TDA4VM: Pre-Fetch Exception R5 main domain

Hi,

On which address are you getting exception? Most likely, this is invalid location or not mapped in MPU region, or region is not executable. 

Regards,

Brijesh

Hi,

After receiving interrupt somehow the Prefetch abort occurs at 0xa2cc11a4 location. This address is not in memory map. 

My vector table is located at 0xc0000000 and copied to 0x00000000.

My Prefetch abort vector function is at 0xc000f64c. this is where the control jumps after Exception. As can be seen below.

I am not sure why is it going to 0xa2cc11a4! Is it possible that TCMs are not mapped. I am not sure what's wrong. 

Hi,

Is this location 0xa2cc11a4 part of the vector table? Can you please check it? 

Can you check which ISR is triggered and put breakpoint on this ISR to see if it is trying to access some invalid location? 

Regards,

Brijesh

No this location is not part of vector table. My vector table is at 0xc0000000 which is quite far away from 0xa2cc11a4. Also this address doesn't exist in memory map.

I have put breakpoint on all ISRs (e.g. IRQ, FIQ, Exceptions). As soon it receives any interrupt (IRQ), it hits the Pre-fetch abort exception. It seems like its jumping to wrong memory location. I have also tried with cache disabled. but no effect. 

Also is it possible there is some register (like vector table pointer register) that is not set or wrongly set?

i was talking about the content of the memory pointed by 0xc0000000. This is where the vector table is. Does it include in any location which starts with 0xa2 ??

Regards,

Brijesh

No it doesn't. neither contents nor any pointer to the 0xa2 memory location...

One more thing: The problem raised after shifting to https://www.ti.com/tool/download/PROCESSOR-SDK-LINUX-J721E/06.02.00.07 from 07.xx SDK (Firmware images and uboot).

Do you mean you did not see this issue in Linux-SDK7.0 and seeing it on Linux-SDK6.2? Does the R5F firmware remain same on both SDKs?

The above link is pointing to Linux and the issue is on main domain R5F. They should not create any issue. 

Yes. The problem was not observed while using R5F firmware images from 07. And a quick comparison via beyond compare the R5 firmware images are not the same in two SDKs.

which firmware are you loading on r5f? 

Is it released as part of SDK?

Loading this script via u-boot for R5F0.
dir_extracted/lib/firmware/app_remoteswitchcfg_server_pdk_mem_map_strip.xer5f though renaming it to  j7-main-r5f0_0-fw

dir_extracted/lib/firmware/pdkipc/ipc_echo_test_mcu2_1_release.strip.xer5f 

dir_extracted/lib/firmware/pdkipc/ipc_echo_test_mcu3_0_release.strip.xer5f

dir_extracted/lib/firmware/pdkipc/ipc_echo_test_mcu3_1_release.strip.xer5f

The above are the Firmware images i am loading.

Hi Brijesh,

Any clue/updates?

Hi Kowalski,

Are you still facing this issue? 

The prefetch abort exception occurs when the core attempts to execute an instruction from an illegal address or incorrect address in the current mapping of the MPU. For the time being can you try mapping this location in the MPU and put breakpoint on this location to see which core/ISR is calling this API?

Regards,

Brijesh