public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: David Laight <david.laight.linux@gmail.com>
To: Willy Tarreau <w@1wt.eu>
Cc: "Thomas Weißschuh" <linux@weissschuh.net>,
	linux-kernel@vger.kernel.org, "Cheng Li" <lechain@gmail.com>
Subject: Re: [PATCH v2 next 03/11] tools/nolibc/printf: Add buffering to vfprintf() callback.
Date: Sat, 7 Feb 2026 23:36:51 +0000	[thread overview]
Message-ID: <20260207233651.577f0fcb@pumpkin> (raw)
In-Reply-To: <aYeSnohSJ78ADFUZ@1wt.eu>

On Sat, 7 Feb 2026 20:29:34 +0100
Willy Tarreau <w@1wt.eu> wrote:

> On Fri, Feb 06, 2026 at 07:11:13PM +0000, david.laight.linux@gmail.com wrote:
> > From: David Laight <david.laight.linux@gmail.com>
> > 
> > Add per-call buffering to the vprintf() callback.
> > While this adds some extra code it will speed things up and
> > makes a massive difference to anyone looking at strace output.  
> 
> This patch alone adds more than 200 extra bytes to the smallest binary
> for something that was never expressed as a need by users:
> 
>  $ size hello-patch*
>    text    data     bss     dec     hex filename
>    1859      48      24    1931     78b hello-patch1
>    2071      48      24    2143     85f hello-patch2
> 
> I doubt it would make sense to have a build option to choose this.
> Or alternately one could decide do disable it when __OPTIMIZE_SIZE__
> is defined. I just tried quickly and it does the job:

That probably makes sense.
For anything non-trivial the extra size is noise.
Actually is would be easier to read with a single #if covering the
whole lot.

> 
>   @@ -390,6 +390,7 @@ struct __nolibc_fprintf_cb_state {
>    
>    static int __nolibc_fprintf_cb(void *v_state, const char *buf, size_t size)
>    {
>   +#if !defined(__OPTIMIZE_SIZE__)
>           struct __nolibc_fprintf_cb_state *state = v_state;
>           unsigned int off = state->buf_offset;
>    
>   @@ -407,16 +408,24 @@ static int __nolibc_fprintf_cb(void *v_state, const char *buf, size_t size)
>                   memcpy(state->buf + off, buf, size);
>           }
>           return 0;
>   +#else
>   +       /* v_state is the stream */
>   +       return size ? _fwrite(buf, size, v_state) : 0;
>   +#endif
>    }
>    
>    static __attribute__((unused, format(printf, 2, 0)))
>    int vfprintf(FILE *stream, const char *fmt, va_list args)
>    {
>   +#if !defined(__OPTIMIZE_SIZE__)
>           struct __nolibc_fprintf_cb_state state;
>    
>           state.stream = stream;
>           state.buf_offset = 0;
>           return __nolibc_printf(__nolibc_fprintf_cb, &state, fmt, args);
>   +#else
>   +       return __nolibc_printf(__nolibc_fprintf_cb, stream, fmt, args);
>   +#endif
>    }
>    
>    static __attribute__((unused, format(printf, 1, 0)))
> 
> > Signed-off-by: David Laight <david.laight.linux@gmail.com>
> > ---
> > 
> > Changes for v2:
> > Formally patch 2, unchanged.
> > 
> >  tools/include/nolibc/stdio.h | 32 +++++++++++++++++++++++++++++---
> >  1 file changed, 29 insertions(+), 3 deletions(-)
> > 
> > diff --git a/tools/include/nolibc/stdio.h b/tools/include/nolibc/stdio.h
> > index 36733ecd4261..552f09d51d82 100644
> > --- a/tools/include/nolibc/stdio.h
> > +++ b/tools/include/nolibc/stdio.h
> > @@ -382,15 +382,41 @@ int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list
> >  	return written;
> >  }
> >  
> > -static int __nolibc_fprintf_cb(void *stream, const char *buf, size_t size)
> > +struct __nolibc_fprintf_cb_state {
> > +	FILE *stream;
> > +	unsigned int buf_offset;
> > +	char buf[128];
> > +};  
> 
> So that's the other state I was wondering if we could merge with the first
> one.

The snprintf state is the 'address to write the next character to'
and 'the amount of space remaining'.
No real overlap at all.

> 
> > +static int __nolibc_fprintf_cb(void *v_state, const char *buf, size_t size)
> >  {
> > -	return size ? _fwrite(buf, size, stream) : 0;
> > +	struct __nolibc_fprintf_cb_state *state = v_state;
> > +	unsigned int off = state->buf_offset;
> > +
> > +	if (off + size > sizeof(state->buf) || buf == NULL) {  
> 
> Please mention that special case of buf==NULL in a comment above the
> function. That's an internal API choice.

ok - it might just go above this line though.

