Messages in this thread | | | Date | Sun, 27 Apr 2008 14:32:52 -0400 | From | Christoph Hellwig <> | Subject | Re: [PATCH] prepare kconfig inline optimization for all architectures |
| |
On Sun, Apr 27, 2008 at 11:24:28AM -0700, Linus Torvalds wrote: > The thing is, the "inline" vs "always_inline" thing _could_ make sense, > but sadly doesn't much. > > Part of it is that gcc imnsho inlines too aggressively anyway in the > absense of "inline", so there's no way "inline" can mean "you might > inline" this, because gcc will do that anyway even without it. As a > result, in _practice_ "inline" and "always_inline" end up being very close > to each other - perhaps more so than they should.
Yes, absolutely. In recent gcc versions it's in fact so bad that xfs had to add a hack to mark every function not explicitly marker inline noinline because otherwise gcc would inline gazillions of big functions which combines with it's horrible liveness analysis led to enormous stack useage. (I think the liveness part is getting better these days to be fair)
> It will take time before we've sorted out all the fall-out, because I bet > there is still code out there that _should_ use __always_inline, but > doesn't.
Given that __always_inline only shows up in non-x86 arch code in homeopathic quantity it's a fair bet that there's a lot of breakage waiting there. Which was one of the initial complains about the way it's done..
| |