Not sure I follow. Given that we have
struct mlx4_lso_seg {
__be32 mss_hdr_size;
__be32 header[0];
};
I don't see much difference between my proposal
> halign = ALIGN(wr->wr.ud.hlen + sizeof *wqe, 16);
and yours
> halign = ALIGN(wr->wr.ud.hlen + 4, 16);
since isn't sizeof *wqe == 4?
> I don't think so, at least in the case that hlen equals 48 which is a
> valid one since the total length used by the LSO segment would be 48 + 4
> which requires 4 * 16 bytes chunks. If we'd use the above statement the
> send would fail.
But the point is that the current code would only bump the wqe pointer
by 48 bytes and the last 4 bytes of the header would be overwritten by
the next data segment.
- R.
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
