Re: [RFC] Get selection to buffer addition

Re: [RFC] Get selection to buffer addition

[Pavel Machek]

Hi!

> I'm sending this mail again with the patch in plain text and not
> gzip+uuencoded, sorry for any inconvenience.
> 
> What I'm looking for:
>   I'm looking for comments and approval for a small addition to the console
> driver (drivers/char/console.c).
> 
> Small description:
>   The included patches adds a couple of new services to the TIOCLINUX ioctl
> call, they are:
> 
> 13 (get selection into a buffer): It copies the contents of the selection
> buffer (maintained in kernel space) into a user space provided buffer. Is
> something like "paste to a buffer"  instead of just paste to the current
> console.
> 
> 14 (get selection length): Returns the length of the selection buffer (0 if
> none selected).

Looks good to me. Now, all I want is utility to share clipboard between
X and text console...
								Pavel
-- 
Philips Velo 1: 1"x4"x8", 300gram, 60, 12MB, 40bogomips, linux, mutt,
details at http://atrey.karlin.mff.cuni.cz/~pavel/velo/index.html.

[RFC] Get selection to buffer addition

[salvador]

Hi all:

What I'm looking for:
  I'm looking for comments and approval for a small addition to the console
driver (drivers/char/console.c).

Small description:
  The included patches adds a couple of new services to the TIOCLINUX ioctl
call, they are:

13 (get selection into a buffer): It copies the contents of the selection
buffer (maintained in kernel space) into a user space provided buffer. Is
something like "paste to a buffer"  instead of just paste to the current
console.

14 (get selection length): Returns the length of the selection buffer (0 if
none selected).
-----------

The included file (compressed and uuencoded) contains diffs that apply
without offsets into 2.4.7 kernel, they are valid for 2.2.18 too (with
offset).
I also include a simple test program that calls the new services. The test
program is really simple and assumes you can open /dev/tty1 (you are
there or you are root ;-)

Additional question: Why the switch uses numbers? Shouldn't these values be
defined as something like TIOCLINUX_GET_SELECTION in a header?

What's the purpose? Just allow pasting in text editors using menues or
without the mouse middle button and avoiding to missinterpret the pasted
text as keystrokes.

SET

begin 664 patchConsole-2.4.7.txt.gz
M'XL("/K,83L"`W!A=&-H0V]N<V]L92TR+C0N-RYT>'0`E57?C]HX$'Z&OV*D
MOA!"0F!9V++M'JM3M^VINH?>MB_5*0K)D%@--N<?['+5_N\W=K*0](+:(@*Q
M/?.-9[[/GB`(H&3</`;3<!8N0B%9/LXDVZ-4X[1(Y#@57(D2P[1W;Q#^,!PF
M4YA<+B<+^L(TBB9]W_>;(&?]"T/^)4SG0,X7%\O9O/)?K2"8S$9S\.GW"E:K
M/L#0/O"[V!T@X1GL$J41-H:GF@D.ZP/<\DSB`[Q+#B5J/7+F`$IL$9`7"4]Q
MBUPK9UJB4H0B!7PT:\99V/>==8X:%)98@3*N!22P-IL-2A>U1)[K@B!J>_OY
M*RGW228DO`GA7HH=)O!*U7,K@F!A+O9A(F\:260(!)T6F'Z%#;EFS(:@[<&>
M92B"-)&9@JU0NCS8#;]-*#*\,[S.ZU4>%#18(89&)T6(F;EQ59M.(U<W^I^/
M)I$K7:_7DZB-Y!!=VU&:*"3.EKWQT"6\R>.:$QB.F^:GA>N^_^QW<?0[5R@"
M\4\@9!D?+>/*9)#(W&M@SCHPZU*WP79&QT810!NULAUX(Q@8KEC.,;-[@J$W
M&%C-T0M%!!\FGF=+\$1/C1B\>?_GY]L/-/O4#WZH_6-(4N^=9'"':X"7,)DO
M+Z+E=/H3ZF\BM/5_V=;_].K*$4E_LTG%8X-%RJ#OC^OSH`L$(DH[>8N-&Y\*
M6;-2<>1T:RL(:I>D6"]6\G_/]^(K58[TQD2JRX'GYHD`6\M.'JT\-!QK7@J>
M@^6V[W\CVJK29ZBTY;K%3,68G:8\>FP#CE%';;74065%H6+_HM@X6CWZ6'4<
MJ;R[_?3AWH&"A;61X34\2Z`#X#1L!O.<-NM]U3J\L16MTXY+7;C`]=KK[]8:
MWJ<%#VQ)W&1*I,5:5.G:38X:`*.Z-E5J';D]`99T:+ZU-A"Y%7KB^'A&_E](
MQ\^92EJ`D\!\PG(">TN'LE-/%?J/E-/BL//0[@7+:KW4X4^U@-^^JRPLCYN#
ML8U+>>I?T#^MN/OS'X/4NVI3K:FE*"52EFC:Z0.C.`[22'<E/S<LY]F5:N>]
MP7A:F@S';J%Q[`MW<;AC']F+@X[]Y;S[XC@/T;XY9M$R6C0ZY\)USD7=.?%1
MHW1WM"MF*9*L-/KLT;UN>6P%22F6N!-2,YY7;)U,[+!E,U!:FE3;FL;UZ]`.
M1@YM;:@Q5[CR\?GE8*77"/E+%\W)]2>51GOOPXL,-XQCU6QC;K;4YTJSY:HW
FV*>NYWVQY-N7O\,LN'&3I?*Z/(D<5+WSCE(\D.-_G[[YF5<)````
`
end
-------------------------------------------
Test program:

