On 22/11/18 17:49, Guenter Roeck wrote:
On 11/22/18 10:07 AM, Murilo Fossa Vicentini wrote:
On 22/11/18 14:33, Guenter Roeck wrote:
drivers/net/ethernet/ibm/ibmvnic.c: In function 'ibmvnic_xmit':
drivers/net/ethernet/ibm/ibmvnic.c:789:40: error: implicit
declaration of function 'skb_vlan_tag_present'
Caused by 5f51bb855fa1 ("ibmvnic: fix accelerated VLAN handling").
Problem is missing include of linux/if_vlan.h, which was added upstream
with commit 6052d5e2a1961b ("ibmvnic: Insert header on VLAN tagged
received frame").
I have no idea if the offending patch fixes 6052d5e2a1961b, if both are
independent, if both are needed, or if none is needed.
Copying the authors of both patches.
Both address different parts of the code within the same VLAN
functionality, patch 5f51bb855fa1 (or upstream e84b47941e15) is making
a change on the transmit side, and the patch I sent upstream
(6052d5e2a196) addresses an issue when the system firmware strips the
VLAN tag on the receive side.
In the context of having proper VLAN support for the ibmvnic driver I
would say both are needed, although potentially there are other
patches needed for this functionality to work properly, for example
upstream commit da75e3b6a029 (ibmvnic: Account for VLAN tag in L2
Header descriptor) and upstream commit 8dff66cc4195 (ibmvnic: Account
for VLAN tag in L2 Header descriptor).
Not my call to make, but it seems to me that it might be better to drop
the patch from 4.9.y and ask people to use a later kernel if they need
vlan support for the ibmvnic driver.
Thanks,
Guenter
Not my call to make as well, but I agree with you, at the very least not
include it in this 4.9.139 release with the limit date for response /
concerns of "Fri Nov 23 18:34:55 UTC 2018" according to Greg.
In case it is needed, adding the maintainer for ibmvnic driver (Thomas
Falcon) in CC for his awareness / input.
Best regards,
Murilo Vicentini
