On 1/14/19 3:14 PM, H.J. Lu wrote:
> On Mon, Jan 14, 2019 at 5:53 AM Richard Biener
> <richard.guent...@gmail.com> wrote:
>>
>> On Mon, Jan 14, 2019 at 2:46 PM H.J. Lu <hongjiu...@intel.com> wrote:
>>>
>>> This patch adds -Waddress-of-packed-member to GCC 9 porting guide.
>>>
>>> OK to install?
>>
>> The docs fail to mention what to do when the unaligned pointer is _not_
>> safe to use. That is, how do I fix
>>
>> struct { char c; int i[4]; } s __attribute__((packed));
>> int foo()
>> {
>> int *p = s.i;
>> return bar (p);
>> }
>> int bar (int *q)
>> {
>> return *q;
>> }
>>
>> for the cases where eliding the pointer isn't easily possible?
>
> You can't have both packed struct and aligned array at the same time.
> The only thing I can say is "don't do it".
>
>> Please also mention the new warning in changes.html
>> (it seems to be enabled by default even?).
>
> I will add a paragraph.
>
> H.J.
>> IIRC the frontends themselves build "bogus" pointer types
>> to aligned data from a simple &s.i[1] because the FIELD_DECLs
>> types are naturally aligned.
>>
>> Richard.
>>
>>> Thanks.
>>>
>>> H.J.
>>> ---
>>> Index: gcc-9/porting_to.html
>>> ===================================================================
>>> RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-9/porting_to.html,v
>>> retrieving revision 1.1
>>> diff -u -r1.1 porting_to.html
>>> --- gcc-9/porting_to.html 11 Jan 2019 18:21:45 -0000 1.1
>>> +++ gcc-9/porting_to.html 14 Jan 2019 13:46:07 -0000
>>> @@ -56,13 +56,36 @@
>>> }
>>> </code></pre>
>>>
>>> +<h2 id="c/cxx">C/C++ language issues</h2>
>>> +
>>> +<h3 id="Waddress-of-packed-member"><code>-Waddress-of-packed-member</code>
>>> +is enabled by default</h3>
>>> +
>>> +<p>
>>> + When address of packed member of struct or union is taken, it may result
>>> + in an unaligned pointer value. A new warning
>>> + <code>-Waddress-of-packed-member</code> was added to check alignment at
>>> + pointer assignment. It warns both unaligned address and unaligned
>>> + pointer.
>>> +</p>
>>> +
>>> +<p>
>>> + If the pointer value is safe to use, you can suppress
>>> + <code>-Waddress-of-packed-member</code> warnings by using pragmas:
>>> +</p>
>>> + <pre><code>
>>> + #pragma GCC diagnostic push
>>> + #pragma GCC diagnostic ignored "-Waddress-of-packed-member"
>>> + /* (code for which the warning is to be disabled) */
>>> + #pragma GCC diagnostic pop
>>> + </code></pre>
>>> +
>>> <!--
>>> <h2 id="cxx">C++ language issues</h2>
>>> -->
>>>
>>> <!--
>>> <h2 id="fortran">Fortran language issues</h2>
>>> --->
>>>
>>> <!--
>>> <h2 id="links">Links</h2>
>
>
>
Thanks for working on that.
Can we please mention a small demonstration of the problem in porting_to?
What about this:
$ cat pack.c
#include <stddef.h>
int main(void) {
struct foo {
char c;
int x;
} __attribute__((packed));
struct foo arr[2] = {{'a', 10}, {'b', 20}};
int *p0 = &arr[0].x;
int *p1 = &arr[1].x;
__builtin_printf("sizeof(struct foo) = %d\n", (int)sizeof(struct foo));
__builtin_printf("offsetof(struct foo, c) = %d\n", (int)offsetof(struct foo,
c));
__builtin_printf("offsetof(struct foo, x) = %d\n", (int)offsetof(struct foo,
x));
__builtin_printf("arr[0].x = %d\n", arr[0].x);
__builtin_printf("arr[1].x = %d\n", arr[1].x);
__builtin_printf("&arr = %p\n", arr);
__builtin_printf("p0 = %p\n", (void *)p0);
__builtin_printf("p1 = %p\n", (void *)p1);
__builtin_printf("*p0 = %d\n", *p0);
__builtin_printf("*p1 = %d\n", *p1);
return 0;
}
$ gcc pack.c -fsanitize=undefined && ./a.out
pack.c: In function ‘main’:
pack.c:9:13: warning: taking address of packed member of ‘struct foo’ may
result in an unaligned pointer value [-Waddress-of-packed-member]
9 | int *p0 = &arr[0].x;
| ^~~~~~~~~
pack.c:10:13: warning: taking address of packed member of ‘struct foo’ may
result in an unaligned pointer value [-Waddress-of-packed-member]
10 | int *p1 = &arr[1].x;
| ^~~~~~~~~
sizeof(struct foo) = 5
offsetof(struct foo, c) = 0
offsetof(struct foo, x) = 1
arr[0].x = 10
arr[1].x = 20
&arr = 0x7fffffffdc26
p0 = 0x7fffffffdc27
p1 = 0x7fffffffdc2c
pack.c:19:3: runtime error: load of misaligned address 0x7fffffffdc27 for type
'int', which requires 4 byte alignment
0x7fffffffdc27: note: pointer points here
00 00 00 61 0a 00 00 00 62 14 00 00 00 2c dc ff ff ff 7f 00 00 27 dc ff ff
ff 7f 00 00 80 12 40
^
*p0 = 10
*p1 = 20
---end---
It presents the new warning, run-time memory layout dump and also sanitizer
error.
Thoughts?
Martin
