lkml.org 
[lkml]   [2021]   [Sep]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC PATCH] x86, vmlinux.lds: Add debug option to force all data sections aligned
On Thu, Sep 23, 2021 at 10:57:20PM +0800, Feng Tang wrote:
> For binary size, I just tested 5.14 kernel with a default desktop
> config from Ubuntu (I didn't use the normal rhel-8.3 config used
> by 0Day, which is more for server):
>
> v5.14
> ------------------------
> text data bss dec hex filename
> 16010221 14971391 6098944 37080556 235cdec vmlinux
>
> v5.14 + 64B-function-align
> --------------------------
> text data bss dec hex filename
> 18107373 14971391 6098944 39177708 255cdec vmlinux
>
> v5.14 + data-align(THREAD_SIZE 16KB)
> --------------------------
> text data bss dec hex filename
> 16010221 57001791 6008832 79020844 4b5c32c vmlinux

That data size increase is indeed excessive. However I wonder if some
other approach (other than SUBALIGN) could be taken. For example, a 4k
alignment for each compilation unit's .data section. That might require
some linker magic at the built-in.o linking level.

Anyway, I suspect the data alignment issues are less common than
function alignment. It might be fine to leave the data alignment as a
debug feature for now, as this current patch does.

> > On a similar vein I think we should re-explore permanently enabling
> > cacheline-sized function alignment i.e. making something like
> > CONFIG_DEBUG_FORCE_FUNCTION_ALIGN_64B the default. Ingo did some
> > research on that a while back:
> >
> > https://lkml.kernel.org/r/20150519213820.GA31688@gmail.com
>
> Thanks for sharing this, from which I learned a lot, and I hope I
> knew this thread when we first check strange regressions in 2019 :)
>
> > At the time, the main reported drawback of -falign-functions=64 was that
> > even small functions got aligned. But now I think that can be mitigated
> > with some new options like -flimit-function-alignment and/or
> > -falign-functions=64,X (for some carefully-chosen value of X).
>
> Will study more about these options.
>
> If they have much less size increase and no regression in performance,
> then maybe it could be turned on by default.

Agreed! I think/hope it would be a net positive change.

I've also been burned by such issues -- like a random one-line code
change causing a measurable performance regression due to changed
i-cache behavior in unrelated code. It doesn't only affect 0-day tests,
it also affects real users.

--
Josh

\
 
 \ /
  Last update: 2021-09-24 03:58    [W:0.082 / U:0.888 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site