Re: linux-next: build failure after merge of the net-next tree

[Eric Dumazet]

On Fri, Mar 1, 2024 at 2:11 AM Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
>
> Hi all,
>
> After merging the net-next tree, today's linux-next build (arm
> multi_v7_defconfig) failed like this:
>
> In file included from <command-line>:
> In function 'tcp_struct_check',
>     inlined from 'tcp_init' at net/ipv4/tcp.c:4700:2:
> include/linux/compiler_types.h:442:45: error: call to '__compiletime_assert_940' declared with attribute error: BUILD_BUG_ON failed: offsetof(struct tcp_sock, __cacheline_group_end__tcp_sock_write_rx) - offsetofend(struct tcp_sock, __cacheline_group_begin__tcp_sock_write_rx) > 99
>   442 |         _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
>       |                                             ^
> include/linux/compiler_types.h:423:25: note: in definition of macro '__compiletime_assert'
>   423 |                         prefix ## suffix();                             \
>       |                         ^~~~~~
> include/linux/compiler_types.h:442:9: note: in expansion of macro '_compiletime_assert'
>   442 |         _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
>       |         ^~~~~~~~~~~~~~~~~~~
> include/linux/build_bug.h:39:37: note: in expansion of macro 'compiletime_assert'
>    39 | #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
>       |                                     ^~~~~~~~~~~~~~~~~~
> include/linux/build_bug.h:50:9: note: in expansion of macro 'BUILD_BUG_ON_MSG'
>    50 |         BUILD_BUG_ON_MSG(condition, "BUILD_BUG_ON failed: " #condition)
>       |         ^~~~~~~~~~~~~~~~
> include/linux/cache.h:108:9: note: in expansion of macro 'BUILD_BUG_ON'
>   108 |         BUILD_BUG_ON(offsetof(TYPE, __cacheline_group_end__##GROUP) - \
>       |         ^~~~~~~~~~~~
> net/ipv4/tcp.c:4687:9: note: in expansion of macro 'CACHELINE_ASSERT_GROUP_SIZE'
>  4687 |         CACHELINE_ASSERT_GROUP_SIZE(struct tcp_sock, tcp_sock_write_rx, 99);
>       |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~
>
> Caused by commit
>
>   99123622050f ("tcp: remove some holes in struct tcp_sock")
>
> I have reverted that commit for today.
>

I have no idea. Maybe this arch has some unusual alignments on
u64/u32/u16 fields ?

The patch should not have changed tcp_sock_write_rx group...

My patch reduced tcp_sock_write_tx on x86_64 from 113 to 105 bytes but
I did not bother changing the assert,
because the assertion triggers if the size of the group is bigger than
the numerical value.

So I could have added the following, but really did not bother.

diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index c82dc42f57c65df112f79080ff407cd98d11ce68..7e1b848398d04f2da2a91c3af97b1e2e3895b8ee
100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -4651,7 +4651,7 @@ static void __init tcp_struct_check(void)
        CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock,
tcp_sock_write_tx, tsorted_sent_queue);
        CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock,
tcp_sock_write_tx, highest_sack);
        CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock,
tcp_sock_write_tx, ecn_flags);
-       CACHELINE_ASSERT_GROUP_SIZE(struct tcp_sock, tcp_sock_write_tx, 113);
+       CACHELINE_ASSERT_GROUP_SIZE(struct tcp_sock, tcp_sock_write_tx, 105);

        /* TXRX read-write hotpath cache lines */
        CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock,
tcp_sock_write_txrx, pred_flags);