On Oct 8, 2008, at 8:17 AM, Jerry Van Baren wrote:
Kumar Gala wrote:
On Sep 29, 2008, at 8:49 AM, Jerry Van Baren wrote:
Kumar Gala wrote:
On Sep 23, 2008, at 10:07 AM, Kumar Gala wrote:
- Use new find_cmd_tbl() to process sub-commands
If this looks good I'll go ahead and clean it up for the other arches and OSes.
common/cmd_bootm.c | 142 ++++++++++++++++++++++++++++- include/image.h | 20 ++++- lib_ppc/bootm.c | 262 +++++++++++++++++++++++++++++++++ +------------------ 3 files changed, 330 insertions(+), 94 deletions(-)
Guys, Any comments on this?
- k
I didn't get to it over the weekend. Here in the North summer is a fleeting thing to be savored while it lasts. :-/
Any more feedback on this towards getting something into -next?
- k
Hi Kumar,
I've been working on the micro side (the bootm [a-z]+ commands) and the macro side (understanding cmd_bootm.c and image.c).
I've been keeping notes on the denx.de wiki (inappropriately) on the FDT page. They aren't entirely coherent either. :-/
I applied your changeset and verified that the traditional bootm syntax worked. I then tried to figure out a working sequence of bootm [a-z]+ commands and failed. http://www.denx.de/wiki/view/U-Boot/UBootFdtInfo#Sequence
- I didn't see any command for disabling interrupts
You have to have the interrupt command built in. than you can do 'interrupts off'
- When I do the "bootm loados" command, my target (MPC8360EMDS)
reboots
- Could be because interrupts are not disabled but we overwrite the
exception vectors.
- Could be I'm doing something wrong.
this is most likely the case.
Do you have a boot script "use case" to illustrate a sequence that works for you?
yes, will post in a follow up.
I have also been looking at the functions defined in cmd_bootm.c and image.c. Image.c is HUGE. Cmd_bootm.c is pretty complex and part of the complexity is that it is closely coupled with image.c functions.
- I would like to separate and understand the control flow separate
from the image handling, fdt handling, bd_t handling, etc.
- Image.c is WAY too large: 2x what it probably should be. Looking
at it, it seems like we should pull all the FIT stuff out of image.c and make a new fit_image.c. (I don't know what sort of coupling there would be between a legacy image.c and a new fit_image.c, I suspect not much.)
Enumerating the functions: <http://www.denx.de/wiki/view/U-Boot/UBootFdtInfo#Refactoring_cmd_bootm_c
http://www.denx.de/wiki/view/U-Boot/UBootFdtInfo#Refactoring_image_c
refactoring image.c is orthogonal to the functionality I'm trying to enable.
- k