CVS status
Hi all!
I tried the CVS version of elks this afternoon and it fails to load init. Is there a known issue?
Harry
On Fri, Jul 12, 2002 at 03:51:32PM +0300, Harry Kalogirou wrote:
Hi Harry
Quoted 2 lines
I tried the CVS version of elks this afternoon and it fails to load init. Is there a known issue?
Just updated my tree and trying to build a kernel with all the defaults, it errors with:
undefined symbol: _rs_ops
This was defined in a #ifdef for CONFIG_CONSOLE_SERIAL and a patch is attached which moves it outside.
I'm just trying to get bochs set up in the office so I can do more testing in work without the laptop
Paul
Index: serial.c
===================================================================
RCS file: /cvsroot/elks/elks/arch/i86/drivers/char/serial.c,v retrieving revision 1.30
serial.c +2 -1
@@ -432,6 +432,8 @@/* Do something */ } +#endif +struct tty_ops rs_ops = { rs_open, rs_release,
@@ -440,4 +442,3 @@ rs_ioctl }; -#endif
Harry
Quoted 2 lines
I tried the CVS version of elks this afternoon and it fails to load init. Is there a known issue?
Got the cvs Image working mostly under bochs (my keyboard sometimes doesn't work) However cvs will boot and load init on the root file system from your site, and load the filesystem and start the EDE stuff.
So I don't see a problem with it (although I did apply my patch but that shouldn't break anything)
Paul
Την Παρ, 12-07-2002 στις 17:20, ο/η Paul Nasrat έγραψε:
Quoted 39 lines · Harry Kalogirou
Hi Harry
Quoted 2 lines
I tried the CVS version of elks this afternoon and it fails to load init. Is there a known issue?
Just updated my tree and trying to build a kernel with all the defaults, it errors with:
undefined symbol:
_rs_opsThis was defined in a #ifdef for
CONFIG_CONSOLE_SERIALand a patch is attached which moves it outside.I'm just trying to get bochs set up in the office so I can do more testing in work without the laptop
Paul
----Index:
serial.c
===================================================================
RCS file: /cvsroot/elks/elks/arch/i86/drivers/char/serial.c,v retrieving revision 1.30
diff -u -r1.30serial.cserial.c +2
-0@@ -432,6 +432,8 @@/* Do something */ } +#endif +struct tty_ops rs_ops = { rs_open, rs_release,
serial.c line 443
Quoted 4 lines
serial.c +0
-1@@ -440,4 +442,3 @@ rs_ioctl }; -#endif
Mmmm.. this might was the problem with ktcp not working all this time(actualy stty before ktcp)...
Harry
Hi Harry.
Quoted 8 lines
Quoted 8 lines
Quoted 2 lines
I tried the CVS version of elks this afternoon and it fails to load init. Is there a known issue?
Just updated my tree and trying to build a kernel with all the defaults, it errors with:
undefined symbol:
_rs_ops
Quoted 8 lines
Quoted 5 lines
This was defined in a #ifdef for
CONFIG_CONSOLE_SERIALand a patch is attached which moves it outside.I'm just trying to get bochs set up in the office so I can do more testing in work without the laptop
Mmmm.. this might was the problem with ktcp not working all this time (actualy stty before ktcp)...
I've now applied this patch and tested it, and on my 286 based laptop, ELKS now boots and ktcp works again, so basic testing says that this problem can now be declared to be fixed.
Harry: I've committed this to the CVS along with the CHANGELOG entry, and relabelled it as kernel 0.1.1-pre3 so if you can do your usual and set up a release thereof, it'd be appreciated. I've also tagged this as elks-0_1_1-pre3 as a version marker.
We're probably close to a 0.1.1 release, but there's a few issues I want to clear up before we do that...
Best wishes from Riley.
Την Παρ, 12-07-2002 στις 18:48, ο/η Paul Nasrat έγραψε:
Quoted 9 lines
Harry
Quoted 2 lines
I tried the CVS version of elks this afternoon and it fails to load init. Is there a known issue?
Got the cvs Image working mostly under bochs (my keyboard sometimes doesn't work) However cvs will boot and load init on the root file system from your site, and load the filesystem and start the EDE stuff.
The problem comes from the new images produced by the elkscmd package. I tried the old made root/boot disks from the site with the new kernel and it works, as you state.
A closer look shows that the new MAKEDEV script in the root_template has some problems(Replacing the new /dev with the /dev from the site floppies solves the problem). I don't have the time see what is wrong right now, but I suggest more carefull and tested commits from now on.
Harry
-
To unsubscribe from this list: send the line "unsubscribe linux-8086" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Την Παρ, 12-07-2002 στις 21:49, ο/η Riley Williams έγραψε:
Quoted 8 lines
Hi Harry.
Quoted 2 lines
Mmmm.. this might was the problem with ktcp not working all this time (actualy stty before ktcp)...
I've now applied this patch and tested it, and on my 286 based laptop, ELKS now boots and ktcp works again, so basic testing says that this problem can now be declared to be fixed.
