igb driver triggers BUILD_BUG_ON on s390

git commit 74e238ead "igb: Support using build_skb in the case that jumbo frames are disabled" added a BUILD_BUG_ON() to the igb drivers which triggers on s390:

drivers/net/ethernet/intel/igb/igb_main.c:6231:2:
error: call to ‘__compiletime_assert_6235’ declared with attribute error:
BUILD_BUG_ON failed: SKB_WITH_OVERHEAD(IGB_RX_BUFSZ) <
(NET_SKB_PAD + NET_IP_ALIGN + IGB_TS_HDR_LEN + ETH_FRAME_LEN + ETH_FCS_LEN)

This happens mainly because s390 has an unusual large L1_CACHE_BYTES aka SMP_CACHE_BYTES define with 256 Bytes.

The BUILD_BUG_ON(..) translates to (taken from .i file):

__cond = !(!(((2048) - (((sizeof(struct skb_shared_info)) + (256 - 1)) & ~(256 - 1))) < (32 + 2 + 16 + 1514 + 4)));

With sizeof(struct skb_shared_info) == 320 we end up with

BUILD_BUG_ON((2048 - 512 = 1536) < 1568)

One possible solution would be to simply disable the driver on s390. Any opinions?

Quoted 9 lines · Heiko Carstens

git commit 74e238ead "igb: Support using build_skb in the case that jumbo frames are disabled" added a BUILD_BUG_ON() to the igb drivers which triggers on s390:

drivers/net/ethernet/intel/igb/igb_main.c:6231:2:
error: call to ‘__compiletime_assert_6235’ declared with attribute error:
BUILD_BUG_ON failed: SKB_WITH_OVERHEAD(IGB_RX_BUFSZ) <
(NET_SKB_PAD + NET_IP_ALIGN + IGB_TS_HDR_LEN + ETH_FRAME_LEN + ETH_FCS_LEN)

Quoted 2 lines

Any opinions?

A fix for this has already been submitted. As it turns out the BUILD_BUG_ON is mostly redundant anyway since we used a similar calculation to determine if we were going to enable the use of build_skb per ring.