Sort of. He’s definitely right that make_u32_from_two_u16 is a terrible function name that obscures the meaning but I don’t think he’s right that the best solution is to inline it. C bit shifting is notoriously error prone - I’ve seen this bug multiple times:
uint32_t a = ...;
uint32_t b = ...;
uint64_t c = (a << 32) | b;
The real problem is the name isn’t very good. E.g. it could be u32_high_low_to_u64 or something. Might clearer. Certainly easily at kernel code levels of clarity.
(Really the naming issue comes from C not having keyword arguments but you can’t do anything about that.)
I am not a C developer, but I found the “helper has a terrible name” and “it’s not clear what the helper is doing” arguments a bit weak.
Who in they right mind does not think the helper creates a 32 bytes word by putting the 16 bytes of the first argument followed by the 16 bytes of the second one ?
Yeah it actually is fairly common to have the high word first because humans unfortunately picked the wrong endianness, and integers are written in big endian.
E.g. what value would you expect from u16x2_to_u32(0x1122, 0x3344)? If you said 0x11223344…
Still, the rant is stupid because all that needs to happen is to fix the name.
Honestly it’s really surprising that the kernel doesn’t already have a library of reliably but manipulation functions for common stuff like this.
It’s bits, not bytes. And endianness is a huge consideration in systems programming. And it’s basically Linus’ whole role at this point to enforce extreme consistency and standards since the project is so large with so many contributors
I mean, he explained what and why is garbage and he’s not wrong, so it’s a valuable lesson at least.
Sort of. He’s definitely right that
make_u32_from_two_u16is a terrible function name that obscures the meaning but I don’t think he’s right that the best solution is to inline it. C bit shifting is notoriously error prone - I’ve seen this bug multiple times:uint32_t a = ...; uint32_t b = ...; uint64_t c = (a << 32) | b;The real problem is the name isn’t very good. E.g. it could be
u32_high_low_to_u64or something. Might clearer. Certainly easily at kernel code levels of clarity.(Really the naming issue comes from C not having keyword arguments but you can’t do anything about that.)
I am not a C developer, but I found the “helper has a terrible name” and “it’s not clear what the helper is doing” arguments a bit weak.
Who in they right mind does not think the helper creates a 32 bytes word by putting the 16 bytes of the first argument followed by the 16 bytes of the second one ?
Yeah it actually is fairly common to have the high word first because humans unfortunately picked the wrong endianness, and integers are written in big endian.
E.g. what value would you expect from
u16x2_to_u32(0x1122, 0x3344)? If you said 0x11223344…Still, the rant is stupid because all that needs to happen is to fix the name.
Honestly it’s really surprising that the kernel doesn’t already have a library of reliably but manipulation functions for common stuff like this.
It’s bits, not bytes. And endianness is a huge consideration in systems programming. And it’s basically Linus’ whole role at this point to enforce extreme consistency and standards since the project is so large with so many contributors