Re: [PATCH] hurd: Implement device memory mapping

[Samuel Thibault]

Joan Lledó, le dim. 19 déc. 2021 12:14:18 +0100, a ecrit:
> El 12/12/21 a les 16:19, Samuel Thibault ha escrit:
> > 
> > Don't we need
> > 
> > else { pager = wobj; } ?
> > 
> > Otherwise pager would be undefined.
> 
> In the case when the two previous conditions are false, what about robj? can
> we be sure it's null and not leaked?

No, the io_map call could return different ports if it wants so, see the
io_map documentation:

“
For objects where read data and write data are the same, these objects
will be equal, otherwise they will be disjoint.
”

Samuel