#include <stdio.h>
#include <sys/ioctl.h>
#include <fcntl.h>
#include <unistd.h>
#include <stdlib.h>

int main(int argc, char *argv[])
{
 int i,res,fd,largo;
 char buffer[sizeof(unsigned int)+2],*s;
 unsigned int *lg=(unsigned int *)(buffer+1);

 fd=open("/dev/tty1",O_RDWR);

 printf("Finding the length of the selection:\n");
 buffer[0]=14;
 res=ioctl(fd,TIOCLINUX,&buffer);
 printf("The iotcl returns %d\n",res);
 printf("Length: %u\n",*lg);
 largo=*lg;
 if (*lg==0) return 0;

 printf("Asking for the selection:\n");
 s=(char *)malloc(largo+sizeof(unsigned int)+2);
 s[0]=13;
 lg=(unsigned int *)(s+1);
 *lg=largo;
 res=ioctl(fd,TIOCLINUX,s);
 printf("The iotcl returns %d\n",res);
 largo=*lg;
 printf("Length returned: %d\n",largo);
 s+=1+sizeof(unsigned int);
 s[largo]=0;

 for (i=0; s[i]; i++)
     if (s[i]==13) s[i]=10;
 printf("Selection: '%s' (%d)\n",s,s[0]);
 return 0;
}


--
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: set-soft@bigfoot.com set@computer.org
                    set@ieee.org
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

[RFC] Get selection to buffer addition

[salvador]

Hi all:

I'm sending this mail again with the patch in plain text and not
gzip+uuencoded, sorry for any inconvenience.

What I'm looking for:
  I'm looking for comments and approval for a small addition to the console
driver (drivers/char/console.c).

Small description:
  The included patches adds a couple of new services to the TIOCLINUX ioctl
call, they are:

13 (get selection into a buffer): It copies the contents of the selection
buffer (maintained in kernel space) into a user space provided buffer. Is
something like "paste to a buffer"  instead of just paste to the current
console.

14 (get selection length): Returns the length of the selection buffer (0 if
none selected).
-----------

The included file contains diffs that apply
without offsets into 2.4.7 kernel, they are valid for 2.2.18 too (with
offset).
I also include a simple test program that calls the new services. The test
program is really simple and assumes you can open /dev/tty1 (you are
there or you are root ;-)

Additional question: Why the switch uses numbers? Shouldn't these values be
defined as something like TIOCLINUX_GET_SELECTION in a header?

What's the purpose? Just allow pasting in text editors using menues or
without the mouse middle button and avoiding to missinterpret the pasted
text as keystrokes.

