On Mon, Jun 10, 2019 at 10:03 AM maowenan <maowenan@xxxxxxxxxx> wrote:
>
>
>
> On 2019/6/10 21:05, Willem de Bruijn wrote:
> > On Mon, Jun 10, 2019 at 8:17 AM Mao Wenan <maowenan@xxxxxxxxxx> wrote:
> >>
> >> The variable 'status' in __packet_lookup_frame_in_block() is never used since
> >> introduction in commit f6fb8f100b80 ("af-packet: TPACKET_V3 flexible buffer
> >> implementation."), we can remove it.
> >> And when __packet_lookup_frame_in_block() calls prb_retire_current_block(),
> >> it can pass macro TP_STATUS_KERNEL instead of 0.
> >>
> >> Signed-off-by: Mao Wenan <maowenan@xxxxxxxxxx>
> >> ---
> >> /* Ok, close the current block */
> >> - prb_retire_current_block(pkc, po, 0);
> >> + prb_retire_current_block(pkc, po, TP_STATUS_KERNEL);
> >
> > I don't think that 0 is intended to mean TP_STATUS_KERNEL here.
> >
> > prb_retire_current_block calls prb_close_block which sets status to
> >
> > TP_STATUS_USER | stat
> >
> > where stat is 0 or TP_STATUS_BLK_TMO.
>
>
> #define TP_STATUS_KERNEL 0
> #define TP_STATUS_BLK_TMO (1 << 5)
>
> Actually, packet_current_rx_frame calls __packet_lookup_frame_in_block with status=TP_STATUS_KERNEL
> in original code.
>
> __packet_lookup_frame_in_block in this function, first is to check whether the currently active block
> has enough space for the packet, which means status of block should be TP_STATUS_KERNEL, then it calls
> prb_retire_current_block to retire this block.
I know. I mean that the status here is what is passed to userspace on
block retire.
It is not intended to be TP_STATUS_USER | TP_STATUS_KERNEL. That makes no sense.
> Since there needs some discussion about means of status, I can send v2 only removing the parameter status of
> __packet_lookup_frame_in_block?
Sounds good.