Ok.
Quoted 4 lines
Harry: I've committed this to the CVS along with the CHANGELOG entry, and relabelled it as kernel 0.1.1-pre3 so if you can do your usual and set up a release thereof, it'd be appreciated. I've also tagged this as elks-0_1_1-pre3 as a version marker.
I'm not going to make a release since I can't create images normaly.
Quoted 5 lines
We're probably close to a 0.1.1 release, but there's a few issues I want to clear up before we do that...
Best wishes from Riley.
Harry
-
To unsubscribe from this list: send the line "unsubscribe linux-8086" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Harry, Paul.
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
Mmmm.. this might was the problem with ktcp not working all this time (actualy stty before ktcp)...
I've now applied this patch and tested it, and on my 286 based laptop, ELKS now boots and ktcp works again, so basic testing says that this problem can now be declared to be fixed.
Ok.
I'll add that this particular bug was as frustrating as they come. I'd been going through each file one at a time and reverting the changes I'd made to that file, then compiling the result, and nothing I did made the slightest difference to getting it working again. I'd already reverted serial.c without success as well !!!
I had determined that the problem was with the stty command and posted an email to that effect to this list some time back. I managed this by tweaking the /etc/rc.d/rc.sysinit script to deal with this case in the same way that the Red Hat Linux scripts deal with multiple-stage items, and the relevant patch was committed to the elkscmd tree some time ago. It's enclosed below for your reference. The resulting display failed to display the stty before panicking, hence my determining that there was at least a problem at this stage...
===============8<=============== CUT ===============>8===============
Index: rootfs_template/etc/rc.d/rc.sysinit
===================================================================
RCS file: /cvsroot/elks/elkscmd/rootfs_template/etc/rc.d/rc.sysinit,v retrieving revision 1.1
etc/rc.d/rc.sysinit +8 -6
@@ -12,26 +12,28 @@ sliptty=/dev/ttys0 ttybaud=4800 if test -f /bin/ktcp then - echo -n "Starting kernel network facilities: TCP/IP" + echo -n 'Starting kernel network facilities:' stty $ttybaud < $sliptty + echo -n ' stty' ktcp $localip $sliptty & + echo -n ' ktcp' - echo . + echo '' - echo -n "Staring network daemons: " + echo -n "Starting network daemons: " - for daemon in httpd + for daemon in ftpd httpd do if test -f /bin/$daemon then - echo -n " $daemon" $daemon & + echo -n " $daemon" fi done - echo . + echo '' fi # # View message of day
===============8<=============== CUT ===============>8===============
In case you haven't guessed, I'm working on an ftpd daemon and have also modified the above for when that is ready...
Quoted 6 lines
Quoted 4 lines
Harry: I've committed this to the CVS along with the CHANGELOG entry, and relabelled it as kernel 0.1.1-pre3 so if you can do your usual and set up a release thereof, it'd be appreciated. I've also tagged this as elks-0_1_1-pre3 as a version marker.
I'm not going to make a release since I can't create images normaly.
Fair enough. The label is there anyway.
Actually, a reasonable release could be made by taking your images.zip file, extracting each image out, mounting it and copying the revised /etc/rc.d/rc.sysinit and the newly compiled kernel over the ones there, then uploading the result.
Quoted 2 lines
Quoted 2 lines
We're probably close to a 0.1.1 release, but there's a few issues I want to clear up before we do that...
With the progress that has been made over the last few days, we can expect to be ready for 0.1.1 before the month finishes...
Best wishes from Riley.
Hi Harry.
Quoted 12 lines
Quoted 7 lines
Quoted 2 lines
I tried the CVS version of elks this afternoon and it fails to load init. Is there a known issue?
Got the cvs Image working mostly under bochs (my keyboard sometimes doesn't work) However cvs will boot and load init on the root file system from your site, and load the filesystem and start the EDE stuff.
The problem comes from the new images produced by the elkscmd package. I tried the old made root/boot disks from the site with the new kernel and it works, as you state.
Quoted 5 lines
A closer look shows that the new MAKEDEV script in the
root_templatehas some problems(Replacing the new /dev with the /dev from the site floppies solves the problem). I don't have the time see what is wrong right now, but I suggest more carefull and tested commits from now on.
I would presume this was my fault as I tweaked the MAKEDEV script a while back, but have to report that the resultant script worked fine on my test system.
The only possible problem that I can see is that the MKSET definition near the top of the file used a relative path for the MAKESET auxilliary script, and that shouldn't be the problem as the remaining mknod lines require that relative path to be correct. However, if the MAKESET script is missing, that would be a problem.
I've replaced the definition with an absolute path to be safe, so if that was it, then it isn't now. I've also included a check that the auxilliary MAKESET script is actually present, displaying an error message and aborting if not.
Harry: As my testing of this is obviously unreliable, can you check this out for me please, and advise regarding the results so we can get this problem sorted out.
Best wishes from Riley.