linux 2.6.26 vt back_color_erase

linux 2.6.26 vt back_color_erase

[Alexander V. Lukyanov]

Hello!

Current linux kernel has broken bce semantics. Now insert_line and
delete_line clear the new lines with default color and not with the current
background as they should, and what is expected by ncurses.

vc->vc_scrl_erase_char was introduced in 2.6.26 for some reason. I beleave
this change should be backed out.

-- 
   Alexander.

Re: linux 2.6.26 vt back_color_erase

[Alexander V. Lukyanov]

On Fri, Oct 03, 2008 at 12:11:47PM +0400, Alexander V. Lukyanov wrote:
> Current linux kernel has broken bce semantics. Now insert_line and
> delete_line clear the new lines with default color and not with the current
> background as they should, and what is expected by ncurses.
> 
> vc->vc_scrl_erase_char was introduced in 2.6.26 for some reason. I beleave
> this change should be backed out.

I think that the test cases Jan provided could be fixed by temporary
changing erase char when doing cr/lf when vc->vc_need_wrap is true.
There is no need to change erase character for all scrolling operations.

--
   Alexander.

Re: linux 2.6.26 vt back_color_erase

[Jan Engelhardt]

On Friday 2008-10-10 12:49, Alexander V. Lukyanov wrote:

>On Fri, Oct 03, 2008 at 12:11:47PM +0400, Alexander V. Lukyanov wrote:
>> vc->vc_scrl_erase_char was introduced in 2.6.26 for some reason. I beleave
>> this change should be backed out.

There is video_erase_char which contains the background fill and is used
for escape codes such as \e[J and \e[K; scrolling also traditionally used
these, which I consider a visual bug. So scrl_erase_char was introduced
that contains the default color, and which is used for scrolling
(the complement of erase ops like J and K).

>I think that the test cases Jan provided could be fixed by temporary
>changing erase char when doing cr/lf when vc->vc_need_wrap is true.
>There is no need to change erase character for all scrolling operations.

Only in the latter of the following two,
	\e[44mfoo\n\e[42mbar
and
	\e[44moverlylonglinemorethan80charsorso\e[42mbar
vc_need_wrap will be set to 1 during the course of printing it to
the console, but the background fill issue happens with both
test strings.

>> Current linux kernel has broken bce semantics. Now insert_line and
>> delete_line clear the new lines with default color and not with the
>> current background as they should, and what is expected by ncurses.

I beg to differ. The Linux VT now does exactly what an xterm does,
namely that newly inserted lines do not get filled with the
background color. And I consider that the right thing, because it
does not leave ugly color patches lying on your screen when `ls -l`
overruns the screen width, and someone decided to use a background
color for his filenames.

ncurses should not expect anything. Except what is in terminfo.
Do we need a terminfo db update for the "linux" term?

Re: linux 2.6.26 vt back_color_erase

[Alexander V. Lukyanov]

On Fri, Oct 10, 2008 at 01:22:08PM -0400, Jan Engelhardt wrote:
> >> Current linux kernel has broken bce semantics. Now insert_line and
> >> delete_line clear the new lines with default color and not with the
> >> current background as they should, and what is expected by ncurses.
> 
> I beg to differ. The Linux VT now does exactly what an xterm does,

I have just tested again: my xterm does erase new lines with current color,
just as linux console before the change. My xterm version: XTerm(196).
Please test yours and check the version. Also, please test other terminals
with bce terminfo attribute. I have tested putty and it also erases new
lines with default colors. And gnome-terminal too.

> namely that newly inserted lines do not get filled with the
> background color. And I consider that the right thing, because it
> does not leave ugly color patches lying on your screen when `ls -l`
> overruns the screen width, and someone decided to use a background
> color for his filenames.

Probably color ls -l is just broken and should be fixed. Why was it
necessary to change the console color semantics that was there for years?

