2.6.27-rc1: softcursor behaviour changed

2.6.27-rc1: softcursor behaviour changed

[Pavel Machek]

Hi!

In 2.6.26, this 

    echo -e "\33[10;5000]\33[11;50]\33[?18;0;136c\33[?102m"

produces some variation on cursor. on 2.6.27-rc1, I have no cursor at
all :-(.
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 2.6.27-rc1: softcursor behaviour changed

[Pavel Machek]

Hi!

> In 2.6.26, this 
> 
>     echo -e "\33[10;5000]\33[11;50]\33[?18;0;136c\33[?102m"
> 
> produces some variation on cursor. on 2.6.27-rc1, I have no cursor at
> all :-(.

The patch that changed it is below... and I must say I liked the old
behaviour better. Unlike blinking, high-brightness background is
actually useful. Can we get that one reverted?

									Pavel

commit 2d04a4a72d7e1519b4838f24bdd4b5d0f3f426dc
tree 2e6278cbc64a338824511c19ad38c7e289fa7112
parent 4a25e41831ee851c1365d8b41decc22493b18e6d
author Stefano Stabellini <stefano.stabellini@eu.citrix.com> Wed, 23
Jul 2008 21:31:48 -0700
committer Linus Torvalds <torvalds@linux-foundation.org> Thu, 24 Jul
2008 10:47:41 -0700

    fbcon: bgcolor fix

    The fourth bit of the background color is the blink property bit,
not the
    intensity bit, as for the foreground color.  Therefore it
shouldn't be
    included in the background color.

    Signed-off-by: Stefano Stabellini
<stefano.stabellini@eu.citrix.com>
    Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
    Cc: "Antonino A. Daplas" <adaplas@pol.net>
    Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
    Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>


-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 2.6.27-rc1: softcursor behaviour changed

[Stefano Stabellini]

Pavel Machek wrote:

> Hi!
> 
>> In 2.6.26, this 
>>
>>     echo -e "\33[10;5000]\33[11;50]\33[?18;0;136c\33[?102m"
>>
>> produces some variation on cursor. on 2.6.27-rc1, I have no cursor at
>> all :-(.
> 
> The patch that changed it is below... and I must say I liked the old
> behaviour better. Unlike blinking, high-brightness background is
> actually useful. Can we get that one reverted?


The problem is not a matter of taste but a matter of correctness: the
build_attr function in drivers/char/vt.c stores the blink attribute as
the forth bit of the background color.
Then in fbcon we are interpreting the same forth bit of the background
color as intensity bit.
My patch fixes that.

Re: 2.6.27-rc1: softcursor behaviour changed

[Pavel Machek]

On Thu 2008-08-07 12:17:21, Stefano Stabellini wrote:
> Pavel Machek wrote:
> 
> > Hi!
> > 
> >> In 2.6.26, this 
> >>
> >>     echo -e "\33[10;5000]\33[11;50]\33[?18;0;136c\33[?102m"
> >>
> >> produces some variation on cursor. on 2.6.27-rc1, I have no cursor at
> >> all :-(.
> > 
> > The patch that changed it is below... and I must say I liked the old
> > behaviour better. Unlike blinking, high-brightness background is
> > actually useful. Can we get that one reverted?
> 
> 
> The problem is not a matter of taste but a matter of correctness: the
> build_attr function in drivers/char/vt.c stores the blink attribute as
> the forth bit of the background color.
> Then in fbcon we are interpreting the same forth bit of the background
> color as intensity bit.

...which works pretty well. We can't reasonably blink in fbcon, so we
highlight it with bright background, instead. Even old VGA had option
to do that.

> My patch fixes that.

AFAICS high bit is now completely ignored now. That makes it
useless...

For example I can no longer see my cursor.

So, if 2.6.26 behaviour is "incorrect", can you show example where it
matters, and how your patch "helps"?

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 2.6.27-rc1: softcursor behaviour changed

[Stefano Stabellini]

Pavel Machek wrote:

>> The problem is not a matter of taste but a matter of correctness: the
>> build_attr function in drivers/char/vt.c stores the blink attribute as
>> the forth bit of the background color.
>> Then in fbcon we are interpreting the same forth bit of the background
>> color as intensity bit.
> 
> ...which works pretty well. We can't reasonably blink in fbcon, so we
> highlight it with bright background, instead. Even old VGA had option
> to do that.
> 
>> My patch fixes that.
> 
> AFAICS high bit is now completely ignored now. That makes it
> useless...

- an application wants to do X
- I don't like X and X is not even implemented here
- let's do Y instead!

I think this is not the correct way of doing things, we could implement
blinking in fbcon instead.
In fact I posted another patch a while ago to implement blinking in
fbcon, but it wasn't accepted, not sure why yet :)
Maybe I'll try to talk with the fbcon devs to understand the problem
better, and reasonably soon you'll be able to see your blinking cursor
in fbcon.

> For example I can no longer see my cursor.
> 
> So, if 2.6.26 behaviour is "incorrect", can you show example where it
> matters, and how your patch "helps"?
> 


We have some ncurses applications that print a blinking white on blue
warning message.
Without the patch the message is static and with a pink background
(applications can change color settings, if they do that only for the
normal colors and not for the intensive colors because they think they
are not using them, this is what happens).

Re: 2.6.27-rc1: softcursor behaviour changed

[Linus Torvalds]

On Fri, 8 Aug 2008, Stefano Stabellini wrote:
> 
> - an application wants to do X
> - I don't like X and X is not even implemented here
> - let's do Y instead!

Well, at the same time:
 - Y is the closest we can get in practice
 - Y is what we've done before (but I don't know how long)

So yeah, I think we should probably revert it.

		Linus