]>
Commit | Line | Data |
---|---|---|
a56bd922 | 1 | |
1eaeb58e | 2 | u-boot for the TI OMAP730 Perseus2 |
a56bd922 | 3 | |
1eaeb58e WD |
4 | Dave Peverley, MPC-Data Limited |
5 | http://www.mpc-data.co.uk | |
a56bd922 WD |
6 | |
7 | ||
8 | Overview : | |
9 | ||
1eaeb58e | 10 | As the OMAP730 is similar to the OMAP1610 in many ways, this port was based |
a56bd922 WD |
11 | on the u-boot port to the OMAP1610 Innovator. Supported features are : |
12 | ||
13 | - Serial terminal support | |
14 | - Onboard NOR Flash | |
15 | - Ethernet via the seperate debug board | |
16 | - Tested on Rev4 and Rev5 boards | |
17 | ||
18 | It has also been tested to work correctly when built with a 'standard' GCC | |
19 | 3.2.1 cross-compiler as well as Montavista Linux CEE 3.1's toolchain. | |
20 | ||
21 | ||
a56bd922 WD |
22 | Hardware Configuration : |
23 | ||
1eaeb58e | 24 | The main dips on the P2 board should be set to 2,3,7 and 9 on with all |
a56bd922 | 25 | others off. On the debug board, dips 1 and 7 should be on with the rest off. |
1eaeb58e WD |
26 | The serial console has been set up to run from the DB9 connector on the |
27 | P2 board at 115200 baud, 8 data bits, no stop bits, 1 parity bit. | |
a56bd922 WD |
28 | |
29 | It should be noted that the P2 board has NOR flash that is addressable via | |
1eaeb58e | 30 | either CS0 or CS3. This mode can be changed via DIP9 on the P2 board. |
a56bd922 WD |
31 | |
32 | ||
33 | Installing u-boot for the P2 : | |
34 | ||
35 | You can simply build u-boot for the Perseus by following the instructions | |
36 | in the main readme file. The target configuration is "omap730p2_config". | |
37 | Once u-boot has been built, you should strip the executable so it can be | |
38 | loaded via CCS (which cant cope with the symbols in the ELF binary) : | |
39 | $ cp u-boot u-boot.out | |
40 | $ arm-linux-strip u-boot.out | |
41 | ||
1eaeb58e | 42 | The method we've used for installing u-boot the first time on a P2 is |
a56bd922 WD |
43 | as follows : |
44 | ||
1eaeb58e | 45 | 1) Configure TI Code Composer Studio to connect to the P2 board via JTAG |
a56bd922 WD |
46 | as described in the Users Guide. |
47 | ||
1eaeb58e WD |
48 | 2) Set up the P2 to boot from CS3, and connect with CCS. Reset the CPU |
49 | and run the "init_mmu" GEL script. | |
a56bd922 WD |
50 | |
51 | 3) Use the "Load Program" option to send the u-boot.out file to the P2 and | |
1eaeb58e | 52 | run. |
a56bd922 WD |
53 | |
54 | At this point, u-boot should run and you will see the boot menu on your | |
55 | serial terminal. You can then load the u-boot image to memory : | |
56 | ||
57 | # loadb 0x10000000 | |
58 | ||
1eaeb58e | 59 | Send the "u-boot.bin" binary via the serial using Kermit. Once loaded |
a56bd922 WD |
60 | you can self-flash u-boot : |
61 | ||
62 | # protect off 1:0 | |
63 | # erase 1:0 | |
64 | # cp.b 0x10000000 0x0 0x20000 | |
a56bd922 | 65 | |
1eaeb58e | 66 | You should now be able to reset the board and run u-boot from flash. |
a56bd922 WD |
67 | |
68 | ||
69 | Alternative flash option : | |
70 | ||
71 | Sometimes, if you've been silly, you can get the board into a state where | |
72 | whats in flash has upset the board so much that you can no longer connect | |
1eaeb58e | 73 | to the P2 via JTAG. However, you can set DIP9 to off to swap the boot mode |
a56bd922 | 74 | of the P2 so that you boot from RAM instead of NOR flash. This moves NOR |
1eaeb58e | 75 | flash up to 0x0C000000. You can build a special version of u-boot to |
a56bd922 WD |
76 | utilise this by the following config : |
77 | ||
78 | $ make omap730p2_cs0boot_config | |
79 | ||
80 | If you load this up via CCS it will detect flash at its alternate location | |
81 | and allow you to programme your u-boot image (which, remember must be built | |
82 | for CS3 boot!) Once you do this, you can revert to CS3 boot and it will work | |
83 | fine again. | |
84 | ||
85 | ||
a56bd922 WD |
86 | Errata : |
87 | ||
88 | 1) It's been observed that sometimes the tftp transfer of kernels to the | |
1eaeb58e WD |
89 | board can have checksum errors or stall. This appears to be an issue |
90 | with the lan91c96.c driver, and can normally be worked around by | |
a56bd922 | 91 | resetting the board and trying again. |