> ncurses should not expect anything. Except what is in terminfo.
> Do we need a terminfo db update for the "linux" term?

Sure, unless the change is reverted. The bce attribute will need to be
disabled, which is a pity as it saves traffic for color full screen
applications. And changing terminfo is very inconvenient for remote logins.

-- 
   Alexander.

Re: linux 2.6.26 vt back_color_erase

[Jan Engelhardt]

On Monday 2008-10-13 01:24, Alexander V. Lukyanov wrote:

>On Fri, Oct 10, 2008 at 01:22:08PM -0400, Jan Engelhardt wrote:
>> >> Current linux kernel has broken bce semantics. Now insert_line and
>> >> delete_line clear the new lines with default color and not with the
>> >> current background as they should, and what is expected by ncurses.
>> 
>> I beg to differ. The Linux VT now does exactly what an xterm does,
>
>I have just tested again: my xterm does erase new lines with current color,
>just as linux console before the change. My xterm version: XTerm(196).
>Please test yours and check the version.

It seems that xterm-235 does indeed erase-with-current on _scrolling_
(i.e. PS1='$'; echo -en "\e[25B\e[44mfoo\n\e[0m") as do urxvt.

>Probably color ls -l is just broken and should be fixed. Why was it
>necessary to change the console color semantics that was there for years?

Because the Linux VT also did erase-with-current on _linewrap_,
which xterm and urxvt certainly do not do.

Mh, this is really unfortunate; I agree something
needs to be backed out.

Re: linux 2.6.26 vt back_color_erase

[Alan Cox]

> Probably color ls -l is just broken and should be fixed. Why was it
> necessary to change the console color semantics that was there for years?

Submit a patch that reverts the change. I would agree it needs reverting.

> 
> > ncurses should not expect anything. Except what is in terminfo.
> > Do we need a terminfo db update for the "linux" term?
> 
> Sure, unless the change is reverted. The bce attribute will need to be
> disabled, which is a pity as it saves traffic for color full screen
> applications. And changing terminfo is very inconvenient for remote logins.

Agreed

Re: linux 2.6.26 vt back_color_erase

[Alexander V. Lukyanov]

On Mon, Oct 13, 2008 at 09:39:24AM +0100, Alan Cox wrote:
> > Probably color ls -l is just broken and should be fixed. Why was it
> > necessary to change the console color semantics that was there for years?
> 
> Submit a patch that reverts the change. I would agree it needs reverting.

Here is the patch to restore status quo.

diff -ru linux-2.6.27/drivers/char/vt.c linux-2.6.27+/drivers/char/vt.c
--- linux-2.6.27/drivers/char/vt.c	2008-10-10 02:13:53.000000000 +0400
+++ linux-2.6.27+/drivers/char/vt.c	2008-10-14 10:09:10.000000000 +0400
@@ -301,7 +301,7 @@
 	d = (unsigned short *)(vc->vc_origin + vc->vc_size_row * t);
 	s = (unsigned short *)(vc->vc_origin + vc->vc_size_row * (t + nr));
 	scr_memmovew(d, s, (b - t - nr) * vc->vc_size_row);
-	scr_memsetw(d + (b - t - nr) * vc->vc_cols, vc->vc_scrl_erase_char,
+	scr_memsetw(d + (b - t - nr) * vc->vc_cols, vc->vc_video_erase_char,
 		    vc->vc_size_row * nr);
 }
 
@@ -319,7 +319,7 @@
 	s = (unsigned short *)(vc->vc_origin + vc->vc_size_row * t);
 	step = vc->vc_cols * nr;
 	scr_memmovew(s + step, s, (b - t - nr) * vc->vc_size_row);
-	scr_memsetw(s, vc->vc_scrl_erase_char, 2 * step);
+	scr_memsetw(s, vc->vc_video_erase_char, 2 * step);
 }
 
 static void do_update_region(struct vc_data *vc, unsigned long start, int count)
@@ -434,7 +434,6 @@
 	              vc->vc_blink, vc->vc_underline,
 	              vc->vc_reverse ^ vc->vc_decscnm, vc->vc_italic);
 	vc->vc_video_erase_char = (build_attr(vc, vc->vc_color, 1, vc->vc_blink, 0, vc->vc_decscnm, 0) << 8) | ' ';
-	vc->vc_scrl_erase_char = (build_attr(vc, vc->vc_def_color, 1, false, false, vc->vc_decscnm, false) << 8) | ' ';
 }
 
