On 05/14/2015 02:35 PM, Johannes Berg wrote:
On Wed, 2015-05-13 at 22:04 -0300, Gaston Gonzalez wrote:
.../rtl8192u/ieee80211/ieee80211_crypt_tkip.c | 53
++--------------------
include/net/mac80211.h | 3 ++
net/mac80211/tkip.c | 9 ++--
3 files changed, 10 insertions(+), 55 deletions(-)
That doesn't seem right to me. Clearly, that's a staging driver that
ships its own 802.11 subsystem, and bears no existing relation to
mac80211. Exporting a mac80211 internal function to make such a driver
happy makes me very suspicious. That function might not be very mac80211
specific, but exporting it from mac80211 doesn't seem right, that
introduces a massive and mostly artificial dependency into the driver.
Now arguably that driver is in staging and that doesn't matter, but then
why even bother cleaning this up? It seems likely that a well-written
driver for this would use mac80211, and not have to bother with this at
all since it could then use ieee80211_get_tkip_rx_p1k() or
ieee80211_get_tkip_p2k() or the update_tkip_key() method.
So I'm not fond of this patch and without a *very very* good reason and
explanation I'm not going to merge the mac80211 part. It certainly
bothers me much less that a crappy staging driver has a few lines of
duplicated code than it would to export mac80211 internals that real
mac80211 drivers don't need.
I agree. No staging driver should ever require changes in any part of mac80211!
If you want to get credit for kernel patches, you are welcome to fuss with the
code, but none of your changes should touch any part of the kernel other than
the driver you are working on. That is an absolute requirement.
One other comment is that I have never seen this hardware. I wonder if anyone
has it anymore.
Larry
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
