From: Daan De Meyer <daan.j.demeyer@xxxxxxxxx>
Date: Wed, 11 Oct 2023 21:09:33 +0200
> > From: Daan De Meyer <daan.j.demeyer@xxxxxxxxx>
> > Date: Wed, 11 Oct 2023 20:37:49 +0200
> > > > > @@ -1483,11 +1488,18 @@ int __cgroup_bpf_run_filter_sock_addr(struct sock *sk,
> > > > > if (!ctx.uaddr) {
> > > > > memset(&unspec, 0, sizeof(unspec));
> > > > > ctx.uaddr = (struct sockaddr *)&unspec;
> > > > > - }
> > > > > + ctx.uaddrlen = 0;
> > > > > + } else
> > > > > + ctx.uaddrlen = *uaddrlen;
> > > > >
> > > > > cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
> > > > > - return bpf_prog_run_array_cg(&cgrp->bpf, atype, &ctx, bpf_prog_run,
> > > > > - 0, flags);
> > > > > + ret = bpf_prog_run_array_cg(&cgrp->bpf, atype, &ctx, bpf_prog_run,
> > > > > + 0, flags);
> > > > > +
> > > > > + if (!ret && uaddrlen)
> > > >
> > > > nit: no need to check uaddrlen here or maybe check ctx.uaddrlen.
> > >
> > > Are you sure? uaddrlen can still be NULL if uaddr is also NULL
> >
> > How? In the patch 2 and 4, it seems uaddrlen always points to an
> > actual variable.
>
> Right, I was assuming we don't know for sure how callers are calling
> this function. It is right that right now no caller calls it with uaddrlen set
> to NULL.
We need not to be defensive for future in-kernel users who should take
care of that properly.
>
> It still seems like a good idea to check for uaddr instead of uaddrlen though,
> to mimic the same check that is done earlier in this function.
Sounds good.