 /* Note: inverting the screen twice should revert to the original state */
diff -ru linux-2.6.27/drivers/video/console/fbcon.c linux-2.6.27+/drivers/video/console/fbcon.c
--- linux-2.6.27/drivers/video/console/fbcon.c	2008-10-10 02:13:53.000000000 +0400
+++ linux-2.6.27+/drivers/video/console/fbcon.c	2008-10-14 10:18:32.000000000 +0400
@@ -1855,7 +1855,6 @@
 	struct fb_info *info = registered_fb[con2fb_map[vc->vc_num]];
 	struct display *p = &fb_display[vc->vc_num];
 	int scroll_partial = info->flags & FBINFO_PARTIAL_PAN_OK;
-	unsigned short saved_ec;
 	int ret;
 
 	if (fbcon_is_inactive(vc, info))
@@ -1869,9 +1868,6 @@
 	 *           whole screen (prevents flicker).
 	 */
 
-	saved_ec = vc->vc_video_erase_char;
-	vc->vc_video_erase_char = vc->vc_scrl_erase_char;
-
 	ret = 0;
 
 	switch (dir) {
@@ -1890,7 +1886,7 @@
 			scr_memsetw((unsigned short *) (vc->vc_origin +
 							vc->vc_size_row *
 							(b - count)),
-				    vc->vc_scrl_erase_char,
+				    vc->vc_video_erase_char,
 				    vc->vc_size_row * count);
 			ret = 1;
 			break;
@@ -1962,7 +1958,7 @@
 			scr_memsetw((unsigned short *) (vc->vc_origin +
 							vc->vc_size_row *
 							(b - count)),
-				    vc->vc_scrl_erase_char,
+				    vc->vc_video_erase_char,
 				    vc->vc_size_row * count);
 			ret = 1;
 			break;
@@ -1982,7 +1978,7 @@
 			scr_memsetw((unsigned short *) (vc->vc_origin +
 							vc->vc_size_row *
 							t),
-				    vc->vc_scrl_erase_char,
+				    vc->vc_video_erase_char,
 				    vc->vc_size_row * count);
 			ret = 1;
 			break;
@@ -2052,14 +2048,13 @@
 			scr_memsetw((unsigned short *) (vc->vc_origin +
 							vc->vc_size_row *
 							t),
-				    vc->vc_scrl_erase_char,
+				    vc->vc_video_erase_char,
 				    vc->vc_size_row * count);
 			ret = 1;
 			break;
 		}
 		break;
 	}
-	vc->vc_video_erase_char = saved_ec;
 	return ret;
 }
 
@@ -2522,9 +2517,6 @@
 			c = vc->vc_video_erase_char;
 			vc->vc_video_erase_char =
 			    ((c & 0xfe00) >> 1) | (c & 0xff);
-			c = vc->vc_scrl_erase_char;
-			vc->vc_scrl_erase_char =
-			    ((c & 0xFE00) >> 1) | (c & 0xFF);
 			vc->vc_attr >>= 1;
 		}
 	} else if (!vc->vc_hi_font_mask && cnt == 512) {
@@ -2555,13 +2547,9 @@
 			if (vc->vc_can_do_color) {
 				vc->vc_video_erase_char =
 				    ((c & 0xff00) << 1) | (c & 0xff);
-				c = vc->vc_scrl_erase_char;
-				vc->vc_scrl_erase_char =
-				    ((c & 0xFF00) << 1) | (c & 0xFF);
 				vc->vc_attr <<= 1;
 			} else {
 				vc->vc_video_erase_char = c & ~0x100;
-				vc->vc_scrl_erase_char = c & ~0x100;
 			}
 		}
 
