Fwd: Re[2]: [U-Boot-Users] at91rm9200 lowlevel_init.S
Zhukov
zhukov at tst.spb.su
Thu Jan 12 16:42:06 CET 2006
>> Does anybody know, what can be tuned in "lowlevel_init" procedure in
>> lowlevel_init.S module for at91rm9200dk? This init suspends u-boot. I
WD> The code is supposed to work fine on the at91rm9200dk board.
WD> Are you really using such a board, or is this some custom hardware?
Yes, my board is customozed - the hardware engineer edited
RomBoot because of some differences in SDRAM system.
But... anyway, the lowlevel_init.S seems so strange, that
some bug in its code is suspected.
Would you explain the meaning of _MTEXT_BASE in fragments like
this:
ldr r0, =SMRDATA1
ldr r1, _MTEXT_BASE
sub r0, r0, r1
add r2, r0, #176
2:
/* the address */
ldr r1, [r0], #4
/* the value */
ldr r3, [r0], #4
str r3, [r1]
cmp r2, r0
bne 2b
I guess _MTEXT_BASE is a start of code, the location just after vectors.
But what's the use of it here?
Suspecting hardware problem, I skip everything after the first step of
the procedure:
/* Get the CKGR Base Address */
ldr r1, =AT91C_BASE_CKGR
/* Main oscillator Enable register */
#ifdef CFG_USE_MAIN_OSCILLATOR
ldr r0, =0x0000FF01 /* Enable main oscillator */
#else
ldr r0, =0x0000FF00 /* Disable main oscillator */
#endif
str r0, [r1, #CKGR_MOR]
... /* delay */
b exit
This cut works fine, u-boot loads. Next, I tried to replace this
with the following:
ldr r0, =CKGRDATA
ldr r1, _MTEXT_BASE
sub r0, r0, r1
add r2, r0, #8 /* 8 bytes */
3:
/* the address */
ldr r1, [r0], #4
/* the value */
ldr r3, [r0], #4
str r3, [r1]
cmp r2, r0
bne 3b
... /* delay */
...
CKGRDATA:
.word AT91C_BASE_CKGR
.word 0x0000FF01
/* 8 bytes long */
It FAILS! Obviously, u-boot crashes here. Something wrong in code...
Best regards,
More information about the U-Boot
mailing list