diff mbox

[U-Boot] spl: mmc: use block device number, not hard-coded 0

Message ID 1449343948-14839-1-git-send-email-eric@nelint.com
State Accepted
Delegated to: Tom Rini
Headers show

Commit Message

Eric Nelson Dec. 5, 2015, 7:32 p.m. UTC
In order to support boot from multiple devices through board_boot_order,
it's necessary to use the block number of a device.

The use of a hard-coded 0 for the device number also creates a need
to re-order block devices for use in SPL like this:
	http://git.denx.de/?p=u-boot.git;a=blob;f=board/freescale/mx6slevk/mx6slevk.c;hb=HEAD#l195

Signed-off-by: Eric Nelson <eric@nelint.com>
---
 common/spl/spl_mmc.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

Comments

Tom Rini Dec. 14, 2015, 1:21 a.m. UTC | #1
On Sat, Dec 05, 2015 at 12:32:28PM -0700, Eric Nelson wrote:
> In order to support boot from multiple devices through board_boot_order,
> it's necessary to use the block number of a device.
> 
> The use of a hard-coded 0 for the device number also creates a need
> to re-order block devices for use in SPL like this:
> 	http://git.denx.de/?p=u-boot.git;a=blob;f=board/freescale/mx6slevk/mx6slevk.c;hb=HEAD#l195

We also do similar'ish things for TI parts (we only register what we're
booting from).  So, before this can be applied don't we also need a
patch for mx6slevk and for TI ones too (but that should be "easier"
since I think it just becomes we always use the normal board_mmc_init) ?
Thanks!
Eric Nelson Dec. 14, 2015, 2:19 a.m. UTC | #2
Hi Tom,

On 12/13/2015 06:21 PM, Tom Rini wrote:
> On Sat, Dec 05, 2015 at 12:32:28PM -0700, Eric Nelson wrote:
>
>> In order to support boot from multiple devices through board_boot_order,
>> it's necessary to use the block number of a device.
>>
>> The use of a hard-coded 0 for the device number also creates a need
>> to re-order block devices for use in SPL like this:
>> 	http://git.denx.de/?p=u-boot.git;a=blob;f=board/freescale/mx6slevk/mx6slevk.c;hb=HEAD#l195
> 
> We also do similar'ish things for TI parts (we only register what we're
> booting from).
>
> So, before this can be applied don't we also need a patch for mx6slevk
> and for TI ones too (but that should be "easier" since I think it
> just becomes we always use the normal board_mmc_init) ?

We don't **need** a patch to support existing use cases because the
patch uses the dev number in the block device directly
(mmc->block_dev.dev).

	https://patchwork.ozlabs.org/patch/553026/

I'm not sure about TI boards, but the i.MX6SL EVK code swaps the
non-SPL ordering of mmc devices when compiled for SPL in order
to handle the hard-coded zero:

	http://git.denx.de/?p=u-boot.git;a=blob;f=board/freescale/mx6slevk/mx6slevk.c;hb=HEAD#l195

In other words, "mmc 0" becomes "mmc 1" and vice versa.

The trouble with this approach is that you can't allow for multiple
MMC devices through board_boot_order, because the second will be
device number 1.

Regards,


Eric
Tom Rini Dec. 15, 2015, 11:28 p.m. UTC | #3
On Sat, Dec 05, 2015 at 12:32:28PM -0700, Eric Nelson wrote:

> In order to support boot from multiple devices through board_boot_order,
> it's necessary to use the block number of a device.
> 
> The use of a hard-coded 0 for the device number also creates a need
> to re-order block devices for use in SPL like this:
> 	http://git.denx.de/?p=u-boot.git;a=blob;f=board/freescale/mx6slevk/mx6slevk.c;hb=HEAD#l195
> 
> Signed-off-by: Eric Nelson <eric@nelint.com>

Applied to u-boot/master, thanks!
diff mbox

Patch

diff --git a/common/spl/spl_mmc.c b/common/spl/spl_mmc.c
index b3c2c64..cc97e47 100644
--- a/common/spl/spl_mmc.c
+++ b/common/spl/spl_mmc.c
@@ -23,12 +23,13 @@  static int mmc_load_image_raw_sector(struct mmc *mmc, unsigned long sector)
 	unsigned long count;
 	u32 image_size_sectors;
 	struct image_header *header;
+	int dev_num = mmc->block_dev.dev;
 
 	header = (struct image_header *)(CONFIG_SYS_TEXT_BASE -
 					 sizeof(struct image_header));
 
 	/* read image header to find the image size & load address */
-	count = mmc->block_dev.block_read(0, sector, 1, header);
+	count = mmc->block_dev.block_read(dev_num, sector, 1, header);
 	debug("read sector %lx, count=%lu\n", sector, count);
 	if (count == 0)
 		goto end;
@@ -45,7 +46,7 @@  static int mmc_load_image_raw_sector(struct mmc *mmc, unsigned long sector)
 			     mmc->read_bl_len;
 
 	/* Read the header too to avoid extra memcpy */
-	count = mmc->block_dev.block_read(0, sector, image_size_sectors,
+	count = mmc->block_dev.block_read(dev_num, sector, image_size_sectors,
 					  (void *)(ulong)spl_image.load_addr);
 	debug("read %x sectors to %x\n", image_size_sectors,
 	      spl_image.load_addr);
@@ -172,7 +173,8 @@  static int mmc_load_image_raw_os(struct mmc *mmc)
 {
 	unsigned long count;
 
-	count = mmc->block_dev.block_read(0,
+	count = mmc->block_dev.block_read(
+		mmc->block_dev.dev,
 		CONFIG_SYS_MMCSD_RAW_MODE_ARGS_SECTOR,
 		CONFIG_SYS_MMCSD_RAW_MODE_ARGS_SECTORS,
 		(void *) CONFIG_SYS_SPL_ARGS_ADDR);