<dormito>
I wrote a custom set of scripts for the lx2160, however multicores don't work too well. Namely having openocd (with my script) running seems to prevent linux from onlining the other, non-boot cores. I modeled my script after targets/ls1046.cfg (and added a core_all_up function to easy bringup of all cores ), Is there something more the ls1046 needs needs to properly allow linux to bring up the
<dormito>
other cores?
<PaulFertser>
Hi dormito
<dormito>
hello
<PaulFertser>
dormito: it's unlikely someone here knows. But in general I guess you do not want to bring up the cores from OpenOCD if you later expect Linux to do it.
JoseT has joined #openocd
<dormito>
alright thanks. while I've tried not configuring the other cores, linux is still unable to bring them up. I'll see if I can find out what openocd is doing that is preventing an IPI
<PaulFertser>
But probably Linux is telling something specific?
<PaulFertser>
Also you have OpenOCD working right, so you should be able to actually _debug_ Linux with GDB and see why/how exactly it fails ;)
<dormito>
"failed to come online" and "failed in unknown state : 0x0", about the most generic errors possible
<PaulFertser>
But you have ELF, you have sources, you have full debug access to the first core that's running that code so
<dormito>
lol. I am trying to debug a linux, but the smp failure, don't actually prevent boot... it just rendners the kernel unstable for anything that might need to deal with other cores (like reboot...)
<PaulFertser>
But you also can debug that SMP bringup failure right away since you already have all the parts.
<PaulFertser>
So instead of non-helpful error message you can see what the kernel tried to do exactly and how it decided that failed.