diff -ru linux-2.6.27/drivers/video/console/mdacon.c linux-2.6.27+/drivers/video/console/mdacon.c
--- linux-2.6.27/drivers/video/console/mdacon.c	2008-10-10 02:13:53.000000000 +0400
+++ linux-2.6.27+/drivers/video/console/mdacon.c	2008-10-14 10:18:06.000000000 +0400
@@ -533,7 +533,7 @@
 
 static int mdacon_scroll(struct vc_data *c, int t, int b, int dir, int lines)
 {
-	u16 eattr = mda_convert_attr(c->vc_scrl_erase_char);
+	u16 eattr = mda_convert_attr(c->vc_video_erase_char);
 
 	if (!lines)
 		return 0;
diff -ru linux-2.6.27/drivers/video/console/sticon.c linux-2.6.27+/drivers/video/console/sticon.c
--- linux-2.6.27/drivers/video/console/sticon.c	2008-10-10 02:13:53.000000000 +0400
+++ linux-2.6.27+/drivers/video/console/sticon.c	2008-10-14 10:17:55.000000000 +0400
@@ -170,12 +170,12 @@
     switch (dir) {
     case SM_UP:
 	sti_bmove(sti, t + count, 0, t, 0, b - t - count, conp->vc_cols);
-	sti_clear(sti, b - count, 0, count, conp->vc_cols, conp->vc_scrl_erase_char);
+	sti_clear(sti, b - count, 0, count, conp->vc_cols, conp->vc_video_erase_char);
 	break;
 
     case SM_DOWN:
 	sti_bmove(sti, t, 0, t + count, 0, b - t - count, conp->vc_cols);
-	sti_clear(sti, t, 0, count, conp->vc_cols, conp->vc_scrl_erase_char);
+	sti_clear(sti, t, 0, count, conp->vc_cols, conp->vc_video_erase_char);
 	break;
     }
 
diff -ru linux-2.6.27/drivers/video/console/vgacon.c linux-2.6.27+/drivers/video/console/vgacon.c
--- linux-2.6.27/drivers/video/console/vgacon.c	2008-10-10 02:13:53.000000000 +0400
+++ linux-2.6.27+/drivers/video/console/vgacon.c	2008-10-14 10:17:38.000000000 +0400
@@ -1350,7 +1350,7 @@
 		} else
 			c->vc_origin += delta;
 		scr_memsetw((u16 *) (c->vc_origin + c->vc_screenbuf_size -
-				     delta), c->vc_scrl_erase_char,
+				     delta), c->vc_video_erase_char,
 			    delta);
 	} else {
 		if (oldo - delta < vga_vram_base) {
@@ -1363,7 +1363,7 @@
 		} else
 			c->vc_origin -= delta;
 		c->vc_scr_end = c->vc_origin + c->vc_screenbuf_size;
-		scr_memsetw((u16 *) (c->vc_origin), c->vc_scrl_erase_char,
+		scr_memsetw((u16 *) (c->vc_origin), c->vc_video_erase_char,
 			    delta);
 	}
 	c->vc_scr_end = c->vc_origin + c->vc_screenbuf_size;
diff -ru linux-2.6.27/include/linux/console_struct.h linux-2.6.27+/include/linux/console_struct.h
--- linux-2.6.27/include/linux/console_struct.h	2008-10-10 02:13:53.000000000 +0400
+++ linux-2.6.27+/include/linux/console_struct.h	2008-10-14 10:07:42.000000000 +0400
@@ -53,7 +53,6 @@
 	unsigned short	vc_hi_font_mask;	/* [#] Attribute set for upper 256 chars of font or 0 if not supported */
 	struct console_font vc_font;		/* Current VC font set */
 	unsigned short	vc_video_erase_char;	/* Background erase character */
-	unsigned short	vc_scrl_erase_char;	/* Erase character for scroll */
 	/* VT terminal data */
 	unsigned int	vc_state;		/* Escape sequence parser state */
 	unsigned int	vc_npar,vc_par[NPAR];	/* Parameters of current escape sequence */

-- 
   Alexander.

