LWN: Comments on "A summary of 2.6.26 API changes"

A summary of 2.6.26 API changes

nix — 2008-06-12T14:14:41+00:00

I'm not sure if a change to MAINTAINERS counts as an API change, really. :)

GCC inline parameter

hjernemadsen — 2008-06-12T07:30:09+00:00

The option have been in the kernel for quite some time, but unfortunately 
it didn't work:

http://lkml.org/lkml/2008/3/3/122

A summary of 2.6.26 API changes

hharrison — 2008-06-12T05:59:24+00:00

There's also the new unaligned access helpers that do byteswapping if
required:

u16 get_unaligned_le16
u32 get_unaligned_le32
u64 get_unaligned_le64
u16 get_unaligned_be16
u32 get_unaligned_be32
u64 get_unaligned_be64

Which returns a cpu-endian value from a (possibly) unaligned pointer
stored in the given endianness.

void put_unaligned_le16
void put_unaligned_le32
void put_unaligned_le64
void put_unaligned_be16
void put_unaligned_be32
void put_unaligned_be64

Which stores a cpu-endian value to a (possibly) unaligned pointer
in the given endianness.

GCC inline parameter

pr1268 — 2008-06-12T03:00:34+00:00

Hasn't the "GCC decide whether to inline" parameter setting been around for a while? I can recall seeing that in

Kernel hacking → Kernel debugging → Force gcc to inline functions marked 'inline'

for at least a year now.

Or, is its default value changing? I assume that leaving it unselected implies that gcc will decide whether to inline (as the change bullet above says).