Re: Aarch64 EXT4FS inode checksum failures - seems to be weak memory ordering issues

[Lukas Wunner]

On Tue, Jan 12, 2021 at 09:28:32AM -0800, Linus Torvalds wrote:
> On Tue, Jan 12, 2021 at 5:20 AM Lukas Wunner <lukas@xxxxxxxxx> wrote:
> > > Variable declarations in for-loops is the only one I can think of. I
> > > think that would clean up some code (and some macros), but might not
> > > be compelling on its own.
> >
> > Anonymous structs/unions.  I used to have a use case for that in
> > struct efi_dev_path in include/linux/efi.h, but Ard Biesheuvel
> > refactored it in a gnu89-compatible way for v5.7 with db8952e7094f.
> 
> We use anonymous structs/unions extensively and all over the place already.

Yes, my apologies, I mixed things up.

Back in 2016 when I authored 46cd4b75cd0e, what I wanted to do was
include an unnamed "struct efi_generic_dev_path;" in struct efi_dev_path:

struct efi_dev_path {
	struct efi_generic_dev_path;
	union {
		struct {
			u32 hid;
			u32 uid;
		} acpi;
		struct {
			u8 fn;
			u8 dev;
		} pci;
	};
} __attribute ((packed));

The alternative is to copy-paste the elements of struct efi_dev_path
or to give it a name such as "header" (which is what db8952e7094f
subsequently did).  Both options seemed inelegant to me.

However it turns out this feature requires -fms-extensions.
It's not part of -std=gnu11.

So coming back to topic, yes there doesn't seem to be too much to
be gained from moving to -std=gnu11 aside from variable declarations
in for-loops.

(And it really has to be -std=gnu11 because -std=c11 fails to compile.)

Thanks,

Lukas