[U-Boot] OMAP3: relocation and bss usage (timer, gpmc, ...)

[U-Boot] OMAP3: relocation and bss usage (timer, gpmc, ...)

[Alexander Holler]

Hello,

browsing through some code searching for a reason why an u-boot compiled 
with gcc 4.5.1 fails to boot on a beagleboard, I've come through another 
place where bss is used before relocation: gpmc_init (gpmc_cfg) in 
arch/arm/cpu/armv7/omap3/mem.c.

I haven't fixed it up to now, but I want to inform the people actually 
working with that board (currently I don't have any to do with that 
board and just touching at from time to time for fun), that there might 
be other places than the already mentioned timer_init() and that 
gpmc_init().

All candidates for such problems could be found in  board_init_f() in 
arch/arm/lib/board.c, escpecially all functions which appear in 
init_sequence in board.c.

Someone should take the time and check them all. ;)

Regards,

Alexander

[U-Boot] OMAP3: relocation and bss usage (timer, gpmc, ...)

[Premi, Sanjeev]

> -----Original Message-----
> From: u-boot-bounces at lists.denx.de 
> [mailto:u-boot-bounces at lists.denx.de] On Behalf Of Alexander Holler
> Sent: Tuesday, November 30, 2010 9:59 PM
> To: u-boot at lists.denx.de
> Subject: [U-Boot] OMAP3: relocation and bss usage (timer, gpmc, ...)
> 
> Hello,
> 
> browsing through some code searching for a reason why an 
> u-boot compiled 
> with gcc 4.5.1 fails to boot on a beagleboard, I've come 
> through another 
> place where bss is used before relocation: gpmc_init (gpmc_cfg) in 
> arch/arm/cpu/armv7/omap3/mem.c.
> 
> I haven't fixed it up to now, but I want to inform the people 
> actually 
> working with that board (currently I don't have any to do with that 
> board and just touching at from time to time for fun), that 
> there might 
> be other places than the already mentioned timer_init() and that 
> gpmc_init().
> 
> All candidates for such problems could be found in  board_init_f() in 
> arch/arm/lib/board.c, escpecially all functions which appear in 
> init_sequence in board.c.

For omap3evm - I was getting multiple errors during compilation.
Most for undefined symbols - which existed in the compiled objects
but, I suspect, not in right sequence.

Haven't still been able to look back at these errors; still surprised
to find that you had the code running on beagle.

~sanjeev
> 
> Someone should take the time and check them all. ;)
> 
> Regards,
> 
> Alexander
> _______________________________________________
> U-Boot mailing list
> U-Boot at lists.denx.de
> http://lists.denx.de/mailman/listinfo/u-boot
> 

[U-Boot] OMAP3: relocation and bss usage (timer, gpmc, ...)

[Alexander Holler]

Hello,

Am 30.11.2010 19:21, schrieb Premi, Sanjeev:

> For omap3evm - I was getting multiple errors during compilation.
> Most for undefined symbols - which existed in the compiled objects
> but, I suspect, not in right sequence.
>
> Haven't still been able to look back at these errors; still surprised
> to find that you had the code running on beagle.

When I'm compiling the current master with gcc 4.3.5 (using binutils 
2.20.1) u.boot currently works (at least it loads and starts the kernel 
from mmc).
But I assume that is just some lucky condition because 4.3.5 does some 
things different than 4.5.1 which causes some sill existent bugs to not 
appear.

Browsing the source to search those problems (which arise when I'm using 
gcc 4.5.1) is just some freetime activity I'm doing when I feel bored, 
without spending much time. For actually working on those problems I 
would have to build a TI<->ARM JTAG cable, which is still on my list of 
low priority todos. ;)

Regards,

Alexander

[U-Boot] OMAP3: relocation and bss usage (timer, gpmc, ...)

[zfsdk]

Maybe in init sequens asm func, i used turn on led to check which func faild.

-- 
View this message in context: http://old.nabble.com/-U-Boot--OMAP3%3A-relocation-and-bss-usage-%28timer%2C-gpmc%2C-...%29-tp30341173p30358424.html
Sent from the Uboot - Users mailing list archive at Nabble.com.

[U-Boot] OMAP3: relocation and bss usage (timer, gpmc, ...)

[Alexander Holler]

Am 02.12.2010 14:33, schrieb zfsdk:
>
> Maybe in init sequens asm func, i used turn on led to check which func faild.
>

Using a LED for debugging is time consuming. And you won't see which 
code still uses BSS before relocation as such code does not have to fail 
(it still might work). I prefer reading source. Using JTAG should help 
too, but I think, using JTAG and a debugger might got somewhat harder, 
at least after the code got relocated.
Anyway, a JTAG-debugger could help a lot searching the stuff which 
currently goes wrong before relocation. I assume single stepping the few 
dozen lines which are executed before relocation could be done in a few 
hours and it should reveal most of the code which accesses BSS before 
relocation.

Regards,

Alexander