
> >
> > There may never be an up-streamed use-case. This is a very small
> > change intended to make it easier for a third-party vendor to
> > build a card that requires a larger staging area in FLASH. They
> > would have to add a new "struct m10bmc_csr_map", but they
> > wouldn't have to refactor this code as part of the change
I'm OK with this description.
Peter, is that what you mean? Or you do have a board type to follow, in
which case you need to submit the new board type as well.
> >
> > This change does not introduce an unused function or variable.
> > It is more of a clean-up, making the code more flexible.
> >
> > Can it not be taken as is?
>
> Would it be acceptable to just change the commit message to something
> like:
>
> Do not hardwire the staging size in the secure update driver. Move
> the staging size to the m10bmc_csr_map structure to make the size
> assignment more flexible.
That would be much better.
Thanks,
Yilun
On Tue, 2024-04-02 at 14:52 +0800, Xu Yilun wrote:
> On Mon, Apr 01, 2024 at 10:19:47AM -0700, Russ Weight wrote:
> >
> > On Mon, Apr 01, 2024 at 10:09:05AM -0700, Russ Weight wrote:
> > > On Mon, Apr 01, 2024 at 05:46:29PM +0800, Xu Yilun wrote:
> > > > On Thu, Mar 28, 2024 at 07:35:59PM -0400, Peter Colberg wrote:
> > > > > From: Ilpo Järvinen <[email protected]>
> > > > >
> > > > > The size of the staging area in FLASH for FPGA updates is dependent on the
> > > > > size of the FPGA. Currently, the staging size is defined as a constant.
> > > > > Larger FPGAs are coming soon and it will soon be necessary to support
> > > >
> > > > Soon? When? You cannot add some feature without a user case. If you do
> > > > have a use case, put the patch in the same patchset.
> > >
> > > There may never be an up-streamed use-case. This is a very small
> > > change intended to make it easier for a third-party vendor to
> > > build a card that requires a larger staging area in FLASH. They
> > > would have to add a new "struct m10bmc_csr_map", but they
> > > wouldn't have to refactor this code as part of the change
>
> I'm OK with this description.
>
> Peter, is that what you mean?
Yes.
> Or you do have a board type to follow, in
> which case you need to submit the new board type as well.
>
> > >
> > > This change does not introduce an unused function or variable.
> > > It is more of a clean-up, making the code more flexible.
> > >
> > > Can it not be taken as is?
> >
> > Would it be acceptable to just change the commit message to something
> > like:
> >
> > Do not hardwire the staging size in the secure update driver. Move
> > the staging size to the m10bmc_csr_map structure to make the size
> > assignment more flexible.
>
> That would be much better.
Thanks Russ, thanks Yilun, I will send a revised patch.
Peter
>
> Thanks,
> Yilun