All of lore.kernel.org
 help / color / mirror / Atom feed
From: Marcin Slusarz <marcin.slusarz@gmail.com>
To: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: LKML <linux-kernel@vger.kernel.org>,
	linux-fbdev-devel@lists.sourceforge.net,
	Antonino Daplas <adaplas@gmail.com>
Subject: Re: [Linux-fbdev-devel] [PATCH 1/2] vgacon: optimize scrolling
Date: Wed, 24 Sep 2008 21:40:00 +0200	[thread overview]
Message-ID: <20080924193943.GA8544@joi> (raw)
In-Reply-To: <20080923235918.114b5147.krzysztof.h1@poczta.fm>

On Tue, Sep 23, 2008 at 11:59:18PM +0200, Krzysztof Helt wrote:
> On Sun, 21 Sep 2008 23:22:17 +0200
> Marcin Slusarz <marcin.slusarz@gmail.com> wrote:
> 
> > On Sun, Sep 21, 2008 at 08:37:57PM +0200, Krzysztof Helt wrote:
> > > On Sun, 21 Sep 2008 18:00:08 +0200
> > > Marcin Slusarz <marcin.slusarz@gmail.com> wrote:
> > > 
> > > > Join multiple scr_memcpyw into 1-3 calls (usually 2).
> > > > (benchmarked average speedup: 1%)
> > > > 
> > > 
> > > The scr_memcpyw is an inline function so there is no real call here.
> > 
> > ... or it's defined to memcpy
> >  
> > > However, have you tested the patch with scroll range > 2 *  vgacon_scrollback_size?
> > >
> > > It seems that your patch only handles scrolling if it is not bigger than 2 * vgacon_scrollback_size.
> > 
> > No. count is always within range <0, c->vc_rows> (look how it's computed
> > just before this loop). Second copy is needed only when end of screen
> > crosses end/beginning of scrollback (it's a circular buffer).
> > 
> 
> This is not what I asked for. If the vgacon_scrollback_cnt is smaller then c->vc_rows it can happen.
> The answer to my question is that count is always < vgacon_scrollback_cnt (set earlier before 
> checking if it is in the <0,c->vc_rows> range).

Yep

> (...)
> 
> > > > +	if (count) {
> > > > +		int copysize;
> > > > +		count *= c->vc_size_row;
> > > > +		/* how much memory to end of buffer left? */
> > > > +		copysize = min(count, vgacon_scrollback_size - soff);
> > > > +		scr_memcpyw(d, vgacon_scrollback + soff, copysize);
> > > > +		d += copysize;
> > > > +		count -= copysize;
> > > > +		
> > > > +		if (count) {
> > > > +			copysize = min(count, vgacon_scrollback_size);
> 
> This line I got confused by. If the count is always smaller then
> the vgacon_scrollback_cnt before (thus the vgacon_scrollback_size 
> inside the if clause) this line always evaluates to copysize = count here.

You are right. I dropped this "call".
Thanks for a review.

---
Subject: [PATCH 1/2] vgacon: optimize scrolling

Join multiple scr_memcpyw into 1-3 calls (usually 2).
(benchmarked average speedup: 1%)

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: linux-fbdev-devel@lists.sourceforge.net
---
 drivers/video/console/vgacon.c |   27 +++++++++++++++------------
 1 files changed, 15 insertions(+), 12 deletions(-)

diff --git a/drivers/video/console/vgacon.c b/drivers/video/console/vgacon.c
index bd1f57b..29d1209 100644
--- a/drivers/video/console/vgacon.c
+++ b/drivers/video/console/vgacon.c
@@ -292,23 +292,26 @@ static int vgacon_scrolldelta(struct vc_data *c, int lines)
 	d = (void *) c->vc_origin;
 	s = (void *) c->vc_screenbuf;
 
-	while (count--) {
-		scr_memcpyw(d, vgacon_scrollback + soff, c->vc_size_row);
-		d += c->vc_size_row;
-		soff += c->vc_size_row;
-
-		if (soff >= vgacon_scrollback_size)
-			soff = 0;
+	if (count) {
+		int copysize;
+		count *= c->vc_size_row;
+		/* how much memory to end of buffer left? */
+		copysize = min(count, vgacon_scrollback_size - soff);
+		scr_memcpyw(d, vgacon_scrollback + soff, copysize);
+		d += copysize;
+		count -= copysize;
+
+		if (count) {
+			scr_memcpyw(d, vgacon_scrollback, count);
+			d += count;
+		}
 	}
 
 	if (diff == c->vc_rows) {
 		vgacon_cursor(c, CM_MOVE);
 	} else {
-		while (diff--) {
-			scr_memcpyw(d, s, c->vc_size_row);
-			d += c->vc_size_row;
-			s += c->vc_size_row;
-		}
+		if (diff)
+			scr_memcpyw(d, s, diff * c->vc_size_row);
 	}
 
 	return 1;
-- 
1.5.6.4

  parent reply	other threads:[~2008-09-24 19:40 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-09-21 16:00 [PATCH 0/2] vgacon_scrolldelta cleanup Marcin Slusarz
2008-09-21 16:00 ` [PATCH 1/2] vgacon: optimize scrolling Marcin Slusarz
2008-09-21 18:37   ` [Linux-fbdev-devel] " Krzysztof Helt
2008-09-21 21:22     ` Marcin Slusarz
     [not found]       ` <20080923235918.114b5147.krzysztof.h1@poczta.fm>
2008-09-24 19:40         ` Marcin Slusarz [this message]
2008-09-21 16:00 ` [PATCH 2/2] vgacon: vgacon_scrolldelta simplification Marcin Slusarz

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=20080924193943.GA8544@joi \
    --to=marcin.slusarz@gmail.com \
    --cc=adaplas@gmail.com \
    --cc=krzysztof.h1@poczta.fm \
    --cc=linux-fbdev-devel@lists.sourceforge.net \
    --cc=linux-kernel@vger.kernel.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.