Re: dangling pointer to '__str' error on ppc64_defconfig, GCC 12.1.0

[Michael Ellerman <mpe@ellerman.id.au>]

Petr Mladek <pmladek@suse.com> writes:
> On Thu 2022-05-26 16:17:15, Bagas Sanjaya wrote:
>> Hi everyone,
>> 
>> Attempting to build ppc64_defconfig kernel with powerpc64-unknown-linux-gnu
>> (GCC 12.1.0) on v5.18, I got build error on ftrace.o:
>> 
>>   CC      arch/powerpc/kernel/trace/ftrace.o
>>   CC      init/init_task.o
>> In file included from ./include/asm-generic/bug.h:22,
>>                  from ./arch/powerpc/include/asm/bug.h:156,
>>                  from ./include/linux/bug.h:5,
>>                  from ./include/linux/thread_info.h:13,
>>                  from ./include/asm-generic/preempt.h:5,
>>                  from ./arch/powerpc/include/generated/asm/preempt.h:1,
>>                  from ./include/linux/preempt.h:78,
>>                  from ./include/linux/spinlock.h:55,
>>                  from arch/powerpc/kernel/trace/ftrace.c:16:
>> arch/powerpc/kernel/trace/ftrace.c: In function 'ftrace_modify_code':
>> ./include/linux/printk.h:446:44: error: using a dangling pointer to '__str' [-Werror=dangling-pointer=]
>>   446 | #define printk(fmt, ...) printk_index_wrap(_printk, fmt, ##__VA_ARGS__)
>>       |                                            ^
>> ./include/linux/printk.h:418:17: note: in definition of macro 'printk_index_wrap'
>>   418 |                 _p_func(_fmt, ##__VA_ARGS__);                           \
>>       |                 ^~~~~~~
>> ./include/linux/printk.h:489:9: note: in expansion of macro 'printk'
>>   489 |         printk(KERN_ERR pr_fmt(fmt), ##__VA_ARGS__)
>>       |         ^~~~~~
>> arch/powerpc/kernel/trace/ftrace.c:75:17: note: in expansion of macro 'pr_err'
>>    75 |                 pr_err("%p: replaced (%s) != old (%s)",
>>       |                 ^~~~~~
>> In file included from ./arch/powerpc/include/asm/code-patching.h:14,
>>                  from arch/powerpc/kernel/trace/ftrace.c:26:
>> ./arch/powerpc/include/asm/inst.h:156:14: note: '__str' declared here
>>   156 |         char __str[PPC_INST_STR_LEN];   \
>>       |              ^~~~~
>
> IMHO, the problem is in the macro:
>
> #define ppc_inst_as_str(x)		\
> ({					\
> 	char __str[PPC_INST_STR_LEN];	\
> 	__ppc_inst_as_str(__str, x);	\
> 	__str;				\
> })
>
> The buffer __str is defined inside the code block {} and
> the macro passes the pointer to the buffer outside.
> IMHO, from the compiler POV, the buffer does not exist
> outside of the code block.
 
This warning seems kind of incompatible with statement expressions. But
I guess the compiler is happy as long as the value returned is not a
pointer. So I'm not sure I agree with the compiler here, but too bad for
me I guess.

This macro has always been a bit awkward, I think I'm inclined to just
get rid of it completely, and print the ulong value. I'll send a patch.

Anyway it's definitely not your problem, sorry you ended up on Cc.

cheers