On Tue, Apr 4, 2023 at 8:24 PM Alexei Starovoitov
<alexei.starovoitov@xxxxxxxxx> wrote:
>
> On Mon, Apr 03, 2023 at 09:36:52PM -0700, Andrii Nakryiko wrote:
> > @@ -1407,6 +1407,11 @@ union bpf_attr {
> > __aligned_u64 fd_array; /* array of FDs */
> > __aligned_u64 core_relos;
> > __u32 core_relo_rec_size; /* sizeof(struct bpf_core_relo) */
> > + /* output: actual total log contents size (including termintaing zero).
> > + * It could be both larger than original log_size (if log was
> > + * truncated), or smaller (if log buffer wasn't filled completely).
> > + */
> > + __u32 log_size_actual;
>
> Naming nit..
> In the networking subsystem there is skb->truesize.
> The concept is exposed to user space through tracepoints and well understood in networking.
> May be call this field 'log_truesize' ?
> With or without underscore.
Sounds good, naming pretty much the only part I wasn't sure about.
log_size_true or log_true_size, any preference? Latter reads more
naturally, so I'm guessing you'll prefer that one? Unless naming
"regularity" of log_size_true is preferred?
>
> Other than this the rest looks good and I believe it addresses Lorenz and Timo concerns.
> Would be good to hear from them.
+1, yep. I'll wait a bit more before resubmitting.
