This is a regression from the out-of-tree Intel IGB driver happened
between 5.3.5.10 and 5.3.5.18.
The condition here should be actually reveresed, but as we always can
be sure to have a MAX/MIN MTU defined we can just remove it,
essentially going back to the previous code state (which also works
with our current 4.15 kernel).
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
rm -rf ${BUILD_DIR}/${MODULES}/${IGBDIR} $@
mkdir -p ${BUILD_DIR}/${MODULES}/${IGBDIR}
tar --strip-components=1 -C ${BUILD_DIR}/${MODULES}/${IGBDIR} -xf ${IGBSRC}
+ cd ${BUILD_DIR}/${MODULES}/${IGBDIR}; patch -p1 < ../../../patches/intel/igb/igb_4.15_mtu.patch
touch $@
${IXGBEDIR}.prepared: ${IXGBESRC}
--- /dev/null
+diff --git a/src/igb_main.c.orig b/src/igb_main.c
+index 3ee1ec7..c8adf04 100644
+--- a/src/igb_main.c.orig
++++ b/src/igb_main.c
+@@ -5888,10 +5888,8 @@ static int igb_change_mtu(struct net_dev
+ while (test_and_set_bit(__IGB_RESETTING, &adapter->state))
+ usleep_range(1000, 2000);
+
+-#ifndef HAVE_NETDEVICE_MIN_MAX_MTU
+ /* igb_down has a dependency on max_frame_size */
+ adapter->max_frame_size = max_frame;
+-#endif
+
+ if (netif_running(netdev))
+ igb_down(adapter);