Re: RES: RES: vt newcons mouse paste issue FIXED
Date: Wed, 22 Jun 2022 16:48:47 UTC
Hi.
Not actually tested, but this can cause breakage on non-ascii cases.
Maybe also (or instead) iswspace() test would be needed.
Possibly additional mbrtowc() in the argument of iswspace().
Please see `man 3 multibyte` and `man 3 iswspace`.
(Possibly more to see.)
Characters in buf can be multibyte or wide char depending on locale,
as vt can show at least UTF-8 characters.
Sorry, not looked into enough how UTF-8 characters outside ascii are
handled internally on vt.
On Wed, 22 Jun 2022 15:01:36 +0000
Ivan Quitschal <tezeka@hotmail.com> wrote:
> Hi Hans
>
> Actually i can , I should've cut off the '\r' 😊 , this is what was causing the term to go bend
>
> This is the correct diff (option -C 9999 doesn’t worked with -u)
> I also included your code for the pts anyway
>
>
>
> --- sys/dev/vt/vt_buf.c.orig 2022-06-22 11:48:39.705597000 -0300
> +++ sys/dev/vt/vt_buf.c 2022-06-22 11:51:05.502415000 -0300
> @@ -41,6 +41,7 @@
> #include <sys/malloc.h>
> #include <sys/mutex.h>
> #include <sys/reboot.h>
> +#include <sys/ctype.h>
>
> #include <dev/vt/vt.h>
>
> @@ -752,6 +753,7 @@
> {
> int i, r, c, cs, ce;
> term_pos_t s, e;
> + term_char_t *end;
>
> /* Swap according to window coordinates. */
> if (POS_INDEX(vtbuf_htw(vb, vb->vb_mark_start.tp_row),
> @@ -772,10 +774,15 @@
> for (c = cs; c < ce; c++) {
> buf[i++] = vb->vb_rows[r][c];
> }
> + for (end = buf + i; end-- != buf; ) {
> + if (isspace((unsigned char)*end) == false)
> + break;
> + *end = '\0';
> + }
> /* Add new line for all rows, but not for last one. */
> if (r != e.tp_row) {
> - buf[i++] = '\r';
> buf[i++] = '\n';
> + buf[i++] = '\0';
> }
> }
> }
>
> Works fine here
>
> Thanks
>
> --tzk
>
> -----Mensagem original-----
> De: Hans Petter Selasky <hps@selasky.org>
> Enviada em: quarta-feira, 22 de junho de 2022 11:02
> Para: Ivan Quitschal <tezeka@hotmail.com>; freebsd-current@freebsd.org
> Assunto: Re: RES: vt newcons mouse paste issue FIXED
>
> On 6/22/22 15:36, Ivan Quitschal wrote:
> > Hi Hans
> >
>
> Hi Ivan,
>
> I think you should upload the diff at:
>
> https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Freviews.freebsd.org%2Fdifferential%2F&data=05%7C01%7C%7C85e3f391dfc941f2852908da5457c75f%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637915033252675801%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=4sxXIR9NhIbwD42gIrP4pYcdnUinqae1SNZAjclr2Aw%3D&reserved=0
>
> Make the diff like this:
>
> diff -u -C 999999 sys/dev/vt/vt_buf.c.orig sys/dev/vt/vt_buf.c > a.diff
>
>
> I see two issues:
>
> 1) Pointer arithmetics is not so good!
>
> > }
> > + end = buf + i - 1;
> > + while (end > buf && isspace((unsigned char)*end))
> > + {
> > + *end = '\0';
> > + end--;
> > + }
> > +
>
> I think this would be better and avoid the ">" with pointers!
>
> for (end = buf + i; end-- != buf; ) {
> if (isspace((unsigned char)*end) == false)
> break;
> *end = '\0';
> }
>
> Can you explain this:
>
> > - buf[i++] = '\r';
> > + buf[i] = '\r';
> > buf[i++] = '\n';
>
> '\r' character is now overwritten by '\n' character.
>
> --HPS
>
--
Tomoaki AOKI <junchoon@dec.sakura.ne.jp>