Received-SPF: pass (google.com: domain of linux-kernel+bounces-187848-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Message-ID: <986294ee-8bb1-4bf4-9f23-2bc25dbad561@efficios.com>
Date: Thu, 23 May 2024 13:53:42 -0400
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Content-Language: en-US
From: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
To: Kent Overstreet <kent.overstreet@linux.dev>,
 Brian Foster <bfoster@redhat.com>
Cc: Kees Cook <keescook@chromium.org>,
 linux-kernel <linux-kernel@vger.kernel.org>, linux-bcachefs@vger.kernel.org
Subject: Use of zero-length arrays in bcachefs structures inner fields
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit

Hi Kent,

Looking around in the bcachefs code for possible causes of this KMSAN
bug report:

https://lore.kernel.org/lkml/000000000000fd5e7006191f78dc@google.com/

I notice the following pattern in the bcachefs structures: zero-length
arrays members are inserted in structures (not always at the end),
seemingly to achieve a result similar to what could be done with a
union:

fs/bcachefs/bcachefs_format.h:

struct bkey_packed {
         __u64           _data[0];

         /* Size of combined key and value, in u64s */
         __u8            u64s;
[...]
};

likewise:

struct bkey_i {
         __u64                   _data[0];

         struct bkey     k;
         struct bch_val  v;
};

(and there are many more examples of this pattern in bcachefs)

AFAIK, the C11 standard states that array declarator constant expression
delimited by [ ] shall have a value greater than zero.

Effectively, we can verify that this code triggers an undefined behavior
with:

#include <stdio.h>

struct z {
         int x[0];
         int y;
         int z;
} __attribute__((packed));

int main(void)
{
         struct z a;

         a.y = 1;
         printf("%d\n", a.x[0]);
}

clang-15 -fsanitize=undefined -o a a.c
/a
a.c:14:17: runtime error: index 0 out of bounds for type 'int[0]'
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior a.c:14:17 in
1

Also, gcc warns that ISO C forbids zero-size arrays when compiling
with -pedantic:

gcc -std=c11 -pedantic -o a a.c
a.c:4:13: warning: ISO C forbids zero-size array ‘x’ [-Wpedantic]
     4 |         int x[0];

And clang states that this is only supported as an extension, even though
accessing it seems to be classified as an undefined behavior by UBSAN.

clang-15 -std=c11 -pedantic -o a a.c
a.c:4:8: warning: zero size arrays are an extension [-Wzero-length-array]
         int x[0];

So I wonder if the issue reported by KMSAN could be caused by this
pattern ?

Thanks,

Mathieu

-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com