Re: [ECOS] tty.c driver bug?

[Jonathan Larmour <jlarmour at redhat dot co dot uk>]

Gary Thomas wrote:
> 
> AFAIK, nothing was done about this.
> 
> On 05-Apr-00 Jonathan Larmour wrote:
> > Prompted by Sergei's recent bug report, I don't recall this mail from
> > Patrick being resolved. Is that right?
> >
> > Patrick O'Grady wrote:
> >>
http://sourceware.cygnus.com/ecos/docs-latest/ref/ecos-ref/tty-driver.html
> >> shows some of the different TTY input and output modes which can be
> >> selected...  it implies that CYG_TTY_IN_FLAGS_CR will translate '\r' into
> >> '\n', and that CYG_TTY_IN_FLAGS_CRLF will translate the pair '\r\n' into a
> >> single '\n'...  however, looking in
> >> packages/io/serial/current/src/common/tty.c, about line 212, shows that
> >> CYG_TTY_IN_FLAGS_CR is ignored, and that CYG_TTY_IN_FLAGS_CRLF will
> >> generate a pair of '\n':
[snip]
> >> Looks like CYG_TTY_IN_FLAGS_CRLF should really be CYG_TTY_IN_FLAGS_CR, and
> >> that a previous character should be stored for use with
> >> CYG_TTY_IN_FLAGS_CRLF.  I'll use CYG_TTY_IN_FLAGS_CRLF for the time
> >> being--all I really care about is when the user presses '\r'.

I think the attached patch should do the trick to fix this. Let me know
what you think Patrick.

>  Might be a
> >> good idea that CYG_TTY_IN_FLAGS_CRLF returns a single '\n' for either
> >> '\r\n' or '\n\r'...  I've seen some lame systems which output those
> >> characters in reverse.  Cheers!

Well, in that mode it's easiest just to pretend the '\r' didn't exist.
That's subtly different behaviour from treating it as '\n\r' because if we
did that, it would mean getting the '\n' and deliberately not returning
from tty_read() until we got the '\r'. We could add that complexity to cope
with such systems, but let's not - by keeping it this way, setting
CYG_TTY_IN_FLAGS_CRLF means it works when connected to systems that send
CRLF *and* ones that send just LF.

Jifl
-- 
Red Hat, 35 Cambridge Place, Cambridge, UK. CB2 1NS  Tel: +44 (1223) 728762
"Plan to be spontaneous tomorrow."  ||  These opinions are all my own fault
Index: tty.c
===================================================================
RCS file: /cvs/ecc/ecc/io/serial/current/src/common/tty.c,v
retrieving revision 1.11
diff -u -5 -p -c -r1.11 tty.c
*** tty.c	2000/04/05 18:55:01	1.11
--- tty.c	2000/04/07 03:03:58
*************** tty_read(cyg_io_handle_t handle, void *_
*** 208,240 ****
              *len = size;
              return res;
          }
          buf[size++] = c;
          if ((priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_BINARY) == 0) {
!             if ((c == '\b') || (c == 0x7F)) {
                  size -= 2;  // erase one character + 'backspace' char
                  if (size < 0) {
                      size = 0;
                  } else if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_ECHO) {
                      clen = 3;
                      cyg_io_write(chan, "\b \b", &clen);
                  }
!             } else if ((c == '\n') || (c == '\r')) {
!                 clen = 2;
                  if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_ECHO) {
!                     cyg_io_write(chan, "\r\n", &clen);
                  }
!                 if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_CRLF) {
                      c = '\n';  // Map CR -> LF
                  }
                  buf[size-1] = c;
!                 break;
!             } else {
                  if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_ECHO) {
                      clen = 1;
                      cyg_io_write(chan, &c, &clen);
                  }
              }
          }
      }
      *len = size;
      return ENOERR;
--- 208,256 ----
              *len = size;
              return res;
          }
          buf[size++] = c;
          if ((priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_BINARY) == 0) {
!             switch (c) {
!             case '\b':    /* drop through */
!             case 0x7f:
                  size -= 2;  // erase one character + 'backspace' char
                  if (size < 0) {
                      size = 0;
                  } else if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_ECHO) {
                      clen = 3;
                      cyg_io_write(chan, "\b \b", &clen);
                  }
!                 break;
!             case '\r':
!                 if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_CRLF) {
!                     /* Don't do anything because a '\n' will come next */
!                     break;
!                 }
!                 /* drop through */
!             case '\n':
                  if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_ECHO) {
!                     if (priv->dev_info.tty_out_flags & CYG_TTY_OUT_FLAGS_CRLF) {
!                         clen = 2;
!                         cyg_io_write(chan, "\r\n", &clen);
!                     } else {
!                         clen = 1;
!                         cyg_io_write(chan, "\n", &clen);
!                     }
                  }
!                 if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_CR) {
                      c = '\n';  // Map CR -> LF
                  }
                  buf[size-1] = c;
!                 *len = size;
!                 return ENOERR;
!             default:
                  if (priv->dev_info.tty_in_flags & CYG_TTY_IN_FLAGS_ECHO) {
                      clen = 1;
                      cyg_io_write(chan, &c, &clen);
                  }
+                 break;
              }
          }
      }
      *len = size;
      return ENOERR;