On 6/14/22 14:52, Nadav Amit wrote:
Actually, I think passing around a struct might be overkill, when you can
simply collapse the various boolean args into a single flags arg. It looked
like a lot of the new args were bools...
Ok. Whatever you prefer. I thought that having something similar to “struct
vm_fault” makes sense, especially since it would allow to avoid propagating
ugly arguments like mmap_changing. But I do not care enough to argue.
Actually, I think you could make a case for passing around a struct.
The argument list is getting longer. I don't actually have an opinion
on that point, other than to point out the obvious: if you usually need
most of the struct fields (args) in each function call, then a struct
might be an improvement. But if you only need a few, then a struct passes
up the opportunity to have some function calls that are much simpler
and shorter.
It's the flags point that I really think you want to stick with, either
a flags arg, or a .flags field.
thanks,
--
John Hubbard
NVIDIA
