Russell King - ARM Linux writes: > On Wed, Sep 16, 2015 at 11:05:11AM -0400, Eric Anholt wrote: >> Lucas Stach writes: >> >> > From: Christian Gmeiner >> >> > +static bool etnaviv_validate_load_state(struct etnaviv_gpu *gpu, u32 *buf, >> > + unsigned int state, unsigned int num) >> > +{ >> > + return true; > ... >> > +} >> >> I was browsing the code, and noticed that it looks like you've got a >> debugging early return in your validation function here. > > That's because it's not quite ready to be enabled yet - the code as > it stands in my tree doesn't have the infrastructure to validate that > a LOADSTATE which touches an address register (or similar) has a > relocation appropriately marked, due to the way things happen. > Hopefully, when the user/kernel API is sorted, this can be resolved. OK. Validation code is painful to write, so I was just browsing what was in here to see if I could pick up anything useful.