Alan:
Here is an updated patch that addresses the issues you raised.
Also, in talking with Aaron Zang, we always want to run fbconsole
on Sparc, so the patch now reflects this as well.
Brian
On 08/20/09 20:28, Alan Coopersmith wrote:
>> +# Check for Solaris fbconsole support
>> +#
>> +AC_MSG_CHECKING(for Solaris fbconsole)
>> +if test -x /usr/X11/bin/fbconsole; then
>> + AC_DEFINE(HAVE_FBCONSOLE)
>> + AC_MSG_RESULT(yes)
>> +else
>> + AC_MSG_RESULT(no)
>> +fi
>
> That looks like it should be AC_CHECK_FILE, but it would probably be
> even better to use AC_PATH_PROG, since we know fbconsole can be
> in a couple different places (/usr/openwin/bin on S10 and older
> Nevada builds, /usr/X11/bin on current nevada builds, possibly
> /usr/bin in the future).
>
>> +#ifdef HAVE_FBCONSOLE
>> +#include<sys/vt.h>
>> +#endif
>
> If you're going to upstream, you'll also need to do an AC_CHECK_HEADERS
> for sys/vt.h and only include that when it's present, since older
> builds/releases won't have it.
>
>> +#define FBCONSOLE "/usr/openwin/bin/fbconsole"
>
> For OpenSolaris/current Nevada, /usr/X11/bin is preferred, so it doesn't
> break if someone breaks the /usr/openwin -> /usr/X11 link by installing
> a legacy SVR4 package - best would be using AC_PATH_PROG as noted above.
>
>
>> + g_debug ("Running on console");
>> + fd = open ("/dev/vt/0", O_WRONLY);
>> + if (ioctl (fd, VT_ENABLED,&vtstat) == 0) {
>> + g_debug ("VT is not enabled");
>> + gdm_exec_fbconsole (server);
>> + } else {
>> + g_debug ("VT is enabled, so not forking
>> fbconsole");
>> + }
>
> For upstreaming, it should also exec fbconsole if the previously mentioned
> check for<sys/vt.h> was false.
>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: gdm-09-fbconsole.diff
URL:
<http://mail.opensolaris.org/pipermail/jds-review/attachments/20090821/711a4d98/attachment.ksh>
