Hi, On Sat, Nov 18, 2017 at 10:55 AM, Jian-Hong Pan <starnight@xxxxxxxxxxxx> wrote: > There are more functions and operations which must be used or implemented > in each IEEE 802.15.4 device driver, but are not mentioned in the Device > drivers API section of Documentation/networking/ieee802154.txt. Therefore, > I want to fulfill the missed part into the documentation with this patch. > > Signed-off-by: Jian-Hong Pan <starnight@xxxxxxxxxxxx> Acked-by: Alexander Aring <aring@xxxxxxxxxxxx> Thanks for doing this kind of job and blame for me that I didn't update documentation... xmit_complete is more a signal to say "frame transmission done", means you can fill the hardware side frame buffer with new frames again. It works together mit xmit_async, as xmit_sync (the old method) waits until framebuffer fill is complete. It queues the work into workqueue and people do "spi_sync" handling... but the ndo_xmit callback of netdev works differently. It means "fill the frame into framebuffer and start transmit" (in context of this callback) what "xmit_async" is made for - not queue into workqueue (what xmit_sync) does. Anyway I see problems when using subsystems e.g. traffic control which assumes that at this time the frame is being to transmitted and not queued somewhere else (in a software queue, e.g. workqueue). People say to me, "spi_sync" "spi_async" - doesn't matter the spi subsystem will do the same "things" afterwards, but then this is a bug inside the spi subsystem to handle softirq context correctly. - Alex -- To unsubscribe from this list: send the line "unsubscribe linux-wpan" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
