[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <aYioTsLEytuc1uue@1wt.eu>
Date: Sun, 8 Feb 2026 16:14:22 +0100
From: Willy Tarreau <w@....eu>
To: David Laight <david.laight.linux@...il.com>
Cc: Thomas Weißschuh <linux@...ssschuh.net>,
linux-kernel@...r.kernel.org, Cheng Li <lechain@...il.com>
Subject: Re: [PATCH v2 next 04/11] tools/nolibc/printf: Output pad characters
in 16 byte chunks
On Sat, Feb 07, 2026 at 11:43:05PM +0000, David Laight wrote:
> On Sat, 7 Feb 2026 20:38:36 +0100
> Willy Tarreau <w@....eu> wrote:
>
> > On Fri, Feb 06, 2026 at 07:11:14PM +0000, david.laight.linux@...il.com wrote:
> > > From: David Laight <david.laight.linux@...il.com>
> > >
> > > Simple to do and saves calls to the callback function.
> >
> > +20 bytes here but OK for me.
>
> I think some of those come back when I change all the variables to 'int'.
> Some, but not all, is because width is 32bit but len is 64bit.
> The final change that did:
> width -= len;
> ...
> while (width > 0)
> saved a surprising amount provided the ... contained some code.
> Breath on the code (or compiler version) and you easily get +/-60 bytes.
Yeah I tried as well with size_t (since compilers generally do not like
mixing data types and tend to place conversions every few instructions),
but some constants become larger and the code inflates as well.
Willy
Powered by blists - more mailing lists