Prev: [PATCH v2 2/3] kernel/power/block_io.c: do not use end_swap_bio_read
Next: [PATCH 4/6] Kill unused header
From: Robert Emanuele on 5 Jun 2010 19:30
Greetings lkml and Jaya,

I've been using the broadsheetfb graphics driver for a while now.
I've been trying to get it working with the fbcon console driver. Has
anyone had any success with that?

While it "works", its performance is unusable as a console. It
redraws the whole screen every new line, which may be acceptable. A
real issue is when it tries to scroll the screen, it redraws each line
partially and then does a full screen update with each partial line.
It seems the driver is capable from Jaya's video at
http://www.youtube.com/watch?v=0Bny6qyRDWw

Are those kind of updates a limit of the broadsheet driver or fbcon?
Or more hopefully, is there a configuration change to improve this?

Thanks for your help,

Rob
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Konrad Rzeszutek Wilk on 7 Jun 2010 12:30
On Sat, Jun 05, 2010 at 04:21:08PM -0700, Robert Emanuele wrote:
> Greetings lkml and Jaya,
>
> I've been using the broadsheetfb graphics driver for a while now.
> I've been trying to get it working with the fbcon console driver. Has
> anyone had any success with that?
>
> While it "works", its performance is unusable as a console. It
> redraws the whole screen every new line, which may be acceptable. A
> real issue is when it tries to scroll the screen, it redraws each line
> partially and then does a full screen update with each partial line.
> It seems the driver is capable from Jaya's video at
> http://www.youtube.com/watch?v=0Bny6qyRDWw
>
> Are those kind of updates a limit of the broadsheet driver or fbcon?

I thought there was an update value on how often it should redraw. Did
you do a modinfo on the driver?
> Or more hopefully, is there a configuration change to improve this?
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Robert Emanuele on 7 Jun 2010 13:20
There are no module parameters for fbcon or broadsheetfb.

On Mon, Jun 7, 2010 at 9:28 AM, Konrad Rzeszutek Wilk
<konrad.wilk(a)oracle.com> wrote:
> On Sat, Jun 05, 2010 at 04:21:08PM -0700, Robert Emanuele wrote:
>> Greetings lkml and Jaya,
>>
>> I've been using the broadsheetfb graphics driver for a while now.
>> I've been trying to get it working with the fbcon console driver. �Has
>> anyone had any success with that?
>>
>> While it "works", its performance is unusable as a console. �It
>> redraws the whole screen every new line, which may be acceptable. �A
>> real issue is when it tries to scroll the screen, it redraws each line
>> partially and then does a full screen update with each partial line.
>> It seems the driver is capable from Jaya's video at
>> http://www.youtube.com/watch?v=0Bny6qyRDWw
>>
>> Are those kind of updates a limit of the broadsheet driver or fbcon?
>
> I thought there was an update value on how often it should redraw. Did
> you do a modinfo on the driver?
>> Or more hopefully, is there a configuration change to improve this?
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Jaya Kumar on 7 Jun 2010 20:30
On Sun, Jun 6, 2010 at 7:21 AM, Robert Emanuele <rob(a)emanuele.us> wrote:
> real issue is when it tries to scroll the screen, it redraws each line
> partially and then does a full screen update with each partial line.

Sounds like what to expect because fbcon is using the slow path with
synchronous updates. It should be relatively straightforward to
improve that by reusing the deferred handler.

> It seems the driver is capable from Jaya's video at
> http://www.youtube.com/watch?v=0Bny6qyRDWw
>
> Are those kind of updates a limit of the broadsheet driver or fbcon?

That demo was horribly slow because it was host limited by the
xres*yres*16 gpio calls. That video was made before using batch
gpiolib support, which is a patchset that's out of tree,
http://thread.gmane.org/gmane.linux.fbdev.devel/13285 .

I had better results here, http://www.youtube.com/watch?v=aoG7XHO7P0s
and I was able to support 8Hz video,
http://www.youtube.com/watch?v=i4Loe5rIu4I (not display limited but
host limited). I had posted all of the code for that for review on
fbdev a while back, but I decided against getting it merged because it
wasn't elegant (needed userspace changes) and I think we can do
better. I'm trying to find time/funding to make those improvements.

Thanks,
jaya
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Konrad Rzeszutek Wilk on 8 Jun 2010 09:30
On Mon, Jun 07, 2010 at 10:19:27AM -0700, Robert Emanuele wrote:
> There are no module parameters for fbcon or broadsheetfb.
>
> On Mon, Jun 7, 2010 at 9:28 AM, Konrad Rzeszutek Wilk
> >> Are those kind of updates a limit of the broadsheet driver or fbcon?
> >
> > I thought there was an update value on how often it should redraw. Did
> > you do a modinfo on the driver?

Please do not post your response on top.

What I was thinking about is the delay variable:

1056 static struct fb_deferred_io broadsheetfb_defio = {
1057 .delay = HZ/4,
1058 .deferred_io = broadsheetfb_dpy_deferred_io,
1059 };
1060


Try changing that to a higher number and seeing if that fixes your
problem. If so, you and I can work on a patch that sets a parameter
for this.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 |  Next  |  Last
Pages: 1 2
Prev: [PATCH v2 2/3] kernel/power/block_io.c: do not use end_swap_bio_read
Next: [PATCH 4/6] Kill unused header