all right, i have some intriguing observations to report. but first, i want to settle this "rhgb" boot-line option. can someone point me to where this option has any effect *at* *all*?
i have gone over these two files:
/etc/sysconfig/init /etc/rc.sysinit
and it *appears* that whether or not you get a graphical boot is tested in /etc/rc.sysinit, and is based *solely* on the following:
RHGB_STARTED=0 GRAPHICAL=yes BOOTUP=color access to /usr/bin/rhgb
there is, AFAICT, no test *anywhere* that looks for the option "rhgb" on the boot line. OTOH, the file /etc/rc.sysinit *does* check for the boot-line option "nogui" to *prevent* the graphical boot. so ... who wants to explain how the option "rhgb" means anything, and where it's tested. in short, i see "nogui" to prevent a graphical boot if it's turned on. i see no "rhgb" test to activate a graphical boot if it's turned off.
second, if you check /etc/rc.sysinit, whether you get a graphical boot early depends on whether /usr is a separate filesystem. if it is, you can't get a graphical boot since /usr/bin/rhgb isn't even available yet. so that first conditional is skipped early in the file.
however, the whole test is done again around line 560 where, by now, /usr is mounted and it should work. at least, that's the theory. in my case, though, what i get when the line "/usr/bin/rhgb" is run is
umount2: Invalid argument. umount: /initrd: not mounted.
printed twice. what the ...? i'd seen these crop up recently and was planning on tracking them down at some point, but it never occurred to me to associate them with the second invocation of rhgb.
can anyone explain what these mean? that might go a long way to clearing this up. well, at least for me.
rday
On Fri, 2003-10-31 at 17:35, Robert P. J. Day wrote:
all right, i have some intriguing observations to report. but first, i want to settle this "rhgb" boot-line option. can someone point me to where this option has any effect *at* *all*?
[snip]
there is, AFAICT, no test *anywhere* that looks for the option "rhgb" on the boot line.
I suppose it's magically happening then :). What's wrong with your /etc/rc.sysinit? Mine has the obvious test on line 91: --- if fgrep rhgb /proc/cmdline > /dev/null 2>&1 && [ "$BOOTUP" = "color" -a "$GRAPHICAL" = "yes" -a -x /usr/bin/rhgb ]; then /usr/bin/rhgb RHGB_STARTED=1 fi ---
# rpm -q --whatprovides /etc/rc.sysinit initscripts-7.42-1
however, the whole test is done again around line 560 where, by now, /usr is mounted and it should work.
It's line 572 for me.
On Fri, 31 Oct 2003, nosp wrote:
On Fri, 2003-10-31 at 17:35, Robert P. J. Day wrote:
all right, i have some intriguing observations to report. but first, i want to settle this "rhgb" boot-line option. can someone point me to where this option has any effect *at* *all*?
[snip]
there is, AFAICT, no test *anywhere* that looks for the option "rhgb" on the boot line.
I suppose it's magically happening then :). What's wrong with your /etc/rc.sysinit? Mine has the obvious test on line 91:
if fgrep rhgb /proc/cmdline > /dev/null 2>&1 && [ "$BOOTUP" = "color" -a "$GRAPHICAL" = "yes" -a -x /usr/bin/rhgb ]; then
what the heck? mine reads:
if ! fgrep -q nogui /proc/cmdline && [ "$BOOTUP" = "color" -a "$GRAPHICAL" = "yes" -a -x /usr/bin/rhgb ]; then /usr/bin/rhgb RHGB_STARTED=1 fi
is one a newer version? i've been letting yum update stuff automatically. which is the industry-approved version? is one version or the other set at install time based on something? how odd.
rday
You must have an older version. Newest version of initscripts is initscripts-7.42-1
what the heck? mine reads:
if ! fgrep -q nogui /proc/cmdline && [ "$BOOTUP" = "color" -a "$GRAPHICAL" = "yes" -a -x /usr/bin/rhgb ]; then /usr/bin/rhgb RHGB_STARTED=1 fi
is one a newer version? i've been letting yum update stuff automatically. which is the industry-approved version? is one version or the other set at install time based on something? how odd.
rday
-- fedora-test-list mailing list fedora-test-list@redhat.com http://www.redhat.com/mailman/listinfo/fedora-test-list
On Fri, 31 Oct 2003, Jaap A. Haitsma wrote:
You must have an older version. Newest version of initscripts is initscripts-7.42-1
what the heck? mine reads:
if ! fgrep -q nogui /proc/cmdline && [ "$BOOTUP" = "color" -a "$GRAPHICAL" = "yes" -a -x /usr/bin/rhgb ]; then /usr/bin/rhgb RHGB_STARTED=1 fi
i;ve just established that -- mine is 7.36-2, which is odd since i thought i'd configured yum to update on a nightly basis. weird.
but it also points out a noticeable change in functionality between the old and newer versions of initscripts, which almost certainly has been causing some of this confusion.
rday
On Fri, 2003-10-31 at 13:11, Robert P. J. Day wrote:
but it also points out a noticeable change in functionality between the old and newer versions of initscripts, which almost certainly has been causing some of this confusion.