SET

Patch (about 77 lines):
-------------------------------------------
--- linux-2.4.7.orig/drivers/char/console.c     Tue Jun 12 15:17:17 2001
+++ linux-2.4.7/drivers/char/console.c  Thu Jul 26 17:33:46 2001
@@ -14,6 +14,8 @@
  *
  * Copy and paste function by Andrew Haylett,
  *   some enhancements by Alessandro Rubini.
+ *   get selection into a buffer and length by
+ *       Salvador E. Tropea <salvador@inti.gov.ar>
  *
  * Code to check for different video-cards mostly by Galen Hunt,
  * <g-hunt@ee.utah.edu>
@@ -2204,6 +2206,10 @@
                        return 0;
                case 12:        /* get fg_console */
                        return fg_console;
+               case 13:        /* get selection into a buffer */
+                       return get_selection_buffer(arg);
+               case 14:        /* get selection length */
+                       return put_user(get_selection_length(), (unsigned int
*)((char *)arg + 1));
        }
        return -EINVAL;
 }
--- linux-2.4.7.orig/drivers/char/selection.c   Fri Feb  9 16:30:22 2001
+++ linux-2.4.7/drivers/char/selection.c        Thu Jul 26 17:53:46 2001
@@ -288,6 +288,41 @@
        return 0;
 }

+/* Copy the contents of the selection buffer into a
+ * user space buffer.
+ * Invoked by ioctl().
+ */
+int get_selection_buffer(const unsigned long arg)
+{
+       char *dest;
+       unsigned int length;
+
+       if (get_user(length, (unsigned int *)(arg + sizeof(char))))
+               return -EFAULT;
+
+       dest = (char *)(arg + sizeof(char) + sizeof(unsigned int));
+
+       if (length > sel_buffer_lth)
+               length = sel_buffer_lth;
+
+       if (sel_buffer) {
+               if (copy_to_user(dest, sel_buffer, length))
+                       return -EFAULT;
+       } else {
+               length = 0;
+       }
+       __put_user(length, (unsigned long *)(arg + sizeof(char)));
+       return 0;
+}
+
+/* Get the selection buffer length.
+ * Invoked by ioctl().
+ */
+unsigned int get_selection_length(void)
+{
+       return sel_buffer ? sel_buffer_lth : 0;
+}
+
 /* Insert the contents of the selection buffer into the
  * queue of the tty associated with the current console.
  * Invoked by ioctl().
--- linux-2.4.7.orig/include/linux/selection.h  Fri Jul 20 16:53:56 2001
+++ linux-2.4.7/include/linux/selection.h       Thu Jul 26 17:40:07 2001
@@ -17,6 +17,8 @@
 extern int sel_loadlut(const unsigned long arg);
 extern int mouse_reporting(void);
 extern void mouse_report(struct tty_struct * tty, int butt, int mrx, int
mry);
+extern int get_selection_buffer(const unsigned long arg);
+extern unsigned int get_selection_length(void);

 #define video_num_columns      (vc_cons[currcons].d->vc_cols)
 #define video_num_lines                (vc_cons[currcons].d->vc_rows)
-------------------------------------------
End of patch.

Test program:
-------------------------------------------
#include <stdio.h>
#include <sys/ioctl.h>
#include <fcntl.h>
#include <unistd.h>
#include <stdlib.h>

int main(int argc, char *argv[])
{
 int i,res,fd,largo;
 char buffer[sizeof(unsigned int)+2],*s;
 unsigned int *lg=(unsigned int *)(buffer+1);

 fd=open("/dev/tty1",O_RDWR);

 printf("Finding the length of the selection:\n");
 buffer[0]=14;
 res=ioctl(fd,TIOCLINUX,&buffer);
 printf("The iotcl returns %d\n",res);
 printf("Length: %u\n",*lg);
 largo=*lg;
 if (*lg==0) return 0;

 printf("Asking for the selection:\n");
 s=(char *)malloc(largo+sizeof(unsigned int)+2);
 s[0]=13;
 lg=(unsigned int *)(s+1);
 *lg=largo;
 res=ioctl(fd,TIOCLINUX,s);
 printf("The iotcl returns %d\n",res);
 largo=*lg;
 printf("Length returned: %d\n",largo);
 s+=1+sizeof(unsigned int);
 s[largo]=0;

 for (i=0; s[i]; i++)
     if (s[i]==13) s[i]=10;
 printf("Selection: '%s' (%d)\n",s,s[0]);
 return 0;
}

--
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: set-soft@bigfoot.com set@computer.org
                    set@ieee.org
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: [RFC] Get selection to buffer addition

[James Simmons]

I plan to make selection go away in 2.5.X. It should be done in userland.
When I get time I will rework gpm.

