to, 2025-04-24 kello 16:19 -0400, Luiz Augusto von Dentz kirjoitti:
> Hi Pauli,
>
> On Thu, Apr 24, 2025 at 3:59 PM Pauli Virtanen <pav@xxxxxx> wrote:
> >
> > Copy timestamp too when allocating new skb for received fragment.
> > Fixes missing RX timestamps with fragmentation.
> >
> > Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> > Signed-off-by: Pauli Virtanen <pav@xxxxxx>
>
> Hmm I though it only matters for TX timestamp but this is actually
> different, anyway I don't think it would be possible to backport to
> that old patch, so we might need to use a more recent one when we
> introduced l2cap_recv_frag to make sure it properly backported:
>
> Fixes: 4d7ea8ee90e4 ("Bluetooth: L2CAP: Fix handling fragmented length")
>
> I can fix it up in place though, so don't bother spinning a new one,
Yes, that's probably first where the patch applies directly. The bug
itself is old.
> btw shall we replace the __net_timestamp with skb_set_delivery_time?
AFAICS the __net_timestamp() in hci_recv_frame() is correct as it is
now.
> > ---
> > net/bluetooth/l2cap_core.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/net/bluetooth/l2cap_core.c b/net/bluetooth/l2cap_core.c
> > index 5ca7ac43c58d..73472756618a 100644
> > --- a/net/bluetooth/l2cap_core.c
> > +++ b/net/bluetooth/l2cap_core.c
> > @@ -7415,6 +7415,9 @@ static int l2cap_recv_frag(struct l2cap_conn *conn, struct sk_buff *skb,
> > return -ENOMEM;
> > /* Init rx_len */
> > conn->rx_len = len;
> > +
> > + skb_set_delivery_time(conn->rx_skb, skb->tstamp,
> > + skb->tstamp_type);
> > }
> >
> > /* Copy as much as the rx_skb can hold */
> > --
> > 2.49.0
> >
> >
>
--
Pauli Virtanen