Re: linux 2.6.26 vt back_color_erase

[Jan Engelhardt]

On Tuesday 2008-10-14 04:16, Alexander V. Lukyanov wrote:

>On Mon, Oct 13, 2008 at 09:39:24AM +0100, Alan Cox wrote:
>> > Probably color ls -l is just broken and should be fixed. Why was it
>> > necessary to change the console color semantics that was there for years?
>> 
>> Submit a patch that reverts the change. I would agree it needs reverting.
>
>Here is the patch to restore status quo.

Tested-by: Jan Engelhardt <jengelh@medozas.de>
Acked-by: Jan Engelhardt <jengelh@medozas.de>

I think this should be documented somewhere though
(perhaps right in the scrup() function in vt.c)
so that no one else tries to change it.

	/*
	 * A linewrap in itself does not cause the next line to be 
	 * erased with video_erase_char. Only when the terminal scrolls 
	 * (possibly as part of a linewrap), this is done.
	 * Other terminal emulators like xterm do the same.
	 * Note that applications rely on this erase-on-scroll behavior.
	 */

>diff -ru linux-2.6.27/drivers/char/vt.c linux-2.6.27+/drivers/char/vt.c
>--- linux-2.6.27/drivers/char/vt.c	2008-10-10 02:13:53.000000000 +0400
>+++ linux-2.6.27+/drivers/char/vt.c	2008-10-14 10:09:10.000000000 +0400
>@@ -301,7 +301,7 @@
> 	d = (unsigned short *)(vc->vc_origin + vc->vc_size_row * t);
> 	s = (unsigned short *)(vc->vc_origin + vc->vc_size_row * (t + nr));
> 	scr_memmovew(d, s, (b - t - nr) * vc->vc_size_row);
>-	scr_memsetw(d + (b - t - nr) * vc->vc_cols, vc->vc_scrl_erase_char,
>+	scr_memsetw(d + (b - t - nr) * vc->vc_cols, vc->vc_video_erase_char,
> 		    vc->vc_size_row * nr);
> }
> 
>@@ -319,7 +319,7 @@
> 	s = (unsigned short *)(vc->vc_origin + vc->vc_size_row * t);
> 	step = vc->vc_cols * nr;
> 	scr_memmovew(s + step, s, (b - t - nr) * vc->vc_size_row);
>-	scr_memsetw(s, vc->vc_scrl_erase_char, 2 * step);
>+	scr_memsetw(s, vc->vc_video_erase_char, 2 * step);
> }
> 
> static void do_update_region(struct vc_data *vc, unsigned long start, int count)
>@@ -434,7 +434,6 @@
> 	              vc->vc_blink, vc->vc_underline,
> 	              vc->vc_reverse ^ vc->vc_decscnm, vc->vc_italic);
> 	vc->vc_video_erase_char = (build_attr(vc, vc->vc_color, 1, vc->vc_blink, 0, vc->vc_decscnm, 0) << 8) | ' ';
>-	vc->vc_scrl_erase_char = (build_attr(vc, vc->vc_def_color, 1, false, false, vc->vc_decscnm, false) << 8) | ' ';
> }
> 
> /* Note: inverting the screen twice should revert to the original state */
>diff -ru linux-2.6.27/drivers/video/console/fbcon.c linux-2.6.27+/drivers/video/console/fbcon.c
>--- linux-2.6.27/drivers/video/console/fbcon.c	2008-10-10 02:13:53.000000000 +0400
>+++ linux-2.6.27+/drivers/video/console/fbcon.c	2008-10-14 10:18:32.000000000 +0400
