linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Matthew Wilcox <willy@infradead.org>
To: Nadav Amit <nadav.amit@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>, linux-mm@kvack.org
Subject: Re: Number of arguments in vmalloc.c
Date: Mon, 3 Dec 2018 14:49:20 -0800	[thread overview]
Message-ID: <20181203224920.GQ10377@bombadil.infradead.org> (raw)
In-Reply-To: <4F09425C-C9AB-452F-899C-3CF3D4B737E1@gmail.com>

On Mon, Dec 03, 2018 at 02:04:41PM -0800, Nadav Amit wrote:
> On Dec 3, 2018, at 8:13 AM, Matthew Wilcox <willy@infradead.org> wrote:
> > On Mon, Dec 03, 2018 at 02:59:36PM +0100, Vlastimil Babka wrote:
> >> On 11/28/18 3:01 PM, Matthew Wilcox wrote:
> >>> Some of the functions in vmalloc.c have as many as nine arguments.
> >>> So I thought I'd have a quick go at bundling the ones that make sense
> >>> into a struct and pass around a pointer to that struct.  Well, it made
> >>> the generated code worse,
> >> 
> >> Worse in which metric?
> > 
> > More instructions to accomplish the same thing.
> > 
> >>> so I thought I'd share my attempt so nobody
> >>> else bothers (or soebody points out that I did something stupid).
> >> 
> >> I guess in some of the functions the args parameter could be const?
> >> Might make some difference.
> >> 
> >> Anyway this shouldn't be a fast path, so even if the generated code is
> >> e.g. somewhat larger, then it still might make sense to reduce the
> >> insane parameter lists.
> > 
> > It might ... I'm not sure it's even easier to program than the original
> > though.
> 
> My intuition is that if all the fields of vm_args were initialized together
> (in the same function), and a 'const struct vm_args *' was provided as
> an argument to other functions, code would be better (at least better than
> what you got right now).
> 
> I’m not saying it is easily applicable in this use-case (since I didn’t
> check).

Your intuition is wrong ...

   text	   data	    bss	    dec	    hex	filename
   9466	     81	     32	   9579	   256b	before.o
   9546	     81	     32	   9659	   25bb	.build-tiny/mm/vmalloc.o
   9546	     81	     32	   9659	   25bb	const.o

indeed, there's no difference between with or without the const, according
to 'cmp'.

Now, only alloc_vmap_area() gets to take a const argument.
__get_vm_area_node() intentionally modifies the arguments.  But feel
free to play around with this; you might be able to make it do something
worthwhile.

  reply	other threads:[~2018-12-03 22:49 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-11-28 14:01 Matthew Wilcox
2018-12-03 13:59 ` Vlastimil Babka
2018-12-03 16:13   ` Matthew Wilcox
2018-12-03 22:04     ` Nadav Amit
2018-12-03 22:49       ` Matthew Wilcox [this message]
2018-12-04  3:12         ` Nadav Amit
2018-12-06  8:28           ` Nadav Amit
2018-12-06 10:25             ` Peter Zijlstra
2018-12-06 11:24               ` Peter Zijlstra
2018-12-06 17:26               ` Nadav Amit
2018-12-07  8:45                 ` Peter Zijlstra
2018-12-07 23:12                   ` Nadav Amit
2018-12-08  0:40                     ` Should this_cpu_read() be volatile? Nadav Amit
2018-12-08 10:52                       ` Peter Zijlstra
2018-12-10  0:57                         ` Nadav Amit
2018-12-10  8:55                           ` Peter Zijlstra
2018-12-11 17:11                             ` Nadav Amit

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20181203224920.GQ10377@bombadil.infradead.org \
    --to=willy@infradead.org \
    --cc=linux-mm@kvack.org \
    --cc=nadav.amit@gmail.com \
    --cc=vbabka@suse.cz \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox