Android no NW - i.MX53 QSB

[Tiago Brito]

Hi, I'm trying to boot Android in the Normal World of an i.MX53 QSB using
U-boot in the Secure World as the bootloader.

I've adapted U-boot in order to switch to Normal World upon a SMC call.
The Android version I'm using is Adeneo BSP for Android Gingerbread v4.3.

Android boots fine when I disable the world switch in U-boot, but when I
try to boot it in the Normal World it freezes in a Kernel Panic before
running the Init.

I followed the patch made by Stefan Kalkowski (as suggested in
https://community.freescale.com/thread/308152 ) because I was experiencing
a similar problem (FAILED TO RELEASE IRAM PARTITION). Unlike Miao Yu in
that post my execution didn't stop there.
I thought this would be enough to run Android in the Secure World (because
this is what Stefan did) but apparently it didn't work.
The problem remains the same.

What happens is the kernel boots successfully and then the following flow
occurs:
(1) init_post() at kernel/init/main.c:826
(2) run_init_process() at kernel/init/main.c:817
(3) kernel_execve() at kernel/arch/arm/kernel/sys_arm.c:81
Inside kernel_execve() when we return to user space by manipulating the
kernel stack (asm code below) it lauches a Kernel Panic and init is never
executed.

ASM CODE:
asm(
...
"b ret_to_user"
:
: "r" (current_thread_info()),
"Ir" (THREAD_START_SP - sizeof(regs)),
"r" (&regs),
"Ir" (sizeof(regs))
: "r0", "r1", "r2", "r3", "ip", "lr", "memory");

In addition we noticed that the execution jumps to do_undefinstr at
kernel/arch/arm/kernel/entry-armv.S:689
b do_undefinstr

Besides this we confirmed that init is never executed by adding a print to
init's main function (src/system/core/init/init.c:713).

Thanks, Tiago Brito

[Stefan Kalkowski]

Hi Tiago,

Post by Tiago Brito
Hi, I'm trying to boot Android in the Normal World of an i.MX53 QSB using
U-boot in the Secure World as the bootloader.
I've adapted U-boot in order to switch to Normal World upon a SMC call.
The Android version I'm using is Adeneo BSP for Android Gingerbread v4.3.
Android boots fine when I disable the world switch in U-boot, but when I
try to boot it in the Normal World it freezes in a Kernel Panic before
running the Init.
I followed the patch made by Stefan Kalkowski (as suggested in
https://community.freescale.com/thread/308152 ) because I was experiencing
a similar problem (FAILED TO RELEASE IRAM PARTITION). Unlike Miao Yu in
that post my execution didn't stop there.
I thought this would be enough to run Android in the Secure World (because
this is what Stefan did) but apparently it didn't work.
The problem remains the same.
What happens is the kernel boots successfully and then the following flow
(1) init_post() at kernel/init/main.c:826
(2) run_init_process() at kernel/init/main.c:817
(3) kernel_execve() at kernel/arch/arm/kernel/sys_arm.c:81
Inside kernel_execve() when we return to user space by manipulating the
kernel stack (asm code below) it lauches a Kernel Panic and init is never
executed.
asm(
...
"b ret_to_user"
: "r" (current_thread_info()),
"Ir" (THREAD_START_SP - sizeof(regs)),
"r" (&regs),
"Ir" (sizeof(regs))
: "r0", "r1", "r2", "r3", "ip", "lr", "memory");
In addition we noticed that the execution jumps to do_undefinstr at
kernel/arch/arm/kernel/entry-armv.S:689
b do_undefinstr
Besides this we confirmed that init is never executed by adding a print to
init's main function (src/system/core/init/init.c:713).

I do not see any relation to the Genode OS framework following your
setup. Would you be so kind to respect the topic of this mailing list?

Just a short remark: not seeing any "printf" results from init does not
mean that no init code is executed at all, it could be one the
instructions in the call-trace of "printf", which caused the undefined
instruction. If you notice that the undefined instruction handler of
Linux is called, maybe you can get the values of SPSR_UND (saved program
status register of undefined mode) and LR_UND (link register of
undefined mode). Those values hint you what instruction (LR == former
instrcution pointer; SPSR gives you former execution mode) caused the
exception exactly.

Regards
Stefan

Post by Tiago Brito
Thanks, Tiago Brito
------------------------------------------------------------------------------
Find and fix application performance issues faster with Applications Manager
Applications Manager provides deep performance insights into multiple tiers of
your business applications. It resolves application problems quickly and
reduces your MTTR. Get your free trial!
https://ad.doubleclick.net/ddm/clk/302982198;130105516;z
_______________________________________________
genode-main mailing list
https://lists.sourceforge.net/lists/listinfo/genode-main