>@@ -1855,7 +1855,6 @@
> 	struct fb_info *info = registered_fb[con2fb_map[vc->vc_num]];
> 	struct display *p = &fb_display[vc->vc_num];
> 	int scroll_partial = info->flags & FBINFO_PARTIAL_PAN_OK;
>-	unsigned short saved_ec;
> 	int ret;
> 
> 	if (fbcon_is_inactive(vc, info))
>@@ -1869,9 +1868,6 @@
> 	 *           whole screen (prevents flicker).
> 	 */
> 
>-	saved_ec = vc->vc_video_erase_char;
>-	vc->vc_video_erase_char = vc->vc_scrl_erase_char;
>-
> 	ret = 0;
> 
> 	switch (dir) {
>@@ -1890,7 +1886,7 @@
> 			scr_memsetw((unsigned short *) (vc->vc_origin +
> 							vc->vc_size_row *
> 							(b - count)),
>-				    vc->vc_scrl_erase_char,
>+				    vc->vc_video_erase_char,
> 				    vc->vc_size_row * count);
> 			ret = 1;
> 			break;
>@@ -1962,7 +1958,7 @@
> 			scr_memsetw((unsigned short *) (vc->vc_origin +
> 							vc->vc_size_row *
> 							(b - count)),
>-				    vc->vc_scrl_erase_char,
>+				    vc->vc_video_erase_char,
> 				    vc->vc_size_row * count);
> 			ret = 1;
> 			break;
>@@ -1982,7 +1978,7 @@
> 			scr_memsetw((unsigned short *) (vc->vc_origin +
> 							vc->vc_size_row *
> 							t),
>-				    vc->vc_scrl_erase_char,
>+				    vc->vc_video_erase_char,
> 				    vc->vc_size_row * count);
> 			ret = 1;
> 			break;
>@@ -2052,14 +2048,13 @@
> 			scr_memsetw((unsigned short *) (vc->vc_origin +
> 							vc->vc_size_row *
> 							t),
>-				    vc->vc_scrl_erase_char,
>+				    vc->vc_video_erase_char,
> 				    vc->vc_size_row * count);
> 			ret = 1;
> 			break;
> 		}
> 		break;
> 	}
>-	vc->vc_video_erase_char = saved_ec;
> 	return ret;
> }
> 
>@@ -2522,9 +2517,6 @@
> 			c = vc->vc_video_erase_char;
> 			vc->vc_video_erase_char =
> 			    ((c & 0xfe00) >> 1) | (c & 0xff);
>-			c = vc->vc_scrl_erase_char;
>-			vc->vc_scrl_erase_char =
>-			    ((c & 0xFE00) >> 1) | (c & 0xFF);
> 			vc->vc_attr >>= 1;
> 		}
> 	} else if (!vc->vc_hi_font_mask && cnt == 512) {
>@@ -2555,13 +2547,9 @@
> 			if (vc->vc_can_do_color) {
> 				vc->vc_video_erase_char =
> 				    ((c & 0xff00) << 1) | (c & 0xff);
>-				c = vc->vc_scrl_erase_char;
>-				vc->vc_scrl_erase_char =
>-				    ((c & 0xFF00) << 1) | (c & 0xFF);
> 				vc->vc_attr <<= 1;
> 			} else {
> 				vc->vc_video_erase_char = c & ~0x100;
>-				vc->vc_scrl_erase_char = c & ~0x100;
> 			}
> 		}
> 
>diff -ru linux-2.6.27/drivers/video/console/mdacon.c linux-2.6.27+/drivers/video/console/mdacon.c
>--- linux-2.6.27/drivers/video/console/mdacon.c	2008-10-10 02:13:53.000000000 +0400
>+++ linux-2.6.27+/drivers/video/console/mdacon.c	2008-10-14 10:18:06.000000000 +0400
>@@ -533,7 +533,7 @@
> 
> static int mdacon_scroll(struct vc_data *c, int t, int b, int dir, int lines)