Correct, and if you'll check the list archives, you'll see that we already had this conversation several times (last week, I think), about the "nogui" vs "rghb" options. :-)
--Jeremy
Robert P. J. Day wrote:
if ! fgrep -q nogui /proc/cmdline && [ "$BOOTUP" = "color" -a "$GRAPHICAL" = "yes" -a -x /usr/bin/rhgb ]; then /usr/bin/rhgb RHGB_STARTED=1 fi
is one a newer version? i've been letting yum update stuff
mine reads the same, on a vanilla test3 installation.
On Fri, 2003-10-31 at 09:41, nosp wrote:
On Fri, 2003-10-31 at 17:35, Robert P. J. Day wrote:
all right, i have some intriguing observations to report. but first, i want to settle this "rhgb" boot-line option. can someone point me to where this option has any effect *at* *all*?
I have a RH9 upgraded system to FCT3, then full rawhide updates, incremental since T3 release day when I upgraded. I haven't ever seen rhgb work on this system, but was fairly certain it was due to the upgraded (and highly modified / borked over system, still kinda acts like redhat I suppose...) :D
The point here, is that many people are apparently still having issues with rhgb, even those who are using the correct command line. For this reason I've shared this experiment, not because I think my system is a proper test bed for this particular feature. You can find my rc.sysinit here to try it yourself, or modify yours alike (initscripts-7.42-1): http://www.andrewfarris.com/fedora/rc.sysinit
I've added some simple echos to take a look at what is going on at this point in startup. On my system /usr is indeed not mounted at the first 'rhgb startup' sequence, however it is mounted before the second, and rhgb thinks it is running but no display change occurs. Only if I add the --force option to rhgb does it actually work, and then only changes to a blank screen with default X cursor. I have no doubt that fresh installs are much better at this, but people do update afterall, and it doesn't quite work for me.
[snip]
there is, AFAICT, no test *anywhere* that looks for the option "rhgb" on the boot line.
I suppose it's magically happening then :). What's wrong with your /etc/rc.sysinit? Mine has the obvious test on line 91: -- if fgrep rhgb /proc/cmdline > /dev/null 2>&1 && [ "$BOOTUP" = "color" -a "$GRAPHICAL" = "yes" -a -x /usr/bin/rhgb ]; then /usr/bin/rhgb RHGB_STARTED=1 fi --
immediately following the above lines (in both locations in rc.sysinit) I have:
[code]
## check rhgb failure? echo -e "\nrhgb failure check:" RHGBFILES="/usr/share/rhgb/fedora-logo.png /usr/share/rhgb/large-computer.png /usr/share/rhgb/throbber-anim.png /usr/bin/rhgb /usr/bin/rhgb-client" echo -e "\trhgb files ( existence checked one at a time ):" for rhgbfile in $RHGBFILES ; do if [ -e $rhgbfile ] ; then echo -e "\t\t$rhgbfile exists" fi done
echo -e "\nrhgb: bootup = $BOOTUP ; graphical = $GRAPHICAL ; rhgb_started = $RHGB_STARTED" if [ -x /usr/bin/rhgb-client ] && /usr/bin/rhgb-client --ping ; then echo "/usr/bin/rhgb-client --ping succeeded" echo -en "manually set rhgb details off:" if /usr/bin/rhgb-client --details=no ; then echo "succeeded" else echo "failed" fi else echo "no /usr/bin/rhgb-client yet" fi echo -e "kernel commandline: `cat /proc/cmdline`\n" ## finished rhgb failure check
[/code]
which produces the following output the second time (first fails in on all points, due to /usr not being available)
[code] rhgb failure check: rhgb files ( existence checked one at a time ): /usr/share/rhgb/fedora-logo.png exists /usr/share/rhgb/large-computer.png exists /usr/share/rhgb/throbber-anim.png exists /usr/bin/rhgb exists /usr/bin/rhgb-client exists
rhgb: bootup = color ; graphical = yes ; rhgb_started = 1 no /usr/bin/rhgb-client yet kernel commandline: ro root=LABEL=/ hdc=ide-scsi hdd=ide-scsi vga=792 rhgb [/code]
Not sure why it fails to start rhgb the second time, but because it does fail the rhgb-client --ping fails as well. So if you can't get rhgb to work for the life of you.. plug that in there and see what she says. Interesting to see anyway. NOTE: you may want to download and copy out of my rc.sysinit, be careful of email linebreaks that don't belong
(and if there are errors or invalid tests there, point em out please, no real idea what I'm doing and just learning bash...)
-- fedora-test-list mailing list fedora-test-list@redhat.com http://www.redhat.com/mailman/listinfo/fedora-test-list
On Fri, 2003-10-31 at 17:23, Andrew Farris wrote:
kernel commandline: ro root=LABEL=/ hdc=ide-scsi hdd=ide-scsi vga=792 rhgb
Forgot to mention that, no the vga setting does not make any difference to this test. The same result occurs in several text modes as well as the framebuffers.
-- fedora-test-list mailing list fedora-test-list@redhat.com http://www.redhat.com/mailman/listinfo/fedora-test-list
-
Andrew Farris -
Jaap A. Haitsma -
Jeremy Portzer -
Matthias Buelow -
nosp -
Robert P. J. Day