Fred Fan wrote:
HI Stefano Babic, Sorry, I just see your and wolfgang's mail. In this days, I am prepare new soc's kernel MSL code. When I just finished my coding, I have same issue too. But during I identify this issue, the issue is disappeared. I does not change lowlevel_init.s. Maybe I removed some definitions of clock register. I will check it ASAP.
Hi Fred,
do not worry, I could find where the problem is located. I started from the IMX51 BSP, changing code to adapt for the u-boot top of tree, and then I applied all changes you sent in your patchset. So I could check step by step where I had to look.
I do not find any issue in the cpu/mx51/clock.c and in the clock definitions, that you rearranged after the first review, so in this file everything seems ok ;)
Definitely I checked with the debugger that the problem arises in the first initialization of the clock registers in lowlevel_init.S (after the init_clock label). I replaced the file with the one I found in the BSP and the board runs now flawlessly.
I have already removed the flashHeader.S, adding a new image format to support the i.MX51 processor and its DCD table.
I have opened another issue, too. I have seen that the name "babbage" is quite disappeared and on the Freescale's website this board is named now "mx51evk". I know you have already changed the name of the board, but I think the name "babbage" can confuse. Better should be to rename the board with the name we find in the official documentation. I know you already did this job, I can only volunteer myself to do that...
What do you think if I take over the job and I will resend all patches to the ML ? In this way I can easier integrate my changes, if you agree. Should be ok for you ?
Best regards, Stefano Babic