Patchwork [U-Boot,V2,3/4] env_mmc: allow negative CONFIG_ENV_OFFSET

login
register
mail settings
Submitter Stephen Warren
Date May 22, 2013, 7:21 p.m.
Message ID <1369250471-9851-3-git-send-email-swarren@wwwdotorg.org>
Download mbox | patch
Permalink /patch/245699/
State Superseded
Headers show

Comments

Stephen Warren - May 22, 2013, 7:21 p.m.
From: Stephen Warren <swarren@nvidia.com>

A negative value of CONFIG_ENV_OFFSET is treated as a backwards offset
from the end of the eMMC device/partition, rather than a forwards offset
from the start.

This is useful when a single board may be stuffed with different eMMC
devices, each of which has a different capacity, and you always want the
environment to be stored at the very end of the device (or eMMC boot
partition for example).

One example of this case is NVIDIA's Ventana reference board.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
---
v2: Also update README to describe the change.
---
 README           |   11 +++++++++++
 common/env_mmc.c |   12 ++++++++++--
 2 files changed, 21 insertions(+), 2 deletions(-)
Peter Korsgaard - May 23, 2013, 7:52 a.m.
>>>>> "S" == Stephen Warren <swarren@wwwdotorg.org> writes:

 S> From: Stephen Warren <swarren@nvidia.com>
 S> A negative value of CONFIG_ENV_OFFSET is treated as a backwards offset
 S> from the end of the eMMC device/partition, rather than a forwards offset
 S> from the start.

 S> This is useful when a single board may be stuffed with different eMMC
 S> devices, each of which has a different capacity, and you always want the
 S> environment to be stored at the very end of the device (or eMMC boot
 S> partition for example).

 S> One example of this case is NVIDIA's Ventana reference board.

 S> Signed-off-by: Stephen Warren <swarren@nvidia.com>
 S> ---
 S> v2: Also update README to describe the change.
 S> ---
 S>  README           |   11 +++++++++++
 S>  common/env_mmc.c |   12 ++++++++++--
 S>  2 files changed, 21 insertions(+), 2 deletions(-)

 S> diff --git a/README b/README
 S> index b9936ca..4335730 100644
 S> --- a/README
 S> +++ b/README
 S> @@ -3627,6 +3627,14 @@ but it can not erase, write this NOR flash by SRIO or PCIE interface.
 S>  	  These two #defines specify the offset and size of the environment
 S>  	  area within the specified MMC device.
 
 S> +	  If offset is positive (the usual case), it is treated as relative to
 S> +	  the start of the MMC partition. If offset is negative, it is treated
 S> +	  as relative to the end of the MMC partition. This can be useful if
 S> +	  your board may be fitted with different MMC devices, which have
 S> +	  different sizes for the MMC partitions, and you always want the
 S> +	  environment placed at the very end of the partition, to leave the
 S> +	  maximum possible space before it, to store other data.
 S> +
 S>  	  These two values must be aligned to an MMC sector boundary.
 
 S>  	- CONFIG_ENV_OFFSET_REDUND (optional):
 S> @@ -3636,6 +3644,9 @@ but it can not erase, write this NOR flash by SRIO or PCIE interface.
 S>  	  valid backup copy in case the other copy is corrupted, e.g. due
 S>  	  to a power failure during a "saveenv" operation.
 
 S> +	  This value may also be positive or negative; this is handled in the
 S> +	  same way as CONFIG_ENV_OFFSET.
 S> +
 S>  	  This value must also be aligned to an MMC sector boundary.
 
 S>  	- CONFIG_ENV_SIZE_REDUND (optional):
 S> diff --git a/common/env_mmc.c b/common/env_mmc.c
 S> index 9ca098f..5d3a769 100644
 S> --- a/common/env_mmc.c
 S> +++ b/common/env_mmc.c
 S> @@ -53,11 +53,19 @@ DECLARE_GLOBAL_DATA_PTR;
 
 S>  __weak int mmc_get_env_addr(struct mmc *mmc, int copy, u32 *env_addr)
 S>  {
 S> -	*env_addr = CONFIG_ENV_OFFSET;
 S> +	s64 offset;
 S> +
 S> +	offset = CONFIG_ENV_OFFSET;
 S>  #ifdef CONFIG_ENV_OFFSET_REDUND
 S>  	if (copy)
 S> -		*env_addr = CONFIG_ENV_OFFSET_REDUND;
 S> +		offset = CONFIG_ENV_OFFSET_REDUND;
 S>  #endif
 S> +
 S> +	if (offset < 0)
 S> +		offset += mmc->capacity;
 S> +
 S> +	*env_addr = offset;
 S> +

This is quite a mix of data types (u32/s64/u64). Not directly related to
this patch, but it would imho be nicer to let env_mmc work directly with
block numbers instead of bytes, that way stuff would also work with >4GB
MMCs.

Patch

diff --git a/README b/README
index b9936ca..4335730 100644
--- a/README
+++ b/README
@@ -3627,6 +3627,14 @@  but it can not erase, write this NOR flash by SRIO or PCIE interface.
 	  These two #defines specify the offset and size of the environment
 	  area within the specified MMC device.
 
+	  If offset is positive (the usual case), it is treated as relative to
+	  the start of the MMC partition. If offset is negative, it is treated
+	  as relative to the end of the MMC partition. This can be useful if
+	  your board may be fitted with different MMC devices, which have
+	  different sizes for the MMC partitions, and you always want the
+	  environment placed at the very end of the partition, to leave the
+	  maximum possible space before it, to store other data.
+
 	  These two values must be aligned to an MMC sector boundary.
 
 	- CONFIG_ENV_OFFSET_REDUND (optional):
@@ -3636,6 +3644,9 @@  but it can not erase, write this NOR flash by SRIO or PCIE interface.
 	  valid backup copy in case the other copy is corrupted, e.g. due
 	  to a power failure during a "saveenv" operation.
 
+	  This value may also be positive or negative; this is handled in the
+	  same way as CONFIG_ENV_OFFSET.
+
 	  This value must also be aligned to an MMC sector boundary.
 
 	- CONFIG_ENV_SIZE_REDUND (optional):
diff --git a/common/env_mmc.c b/common/env_mmc.c
index 9ca098f..5d3a769 100644
--- a/common/env_mmc.c
+++ b/common/env_mmc.c
@@ -53,11 +53,19 @@  DECLARE_GLOBAL_DATA_PTR;
 
 __weak int mmc_get_env_addr(struct mmc *mmc, int copy, u32 *env_addr)
 {
-	*env_addr = CONFIG_ENV_OFFSET;
+	s64 offset;
+
+	offset = CONFIG_ENV_OFFSET;
 #ifdef CONFIG_ENV_OFFSET_REDUND
 	if (copy)
-		*env_addr = CONFIG_ENV_OFFSET_REDUND;
+		offset = CONFIG_ENV_OFFSET_REDUND;
 #endif
+
+	if (offset < 0)
+		offset += mmc->capacity;
+
+	*env_addr = offset;
+
 	return 0;
 }