Pelles C > FAQ

Unreachable code

(1/1)

Grincheux:
Could you make a thread on that subject.
Because I don't understand anything.
I have tested many compilers and Pelle is the only one that emits that kind of warning. I suppose it is is a good thing. But and can I correct this warning in my code.

Thank You

frankie:
Unreachable, or dead, code is that code that will never be executed, and for this reason will not be emitted by the compiler.
These parts are normally generated when a conditional execution resolves to a constant comparison that allows the compiler to evaluate that the code will never execute during compilation. I.e.

--- Code: ---    const a=2;    //This is a constant value
    ...
    if (a ==10)    //This condition will never be true. The code will be removed
    {
        printf("Dead code. Will never be executed!\");
    }

--- End code ---

Another example is when there is code after a return statement:

--- Code: ---void foo(void)
{
    printf("foo()\n");
    return;
    printf("Dead code. Will never be executed!\");
}

--- End code ---

The warn is very useful if you put the code there by mistake.

Grincheux:
Ok Frankie, I had understood this but this is not always the case.
Specially this one :


--- Code: ---uint64_t crc64speed(uint64_t crc, const void *s, const uint64_t l)
{
/* Quickly check if CRC table is initialized to little endian correctly. */
#ifndef CRC64SPEED_DUAL
    check_init(crc64_table, LITTLE1);
#else
    check_init(crc64_table_little, LITTLE1);
#endif
    return crcspeed64little(dual ? crc64_table_little : crc64_table, crc,(void *)s, l);
}

--- End code ---

--- Quote ---return crcspeed64little(dual ? crc64_table_little : crc64_table, crc,(void *)s, l);
--- End quote ---

https://github.com/mattsta/crcspeed
https://github.com/mattsta/crc64-compare
https://matt.sh/redis-crcspeed

frankie:
What is the exact warning text?
On which line? In the following line?

--- Code: ---return crcspeed64little(dual ? crc64_table_little : crc64_table, crc,(void *)s, l);

--- End code ---
It could also be a bug...

Navigation

[0] Message Index