Closed ghibo closed 7 years ago
Hi, thanks for the feedback. This patch will never be merged back into 4.9 upstream. Off the top of my head I don't remember why I added it (probably a prerequisite for others to apply cleanly), but out-of-tree drivers have always been at a disadvantage. That being said, I'll look into it.
So it turns out that this patch is not a prerequisite and can be reverted without problem. I think it just looked good at the time. ;-) I'll revert it for the next release (.39, currently in -rc1), due tomorrow or so.
OK. Should be necessary in the future, and the copy of jiffies are really cosmetic or abused, maybe the only part altering the structure members in include/linux/netdevice.h could be left out. Anyway for .39, just in the case you haven't yet prepared it, it seems that these patches:
mm-20170627-fix-crash-due-to-race-in-MADV_FREE-handling.patch net-20170621-avoid-unregistering-inet6_dev-for-loopback.patch net-20170621-only-call-ip6_route_dev_notify-once-for-NETDEV_UNREGISTER.patch net-20170624-reset-sk_rx_dst-in-tcp_disconnect.patch net-20170629-handle-NAPI_GRO_FREE_STOLEN_HEAD-case-also-in-napi_frags_finish.patch net-20170703-fix-slab-out-of-bounds-in-netdev_stats_to_stats64.patch sched-20170710-001-export-for_each_cpu_wrap.patch
need to be removed to as they are going to be merged upstream.
All already done. :)
:-)
Would be possible to revert/remove from the patchset the patch net-20170118-remove-usage-of-net_device-last_rx-member.patch, at least until it's not merged into upstream 4.9.x? Actually it changes the common structure net_device (see file include/linux/netdevice.h) removing one member, i.e. 'last_rx'. As a side effect it breaks the building of some 3rd party driver (which needs to be fixed in the same ways) such as ndiswrapper or broadcom-wl which still refers to last_rx member somewhere.