base-hw ,,, What is CACHE_LINE_SIZE_LOG2?

bob robjsstewart at ...196...
Mon Nov 18 16:22:27 CET 2013

On 11/18/2013 06:04 AM, Martin Stein wrote:
> On 18.11.2013 11:42, Norman Feske wrote:
>> Maybe your platform starts the kernel in
>> non-secure mode? If I remember right, the secure/non-secure mode must be
>> distinguished when setting up mappings in the page table.
> On ARMv7 a complete base-hw page-table is either secure or non-secure. A
> page table sets its secure bit automatically according to the mode that
> is active while the page table gets constructed (see
> base-hw/src/core/tlb/arm_v7.h). The decision, whether secure or
> non-secure mode is active (Arm_v7::Cpu::secure_mode()) depends on
> Board_base::SECURITY_EXTENSION (in
> base/include/platform/$YOUR_BOARD/drivers/board_base.h). So it might be
> that this parameter is set inappropriate for your board. When adding
> mappings to page tables, secure-mode setting doesn't have to be considered.
> Martin
> ------------------------------------------------------------------------------
> DreamFactory - Open Source REST & JSON Services for HTML5 & Native Apps
> OAuth, Users, Roles, SQL, NoSQL, BLOB Storage and External API Access
> Free app hosting. Or install the open source package on any LAMP server.
> Sign up and see examples for AngularJS, jQuery, Sencha Touch and Native!
> _______________________________________________
> Genode-main mailing list
> Genode-main at
Thanks for the responses Normal and Martin.

Trying the suggestion that the uart set-up may be the issue showed that 
that was not the issue.

However, setting the SECURITY_EXTENSION to 1 did produce an interesting 
result, in that the kernel crashed with an "undefined instruction" 
message. I traced this down to a call to the Security Control register, 
specifically to the assembly code for the read which is "asm volatile 
("mrc p15, 0, %[v], c1, c1, 0" : [v]"=r"(v) ::);". This is used in the 
method to get the security mode setting in the cpu object.According to 
the Armv7-A Architecture Reference Manual, the assembly code appears to 
be correct. I'm not sure how to interpret that. Does it mean that the 
Security control is just not implemented?



More information about the users mailing list