Re: [PATCH 14/24] thunderbolt: Let the connection manager handle all notifications

From: Lukas Wunner
Date: Wed May 24 2017 - 10:01:01 EST


On Thu, May 18, 2017 at 05:39:04PM +0300, Mika Westerberg wrote:
> @@ -320,18 +330,42 @@ static void tb_ctl_rx_callback(struct tb_ring *ring, struct ring_frame *frame,
> }
>
> frame->size -= 4; /* remove checksum */
> - if (*(__be32 *) (pkg->buffer + frame->size)
> - != tb_crc(pkg->buffer, frame->size)) {
> - tb_ctl_err(pkg->ctl,
> - "RX: checksum mismatch, dropping packet\n");
> - goto rx;
> - }
> + crc32 = tb_crc(pkg->buffer, frame->size);
> be32_to_cpu_array(pkg->buffer, pkg->buffer, frame->size / 4);
>
> - if (frame->eof == TB_CFG_PKG_EVENT) {
> - tb_ctl_handle_plug_event(pkg->ctl, pkg);
> + switch (frame->eof) {
> + case TB_CFG_PKG_READ:
> + case TB_CFG_PKG_WRITE:
> + case TB_CFG_PKG_ERROR:
> + case TB_CFG_PKG_OVERRIDE:
> + case TB_CFG_PKG_RESET:
> + if (*(__be32 *)(pkg->buffer + frame->size) != crc32) {
> + tb_ctl_err(pkg->ctl,
> + "RX: checksum mismatch, dropping packet\n");
> + goto rx;
> + }

Any harm keeping the crc32 check above the switch/case statement?
(And thus also execute it for unknown packets?)


> + case TB_CFG_PKG_EVENT:
> + if (*(__be32 *)(pkg->buffer + frame->size) != crc32) {
> + tb_ctl_err(pkg->ctl,
> + "RX: checksum mismatch, dropping packet\n");
> + goto rx;
> + }
> + tb_ctl_handle_event(pkg->ctl, frame->eof, pkg, frame->size);
> + goto rx;
> +
> + default:
> + tb_ctl_dbg(pkg->ctl, "RX: unknown package %#x, dropping\n",

The packet / package terminology is a bit inconsistent here.
Andreas originally used package. What's the term used by the TB spec?

Thanks,

Lukas