> {
>-	u16 eattr = mda_convert_attr(c->vc_scrl_erase_char);
>+	u16 eattr = mda_convert_attr(c->vc_video_erase_char);
> 
> 	if (!lines)
> 		return 0;
>diff -ru linux-2.6.27/drivers/video/console/sticon.c linux-2.6.27+/drivers/video/console/sticon.c
>--- linux-2.6.27/drivers/video/console/sticon.c	2008-10-10 02:13:53.000000000 +0400
>+++ linux-2.6.27+/drivers/video/console/sticon.c	2008-10-14 10:17:55.000000000 +0400
>@@ -170,12 +170,12 @@
>     switch (dir) {
>     case SM_UP:
> 	sti_bmove(sti, t + count, 0, t, 0, b - t - count, conp->vc_cols);
>-	sti_clear(sti, b - count, 0, count, conp->vc_cols, conp->vc_scrl_erase_char);
>+	sti_clear(sti, b - count, 0, count, conp->vc_cols, conp->vc_video_erase_char);
> 	break;
> 
>     case SM_DOWN:
> 	sti_bmove(sti, t, 0, t + count, 0, b - t - count, conp->vc_cols);
>-	sti_clear(sti, t, 0, count, conp->vc_cols, conp->vc_scrl_erase_char);
>+	sti_clear(sti, t, 0, count, conp->vc_cols, conp->vc_video_erase_char);
> 	break;
>     }
> 
>diff -ru linux-2.6.27/drivers/video/console/vgacon.c linux-2.6.27+/drivers/video/console/vgacon.c
>--- linux-2.6.27/drivers/video/console/vgacon.c	2008-10-10 02:13:53.000000000 +0400
>+++ linux-2.6.27+/drivers/video/console/vgacon.c	2008-10-14 10:17:38.000000000 +0400
>@@ -1350,7 +1350,7 @@
> 		} else
> 			c->vc_origin += delta;
> 		scr_memsetw((u16 *) (c->vc_origin + c->vc_screenbuf_size -
>-				     delta), c->vc_scrl_erase_char,
>+				     delta), c->vc_video_erase_char,
> 			    delta);
> 	} else {
> 		if (oldo - delta < vga_vram_base) {
>@@ -1363,7 +1363,7 @@
> 		} else
> 			c->vc_origin -= delta;
> 		c->vc_scr_end = c->vc_origin + c->vc_screenbuf_size;
>-		scr_memsetw((u16 *) (c->vc_origin), c->vc_scrl_erase_char,
>+		scr_memsetw((u16 *) (c->vc_origin), c->vc_video_erase_char,
> 			    delta);
> 	}
> 	c->vc_scr_end = c->vc_origin + c->vc_screenbuf_size;
>diff -ru linux-2.6.27/include/linux/console_struct.h linux-2.6.27+/include/linux/console_struct.h
>--- linux-2.6.27/include/linux/console_struct.h	2008-10-10 02:13:53.000000000 +0400
>+++ linux-2.6.27+/include/linux/console_struct.h	2008-10-14 10:07:42.000000000 +0400
>@@ -53,7 +53,6 @@
> 	unsigned short	vc_hi_font_mask;	/* [#] Attribute set for upper 256 chars of font or 0 if not supported */
> 	struct console_font vc_font;		/* Current VC font set */
> 	unsigned short	vc_video_erase_char;	/* Background erase character */
>-	unsigned short	vc_scrl_erase_char;	/* Erase character for scroll */
> 	/* VT terminal data */
> 	unsigned int	vc_state;		/* Escape sequence parser state */
> 	unsigned int	vc_npar,vc_par[NPAR];	/* Parameters of current escape sequence */
>
>-- 
>   Alexander.
>