	David

> 
> > +		state->buf_offset = 0;
> > +		if (off && _fwrite(state->buf, off, state->stream))
> > +			return -1;
> > +		if (size > sizeof(state->buf))
> > +			return _fwrite(buf, size, state->stream);
> > +		off = 0;
> > +	}
> > +
> > +	if (size) {
> > +		state->buf_offset = off + size;
> > +		memcpy(state->buf + off, buf, size);
> > +	}
> > +	return 0;
> >  }
> >  
> >  static __attribute__((unused, format(printf, 2, 0)))
> >  int vfprintf(FILE *stream, const char *fmt, va_list args)
> >  {
> > -	return __nolibc_printf(__nolibc_fprintf_cb, stream, fmt, args);
> > +	struct __nolibc_fprintf_cb_state state;
> > +
> > +	state.stream = stream;
> > +	state.buf_offset = 0;
> > +	return __nolibc_printf(__nolibc_fprintf_cb, &state, fmt, args);
> >  }
> >  
> >  static __attribute__((unused, format(printf, 1, 0)))  
> 
> Thanks,
> Willy


  reply	other threads:[~2026-02-07 23:36 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-02-06 19:11 [PATCH v2 next 00/11] tools/nolibc: Enhance printf() david.laight.linux
2026-02-06 19:11 ` [PATCH v2 next 01/11] tools/nolibc/printf: Change variable used for format chars from 'c' to 'ch' david.laight.linux
2026-02-07 18:51   ` Willy Tarreau
2026-02-16 18:52   ` Thomas Weißschuh
2026-02-06 19:11 ` [PATCH v2 next 02/11] tools/nolibc/printf: Move snprintf length check to callback david.laight.linux
2026-02-07 19:12   ` Willy Tarreau
2026-02-07 23:28     ` David Laight
2026-02-08 15:12       ` Willy Tarreau
2026-02-08 22:49         ` David Laight
2026-02-06 19:11 ` [PATCH v2 next 03/11] tools/nolibc/printf: Add buffering to vfprintf() callback david.laight.linux
2026-02-07 19:29   ` Willy Tarreau
2026-02-07 23:36     ` David Laight [this message]
2026-02-16 19:07       ` Thomas Weißschuh
2026-02-17 11:51         ` David Laight
2026-02-18 17:52           ` Thomas Weißschuh
2026-02-06 19:11 ` [PATCH v2 next 04/11] tools/nolibc/printf: Output pad characters in 16 byte chunks david.laight.linux
2026-02-07 19:38   ` Willy Tarreau
2026-02-07 23:43     ` David Laight
2026-02-08 15:14       ` Willy Tarreau
2026-02-16 19:30   ` Thomas Weißschuh
2026-02-16 22:29     ` David Laight
2026-02-18 17:30       ` Thomas Weißschuh
2026-02-06 19:11 ` [PATCH v2 next 05/11] tools/nolibc/printf: Simplify __nolibc_printf() david.laight.linux
2026-02-07 20:05   ` Willy Tarreau
2026-02-07 23:50     ` David Laight
2026-02-08 12:20       ` David Laight
2026-02-08 14:44         ` Willy Tarreau
2026-02-08 16:54           ` David Laight
2026-02-08 17:06             ` Willy Tarreau
2026-02-06 19:11 ` [PATCH v2 next 06/11] tools/nolibc/printf: Use bit-masks to hold requested flag, length and conversion chars david.laight.linux
2026-02-08 15:22   ` Willy Tarreau
2026-02-16 19:52   ` Thomas Weißschuh
2026-02-16 22:47     ` David Laight
2026-02-18 17:36       ` Thomas Weißschuh
2026-02-18 22:57         ` David Laight
2026-02-06 19:11 ` [PATCH v2 next 07/11] tools/nolibc/printf: Add support for conversion flags "#- +" and format "%X" david.laight.linux
2026-02-08 15:47   ` Willy Tarreau
2026-02-08 17:14     ` David Laight
2026-02-08 16:06   ` Willy Tarreau
2026-02-16 19:57   ` Thomas Weißschuh
2026-02-16 22:50     ` David Laight
2026-02-18 17:39       ` Thomas Weißschuh
2026-02-16 20:11   ` Thomas Weißschuh
2026-02-16 22:52     ` David Laight
2026-02-06 19:11 ` [PATCH v2 next 08/11] tools/nolibc/printf: Add support for zero padding and field precision david.laight.linux
2026-02-08 16:16   ` Willy Tarreau
2026-02-08 17:31     ` David Laight
2026-02-06 19:11 ` [PATCH v2 next 09/11] selftests/nolibc: Improve reporting of vfprintf() errors david.laight.linux
2026-02-16 20:05   ` Thomas Weißschuh
2026-02-17 10:48     ` David Laight
2026-02-18 17:48       ` Thomas Weißschuh
2026-02-06 19:11 ` [PATCH v2 next 10/11] selftests/nolibc: Increase coverage of printf format tests david.laight.linux
2026-02-16 20:14   ` Thomas Weißschuh
2026-02-16 20:23   ` Thomas Weißschuh
2026-02-16 22:54     ` David Laight
2026-02-18 17:41       ` Thomas Weißschuh
2026-02-06 19:11 ` [PATCH v2 next 11/11] selftests/nolibc: Use printf("%.*s", n, "") to align output david.laight.linux
2026-02-08 16:20   ` Willy Tarreau
2026-02-16 20:22   ` Thomas Weißschuh
2026-02-06 21:36 ` [PATCH v2 next 00/11] tools/nolibc: Enhance printf() David Laight

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=20260207233651.577f0fcb@pumpkin \
    --to=david.laight.linux@gmail.com \
    --cc=lechain@gmail.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux@weissschuh.net \
    --cc=w@1wt.eu \
    /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