[OpenWrt-Devel] lantiq vr9 and fxs

[John Crispin]

On 15/10/2014 12:10, thomas.langer at lantiq.com wrote:
> Hello Eddi,
> 
> Eddi De Pieri wrote on 2014-10-14:
>> Hi to all,
>> 
>> I've applied to kernel patch for vpe committed for falcon, fixed
>> for vr9 ltq-vmmc, modified VGV7519 dts for passing new
>> parameters
>> 
>> console=ttyLTQ0,115200 init=/etc/preinit maxvpes=1 maxtcs=1
>> phym=64M mem=62M panic=1 vpe1_load_addr=0x83e00000 vpe1_mem=2M 
>> vpe1_wired_tlb_entries=1
> 
> Why do you reserve 2M? The voice FW only needs 1M
> 
>> 
>> Now I get:
>> 
>> [  175.672000] Here I am: #1 [  175.672000] Here I am: #2 [
>> 175.676000] Here I am: #3 [  175.680000] Here I am: #4 [
>> 175.680000] Here I am: #5 [  175.684000] Here I am:
>> cpu1_base_addr: 83e00000 #6 [  175.688000] Here I am: #7 [
>> 175.692000] Here I am: #8 [  175.700000] Unhandled kernel
>> unaligned access[#1]: [  175.700000] CPU: 0 PID: 1516 Comm:
>> tapidemo Not tainted 3.14.18 #40 [  175.700000] task: 829819b0
>> ti: 82b54000 task.ti: 82b54000 [  175.700000] $ 0   : 00000000
>> 77d95290 bf001f38 cc123456 [  175.700000] $ 4   : a3e00000
>> 00082280 ffff0000 80000000 [  175.700000] $ 8   : a6e5d6ef
>> 00080080 ffff00ff 91008110 [  175.700000] $12   : 807f807f
>> 293f1500 00000000 00fe00fe [  175.700000] $16   : 82b55da0
>> 82960000 82960000 00000000 [  175.700000] $20   : 00000001
>> 82b55e88 00000000 00000000 [  175.700000] $24   : 00000000
>> 00000000 [  175.700000] $28   : 82b54000 82b55d60 7fd3aec8
>> 8295ee84 [  175.700000] Hi    : 00001b08 [  175.700000] Lo    :
>> 00006c20 [  175.700000] epc   : bf001f44 0xbf001f44
> 
> This looks like a crash in the decrypt-routine, located in the
> bootrom.
> 
>> 
>> Any suggestion?
>> 
> 
> Please check the vmmc driver sources (I don't know which version is
> available in OpenWrt), if it already contains something like this:
> 
> const void *ifx_mps_get_decrypt_entrypoint(void) { #if
> defined(SYSTEM_AR9) return (const void *)0xbf0017c4; #elif
> defined(SYSTEM_VR9) u32 part_number;
> 
> part_number = IFX_MPS_CHIPID_PARTNUM_GET(*IFX_MPS_CHIPID); if
> ((part_number & 0xFFF0) == 0x01C0) return (const void
> *)0xbf001ea4; else return (const void *)0xbf001f38; #elif
> defined(SYSTEM_...) ... #endif /* SYSTEM_... */


you can skip this. its lantiqs attempt at obfuscation if i recall, its
a simple xor which can be done on the build host.


	John






> }
> 
> I don't know, why the people from the bootrom team did not manage
> to provide a constant entry point across all chips/chip-version,
> but this is another story.
> 
>> On Sat, Oct 11, 2014 at 5:35 PM, Eddi De Pieri <eddi at depieri.net>
>> wrote:
>>> Hi to all,
>>> 
>>> I've started to figure out how much work is needed to support
>>> fxs on vr9 board like mine.
>>> 
>>> On  latest trunk : - owsip is missing (i know is bugged with
>>> >32mbram) - ltq-tapidemo is m issing - lantiq falcon fxs
>>> support seems broken
> 
> Falcon and vr9 are different SoCs!
> 
>>> 
>>> It seems to me that by rebasing some patch for falcon should be
>>> possible to make vmmc for vr9 working (or at least to build for
>>> now...).
>>> 
>>> You agree?
>>> 
>>> Eddi
> 
> Best Regards, Thomas --- There are two hard things in computer
> science: cache invalidation, naming things, and off-by-one errors. 
> --- _______________________________________________ openwrt-devel
> mailing list openwrt-devel at lists.openwrt.org 
> https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
> 
_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel