Dear Benoît Thébaudeau,
Dear Marek Vasut,
On Friday, April 19, 2013 1:55:31 PM, Benoît Thébaudeau wrote:
Dear Marek Vasut,
On Friday, April 19, 2013 1:14:16 PM, Marek Vasut wrote:
Dear Benoît Thébaudeau,
On Friday, April 19, 2013 10:38:48 AM, Benoît Thébaudeau wrote:
Dear Marek Vasut,
On Friday, April 19, 2013 6:10:51 AM, Marek Vasut wrote:
Add support for generic NAND SPL via the SPL framework into the mxc_nand_spl driver. This is basically just a simple rename and publication of the already implemented functions. To avoid the old function which are used with the nand_spl/ stuff getting in the way of NAND SPL framework, the macro CONFIG_SPL_NAND_LEGACY was introduced and two remaining legacy boards were adjusted. These board need to be either fixed or removed in the long run, but I don't have either.
Also make sure the requested payload is aligned to full pages, otherwise this simple driver fails to load the last page.
Signed-off-by: Marek Vasut marex@denx.de Cc: Albert ARIBAUD albert.u.boot@aribaud.net Cc: Benoît Thébaudeau benoit.thebaudeau@advansee.com Cc: Fabio Estevam fabio.estevam@freescale.com Cc: Scott Wood scottwood@freescale.com Cc: Stefano Babic sbabic@denx.de Cc: Tom Rini trini@ti.com
drivers/mtd/nand/mxc_nand_spl.c | 13 ++++++++++--- include/configs/mx31pdk.h | 1 + include/configs/tx25.h | 1 + 3 files changed, 12 insertions(+), 3 deletions(-)
diff --git a/drivers/mtd/nand/mxc_nand_spl.c b/drivers/mtd/nand/mxc_nand_spl.c index 09f23c3..8ff03c9 100644 --- a/drivers/mtd/nand/mxc_nand_spl.c +++ b/drivers/mtd/nand/mxc_nand_spl.c @@ -290,7 +290,7 @@ static int is_badblock(int pagenumber)
return 0;
}
-static int nand_load(unsigned int from, unsigned int size, unsigned char *buf) +int nand_spl_load_image(uint32_t from, unsigned int size, void *buf)
{
int i; unsigned int page;
@@ -303,6 +303,7 @@ static int nand_load(unsigned int from, unsigned int size, unsigned char *buf)
page = from / CONFIG_SYS_NAND_PAGE_SIZE; i = 0;
size = roundup(size, CONFIG_SYS_NAND_PAGE_SIZE);
while (i < size / CONFIG_SYS_NAND_PAGE_SIZE) {
if (nfc_read_page(page, buf) < 0)
return -1;@@ -332,6 +333,7 @@ static int nand_load(unsigned int from, unsigned int size, unsigned char *buf)
return 0;
}
+#ifdef CONFIG_SPL_NAND_SUPPORT_LEGACY
/*
- The main entry for NAND booting. It's necessary that SDRAM is
already * configured and available since this code loads the main U-Boot image
@@ -345,8 +347,9 @@ void nand_boot(void)
* CONFIG_SYS_NAND_U_BOOT_OFFS and CONFIG_SYS_NAND_U_BOOT_SIZE must * be aligned to full pages */
- if (!nand_load(CONFIG_SYS_NAND_U_BOOT_OFFS,
CONFIG_SYS_NAND_U_BOOT_SIZE, - (uchar *)CONFIG_SYS_NAND_U_BOOT_DST)) {
if (!nand_spl_load_image(CONFIG_SYS_NAND_U_BOOT_OFFS,
CONFIG_SYS_NAND_U_BOOT_SIZE,(uchar *)CONFIG_SYS_NAND_U_BOOT_DST)) {/* Copy from NAND successful, start U-boot */ uboot = (void *)CONFIG_SYS_NAND_U_BOOT_START; uboot();
@@ -364,3 +367,7 @@ void hang(void)
/* Loop forever */ while (1) ;
}
+#endif
+void nand_init(void) {} +void nand_deselect(void) {} diff --git a/include/configs/mx31pdk.h b/include/configs/mx31pdk.h index 1754595..217552e 100644 --- a/include/configs/mx31pdk.h +++ b/include/configs/mx31pdk.h @@ -50,6 +50,7 @@
#define CONFIG_SPL_LDSCRIPT "arch/$(ARCH)/cpu/u-boot-
spl.lds"
#define CONFIG_SPL_MAX_SIZE 2048 #define CONFIG_SPL_NAND_SUPPORT
+#define CONFIG_SPL_NAND_SUPPORT_LEGACY
#define CONFIG_SPL_TEXT_BASE 0x87dc0000 #define CONFIG_SYS_TEXT_BASE 0x87e00000
diff --git a/include/configs/tx25.h b/include/configs/tx25.h index e72f8f6..7c362d0 100644 --- a/include/configs/tx25.h +++ b/include/configs/tx25.h @@ -37,6 +37,7 @@
#define CONFIG_SPL_LDSCRIPT "arch/$(ARCH)/cpu/u-
boot-
spl.lds"
#define CONFIG_SPL_MAX_SIZE 2048 #define CONFIG_SPL_NAND_SUPPORT
+#define CONFIG_SPL_NAND_SUPPORT_LEGACY
#define CONFIG_SPL_TEXT_BASE 0x810c0000 #define CONFIG_SYS_TEXT_BASE 0x81200000
-- 1.7.11.7
This is not about legacy vs. non-legacy. This is about basic vs. more featured SPL because of SPL size constraints. So what about dropping CONFIG_SPL_NAND_SUPPORT_LEGACY and testing for CONFIG_SPL_FRAMEWORK definition instead?
I was thinking about that, but the symbol is unrelated to NAND.
Well, it's CONFIG_SPL_FRAMEWORK + CONFIG_SPL_NAND_SUPPORT that defines the other implementation, and CONFIG_SPL_NAND_SUPPORT triggers the build of mxc_nand_spl.c for SPL, so the common point is CONFIG_SPL_FRAMEWORK.
I still think it's either a matter of fixing for new SPL or removing those two boards. The nand_spl/ stuff shall be removed ASAP.
Removing those boards is not a solution.
Is it really about "new" SPL? The generic SPL is enabled by CONFIG_SPL, and CONFIG_SPL_FRAMEWORK is a sub-option. If the generic SPL rules imposed to use the SPL framework functions, there would be no such sub-option. So it looks like these boards are complying to the new SPL rules.
We could see if using CONFIG_SPL_FRAMEWORK would still allow the SPL to fit in 2 kiB in order to drop this function, but if it does not fit, the new SPL rules should still make it possible to have a solution for any board having SPL size constraints.
I've just quickly build-tested the hack-patch below to have a guess of the SPL size for mx31pdk with CONFIG_SPL_FRAMEWORK, and I get 3.9 kiB, so way more than the maximum 2 kiB. Hence, these boards can't unfortunately be made more generic, but it's not a reason to remove them. The generic SPL must allow such hardware to be supported by U-Boot.
See the other email for my other idea. Either we can go with that or I can focus on making use of the regular MXC NAND driver on mx5, since there's no 2KiB limit there. [...]