summaryrefslogtreecommitdiffstats
path: root/qemu/roms/u-boot/doc/README.davinci.nand_spl
diff options
context:
space:
mode:
authorYang Zhang <yang.z.zhang@intel.com>2015-08-28 09:58:54 +0800
committerYang Zhang <yang.z.zhang@intel.com>2015-09-01 12:44:00 +0800
commite44e3482bdb4d0ebde2d8b41830ac2cdb07948fb (patch)
tree66b09f592c55df2878107a468a91d21506104d3f /qemu/roms/u-boot/doc/README.davinci.nand_spl
parent9ca8dbcc65cfc63d6f5ef3312a33184e1d726e00 (diff)
Add qemu 2.4.0
Change-Id: Ic99cbad4b61f8b127b7dc74d04576c0bcbaaf4f5 Signed-off-by: Yang Zhang <yang.z.zhang@intel.com>
Diffstat (limited to 'qemu/roms/u-boot/doc/README.davinci.nand_spl')
-rw-r--r--qemu/roms/u-boot/doc/README.davinci.nand_spl141
1 files changed, 141 insertions, 0 deletions
diff --git a/qemu/roms/u-boot/doc/README.davinci.nand_spl b/qemu/roms/u-boot/doc/README.davinci.nand_spl
new file mode 100644
index 000000000..f46721a00
--- /dev/null
+++ b/qemu/roms/u-boot/doc/README.davinci.nand_spl
@@ -0,0 +1,141 @@
+With this approach, we don't need the UBL any more on DaVinci boards.
+A "make boardname" will compile a u-boot.ubl, with UBL Header, which is
+needed for the RBL to find the "UBL", which actually is a UBL-compatible
+header, nand spl code and u-boot code.
+
+
+As the RBL uses another read function as the "standard" u-boot,
+we need a command, which switches between this two read/write
+functions, so we can write the UBL header and the spl
+code in a format, which the RBL can read. This is realize
+(at the moment in board specific code) in the u-boot command
+nandrbl
+
+nandrbl without arguments returns actual mode (rbl or uboot).
+with nandrbl mode (mode = "rbl" or "uboot") you can switch
+between the two NAND read/write modes.
+
+
+To set up mkimage you need a config file for mkimage, example:
+board/ait/cam_enc_4xx/ublimage.cfg
+
+For information about the configuration please see:
+doc/README.ublimage
+
+Example for the cam_enc_4xx board:
+On the cam_enc_4xx board we have a NAND flash with blocksize = 0x20000 and
+pagesize = 0x800, so the u-boot.ubl image (which you get with:
+"make cam_enc_4xx") looks like this:
+
+00000000 00 ed ac a1 20 00 00 00 06 00 00 00 05 00 00 00 |.... ...........|
+00000010 00 00 00 00 20 00 00 00 ff ff ff ff ff ff ff ff |.... ...........|
+00000020 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff |................|
+*
+00000800 14 00 00 ea 14 f0 9f e5 10 f0 9f e5 0c f0 9f e5 |................|
+00000810 08 f0 9f e5 04 f0 9f e5 00 f0 9f e5 04 f0 1f e5 |................|
+00000820 00 01 00 00 78 56 34 12 78 56 34 12 78 56 34 12 |....xV4.xV4.xV4.|
+[...]
+*
+00001fe0 00 00 00 00 00 00 00 00 ff ff ff ff ff ff ff ff |................|
+00001ff0 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff |................|
+*
+00003800 14 00 00 ea 14 f0 9f e5 14 f0 9f e5 14 f0 9f e5 |................|
+00003810 14 f0 9f e5 14 f0 9f e5 14 f0 9f e5 14 f0 9f e5 |................|
+00003820 80 01 08 81 e0 01 08 81 40 02 08 81 a0 02 08 81 |........@.......|
+
+In the first "page" of the image, we have the UBL Header, needed for
+the RBL to find the spl code.
+
+The spl code starts in the second "page" of the image, with a size
+defined by:
+
+#define CONFIG_SYS_NROF_PAGES_NAND_SPL 6
+
+After the spl code, there comes the "real" u-boot code
+@ (6 + 1) * pagesize = 0x3800
+
+------------------------------------------------------------------------
+Setting up spl code:
+
+/*
+ * RBL searches from Block n (n = 1..24)
+ * so we can define, how many UBL Headers
+ * we write before the real spl code
+ */
+#define CONFIG_SYS_NROF_UBL_HEADER 5
+#define CONFIG_SYS_NROF_PAGES_NAND_SPL 6
+
+#define CONFIG_SYS_NAND_U_BOOT_OFFS ((CONFIG_SYS_NROF_UBL_HEADER * \
+ CONFIG_SYS_NAND_BLOCK_SIZE) + \
+ (CONFIG_SYS_NROF_PAGES_NAND_SPL) * \
+ CONFIG_SYS_NAND_PAGE_SIZE)
+------------------------------------------------------------------------
+
+Burning into NAND:
+
+step 1:
+The RBL searches from Block n ( n = 1..24) on page 0 for valid UBL
+Headers, so you have to burn the UBL header page from the u-boot.ubl
+image to the blocks, you want to have the UBL header.
+!! Don;t forget to switch to rbl nand read/write functions with
+ "nandrbl rbl"
+
+step 2:
+You need to setup in the ublimage.cfg, where the RBL can find the spl
+code, and how big it is.
+
+!! RBL always starts reading from page 0 !!
+
+For the AIT board, we have:
+PAGES 6
+START_BLOCK 5
+
+So we need to copy the spl code to block 5 page 0
+!! Don;t forget to switch to rbl nand read/write functions with
+ "nandrbl rbl"
+
+step 3:
+You need to copy the u-boot image to the block/page
+where the spl code reads it (CONFIG_SYS_NAND_U_BOOT_OFFS)
+!! Don;t forget to switch to rbl nand read/write functions with
+ "nandrbl uboot", which is default.
+
+On the cam_enc_4xx board it is:
+#define CONFIG_SYS_NAND_U_BOOT_OFFS (0xc0000)
+
+-> this results in following NAND usage on the cam_enc_4xx board:
+
+addr
+
+20000 possible UBL Header
+40000 possible UBL Header
+60000 possible UBL Header
+80000 possilbe UBL Header
+a0000 spl code
+c0000 u-boot code
+
+The above steps are executeed through the following environment vars:
+(using 80000 as address for the UBL header)
+
+pagesz=800
+uboot=/tftpboot/cam_enc_4xx/u-boot.ubl
+load=tftp 80000000 ${uboot}
+writeheader nandrbl rbl;nand erase 80000 ${pagesz};nand write 80000000 80000 ${pagesz};nandrbl uboot
+writenand_spl nandrbl rbl;nand erase a0000 3000;nand write 80000800 a0000 3000;nandrbl uboot
+writeuboot nandrbl uboot;nand erase c0000 5d000;nand write 80003800 c0000 5d000
+update=run load writeheader writenand_spl writeuboot
+
+If you do a "run load update" u-boot, spl + ubl header
+are magically updated ;-)
+
+Note:
+- There seem to be a bug in the RBL code (at least on my HW),
+ In the UBL block, I can set the page to values != 0, so it
+ is possible to burn step 1 and step 2 in one step into the
+ flash, but the RBL ignores the page settings, so I have to
+ burn the UBL Header to a page 0 and the spl code to
+ a page 0 ... :-(
+- If we make the nand read/write functions in the RBL equal to
+ the functions in u-boot (as I have no RBL code, it is only
+ possible in u-boot), we could burn the complete image in
+ one step ... that would be nice ...