Nios® II Embedded Design Suite (EDS)
Support for Embedded Development Tools, Processors (SoCs and Nios® II processor), Embedded Development Suites (EDSs), Boot and Configuration, Operating Systems, C and C++
Announcements
Intel Support hours are Monday-Fridays, 8am-5pm PST, except Holidays. Thanks to our community members who provide support during our down time or before we get to your questions. We appreciate you!

Need Forum Guidance? Click here
Search our FPGA Knowledge Articles here.

Device tree corrupted

Altera_Forum
Honored Contributor II
1,101 Views

Hi all, 

 

I use a 3.2.0 kernel from Altera git on a nios2, I have no bootloader and an initramfs. the kernel is loaded directly from Flash using altera_nios2/boot_loader_cfi.srec 

 

The kernel succeed to boot however I have corrupted values in /proc/device-tree. For instance in /proc/device-tree/sopc@0/gpio@0x2000120compatible : 

hexdump -C /proc/device-tree/sopc@0/gpio@0x2000120/compatible 00000000 68 69 73 20 68 65 6c 70 20 6d 65 73 73 61 67 65 |his help message| 00000010 20 61 6e 64 20 65 78 69 74 00 00 | and exit..|  

 

It is the same for each of my devices in the device-tree. I search for those corrupted values and found that they comes from rootfs.initramfs file generated by uClinux 

 

I really don't know how to fix this. It is a big problem since I can't even drive a GPIO :(. 

 

I found some old forums that talk about a similar issue : 

http://comments.gmane.org/gmane.linux.uclinux.nios2.devel/692 

There is no solution and this forum seems dead. 

 

Does someone had the same issue ?  

And does one find a solution ? 

 

Best regards
0 Kudos
1 Reply
Altera_Forum
Honored Contributor II
75 Views

Hi, 

 

So after (a long) investigation I think I can explain what happens, but I don't know how to correct since it may come from the kernel sources (I'm not kernel developper). 

 

If I dump device-tree values before starting any application the device-tree (DT) is not corrupted. 

If I start an application that requires quite a lot of memory (such as net-snmp with V3 and encryption/authentication enabled) the device tree becomes corrupted. 

 

It seems that applications that requires quite a lot of memory are overwritting DT values. 

So I configure my GPIOs before DT corruption and everything works fine. 

 

But this is not really a solution. 

So I think that the DT area is not protected from a kernel point of vu, so the memory used by the DT is overwritten when kernel needs more memory. 

 

I don't know if the issue comes from the kernel itself or from uClinux stuffs. 

 

Does someone has any idea ? 

 

Best regards 

 

PS : the more strange behaviour is that with a 16MB RAM the DT needs more applications running to be corrupted, with a 32MB RAM the DT is corrupted after the first application launched?
Reply