-----------------------------------------------------------------------------
Crap can work. Given enough thrust pigs will fly, but it's not necessary a
good idea.                                 [ Alexander Viro on linux-kernel ]

Re: [RFC] Get selection to buffer addition

[salvador]

James Simmons wrote:

> I plan to make selection go away in 2.5.X. It should be done in userland.
> When I get time I will rework gpm.

Hmmm... ok, it changes all the thing. So I doubt the patch will be accepted. In
fact I started looking gpm sources when I wanted to do it ;-)
I would like to know when you start with it and I could contribute some
ideas/code.

SET

--
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: set-soft@bigfoot.com set@computer.org
                    set@ieee.org
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013

Re: [RFC] Get selection to buffer addition

[James Simmons]

> > I plan to make selection go away in 2.5.X. It should be done in userland.
> > When I get time I will rework gpm.
> 
> Hmmm... ok, it changes all the thing. So I doubt the patch will be accepted. In
> fact I started looking gpm sources when I wanted to do it ;-)
> I would like to know when you start with it and I could contribute some
> ideas/code.

Look at http://www.sf.net/projects/linuxconsole

We have patches for gpm in CVS. It needs alot of work.

Re: [RFC] Get selection to buffer addition

[Russell King]

On Sat, Jan 01, 2000 at 01:24:46AM +0000, Pavel Machek wrote:
> Hi!
> 
> > I'm sending this mail again with the patch in plain text and not
> > gzip+uuencoded, sorry for any inconvenience.
> > 
> > What I'm looking for:
> >   I'm looking for comments and approval for a small addition to the console
> > driver (drivers/char/console.c).
> > 
> > Small description:
> >   The included patches adds a couple of new services to the TIOCLINUX ioctl
> > call, they are:
> > 
> > 13 (get selection into a buffer): It copies the contents of the selection
> > buffer (maintained in kernel space) into a user space provided buffer. Is
> > something like "paste to a buffer"  instead of just paste to the current
> > console.
> > 
> > 14 (get selection length): Returns the length of the selection buffer (0 if
> > none selected).
> 
> Looks good to me. Now, all I want is utility to share clipboard between
> X and text console...

Umm, silly question, but why not put this stuff into something similar to
gpm, rather than have unswappable kernel memory sucked up just for cut and
paste (possibly very large cut and paste under X).

--
Russell King (rmk@arm.linux.org.uk)                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html

Re: [RFC] Get selection to buffer addition

[salvador]

Russell King wrote:

> On Sat, Jan 01, 2000 at 01:24:46AM +0000, Pavel Machek wrote:

[snip]

> > > What I'm looking for:
> > >   I'm looking for comments and approval for a small addition to the console
> > > driver (drivers/char/console.c).

[snip]

> > Looks good to me. Now, all I want is utility to share clipboard between
> > X and text console...
>
> Umm, silly question, but why not put this stuff into something similar to
> gpm, rather than have unswappable kernel memory sucked up just for cut and
> paste (possibly very large cut and paste under X).

According to James Simmons (he is maintaining a console development CVS tree in
Sourge Force) the selection stuff will go away in 2.5.x series and the code will
be moved to user space (gpm) as Russell sugest.
As things will be changing in 2.5.x I don't think we should introduce my change.
My idea was towards what Pavel says: adding some kind of clipboard facility, after
all it was already there but not accesible. I sent this patch first and already
designed the IOCTL calls for the reverse: set the selection buffer.
When I started it I thinked this was part of gpm. I was very surprised when I
discovered the kernel was doing the real job.
As the 2.5.x code will remove it I don't know if we should add features that will
go away. The only problem for me is how many time I'll must wait before 2.6.x is
released and the users of my text editor will enjoy this facility (a clipboard
between consoles and different applications).

SET

--
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Visit my home page: http://welcome.to/SetSoft or
http://www.geocities.com/SiliconValley/Vista/6552/
Alternative e-mail: set-soft@bigfoot.com set@computer.org
                    set@ieee.org
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013