[BUG] etnaviv: broken timeouts

[Kyle Russell]

I've observed the timeouts issue originally reported in this thread on a 
4.17 aarch64 kernel, and it has reproduced reliably every boot.  I tried 
the "normal" timespec patch that Russell provided, and that seems to 
have fixed it up.  Before finding this thread, I similarly considered 
using ktime_get_ts instead of timespec_to_jiffies while debugging the 
issue.  There don't appear to be many clients of timespec_to_jiffies, at 
least in 4.17.

Is this patch acceptable for now, or is it worth holding off for a more 
fundamental fix?

I haven't inspected the other usages of timespec_to_jiffies in detail, 
or searched newer kernel versions to see if usage of timespec_to_jiffies 
has changed, but if ktime_get_ts addresses some subtleties, maybe it's 
worth removing timespec_to_jiffies?

I'll be glad to help however I can.