Thanks Arnd and Ben.
We have changed our code by the review, and submit the patch again.
1. Use variable type integer to describe umal_speed instead of enum.
2. Changed the name ETHER_ADDR_LEN to ETH_ALEN.
3. Renamed those members in struct umaldmadscr to fit the naming convention.
4. Modified our code skb_reserve(sb_new, 2) to skb_reserve(sb_new, SMP_CACHE_BYTES + NET_IP_ALIGN)
in function umaldma_add_rcvbuffer.
5. Put judgment statement if (!netif_device_present(dev)) before lock the sc->umal_lock
in function umaldma_tx_process.
6. Reorder those functions in driver and delete all prototypes.
7. Modified our code comment to fit the demands of kernel-doc.
For the mac address, we use the fixed, single mac address just for board test, and this
mac address is just assigned to our company, so we didn't choose random_ether_addr().
And NAPI work is just on the way.
Tested-and-Signed-off-by: Zhang Shu <davidshuzhang@gmail.com> Signed-off-by: Chu Xiaowei <chuxiaowei@mprc.pku.edu.cn> Signed-off-by: Su Yonggang <tonyhook.su@gmail.com> Signed-off-by: Guan Xuetao <gxt@mprc.pku.edu.cn> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Ben Hutchings <bhutchings@solarflare.com> Cc: netdev@vger.kernel.org