Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] afs: fix no return statement in function returning
 non-void
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Zheng Zengkai <zhengzengkai@huawei.com>,
        Randy Dunlap <rdunlap@infradead.org>,
        David Howells <dhowells@redhat.com>,
        Hulk Robot <hulkci@huawei.com>, linux-afs@lists.infradead.org,
        Marc Dionne <marc.dionne@auristor.com>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
References: <162375813191.653958.11993495571264748407.stgit@warthog.procyon.org.uk>
 <CAHk-=whARK9gtk0BPo8Y0EQqASNG9SfpF1MRqjxf43OO9F0vag@mail.gmail.com>
 <f2764b10-dd0d-cabf-0264-131ea5829fed@infradead.org>
 <CAHk-=whPPWYXKQv6YjaPQgQCf+78S+0HmAtyzO1cFMdcqQp5-A@mail.gmail.com>
 <c2002123-795c-20ae-677c-a35ba0e361af@infradead.org>
 <051421e0-afe8-c6ca-95cd-4dc8cd20a43e@huawei.com>
 <200ea6f7-0182-9da1-734c-c49102663ccc@redhat.com>
 <CAHk-=wjEThm5Kyockk1kJhd_K-P+972t=SnEj-WX9KcKPW0-Qg@mail.gmail.com>
From:   Tom Rix <trix@redhat.com>
Message-ID: <9d7873b6-e35c-ae38-9952-a1df443b2aea@redhat.com>
Date:   Wed, 16 Jun 2021 09:22:02 -0700
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.8.1
MIME-Version: 1.0
In-Reply-To: <CAHk-=wjEThm5Kyockk1kJhd_K-P+972t=SnEj-WX9KcKPW0-Qg@mail.gmail.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Content-Language: en-US
Precedence: bulk


On 6/16/21 7:34 AM, Linus Torvalds wrote:
> On Wed, Jun 16, 2021 at 5:56 AM Tom Rix <trix@redhat.com> wrote:
>> A fix is to use the __noreturn attribute on this function
> That's certainly a better thing. It would be better yet to figure out
> why BUG() didn't do it automatically.
>
> Without CONFIG_BUG, it looks like powerpc picks up
>
>    #ifndef HAVE_ARCH_BUG
>    #define BUG() do {} while (1)
>
> which should still make it pointless to have the return.  But I might
> have missed something.

This looks like a problem the generic BUG().

with CONFIG_BUG=y, the *.i is

static int afs_dir_set_page_dirty(struct page *page)
{
  do { __asm__ __volatile__( "1:    " "twi 31, 0, 0" "\n" ".section 
__bug_table,\"aw\"\n" "2:\t.4byte 1b - 2b, %0 - 2b\n" "\t.short %1, 
%2\n" ".org 2b+%3\n" ".previous\n" : : "i" ("fs/afs/dir.c"), "i" (50), 
"i" (0), "i" (sizeof(struct bug_entry))); do { ; asm volatile(""); 
__builtin_unreachable(); } while (0); } while (0);
}
BUG() expanded from
#define BUG() do {                        \
     BUG_ENTRY("twi 31, 0, 0", 0);                \
     unreachable();                        \
} while (0)


with CONFIG_BUG=n, the *.i is

static int afs_dir_set_page_dirty(struct page *page)
{
  do {} while (1);
}

BUG() expanded from
  do {} while (1)

to fix, add an unreachable() to the generic BUG()

diff --git a/include/asm-generic/bug.h b/include/asm-generic/bug.h
index f152b9bb916f..b250e06d7de2 100644
--- a/include/asm-generic/bug.h
+++ b/include/asm-generic/bug.h
@@ -177,7 +177,10 @@ void __warn(const char *file, int line, void 
*caller, unsigned taint,

  #else /* !CONFIG_BUG */
  #ifndef HAVE_ARCH_BUG
-#define BUG() do {} while (1)
+#define BUG() do {                                             \
+               do {} while (1);                                \
+               unreachable();                                  \
+       } while (0)
  #endif

the new *.i

static int afs_dir_set_page_dirty(struct page *page)
{
  do { do {} while (1); do { ; asm volatile(""); 
__builtin_unreachable(); } while (0); } while (0);
}

The assembly is unchanged.

The key being the unreachable builtin

ref: gcc docs https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html

" ... without the __builtin_unreachable, GCC issues a
   warning that control reaches the end of a non-void function."

Tom

>
>               Linus
>