Re: pre-R5&miscs

Re: pre-R5&miscs

[Martin Costabel]

"G. Laurès" wrote:
> 
> Hi guys, I just discovered the pre-R5 directory @ dev.linuxppc.org.
> 
>  It does include the latest version of Bootx, but not ramdisk.image.gz as in
> previous distributions (R4.1 for example). Note that this is also the case for R4.7.

What and where is R4.7?

> 
>   I've seen an initrd.img and a supp.img but I'm not sure if this should be
> used for installation. As I could not see installation tips anywhere, ????

This is a new installer (Redhat 5.1, I guess). The old one doesn't work
with the files they put into the base folder. It is so new that
apparently they have not yet had the time to write instructions. The
following worked for me:

1. Rename initrd.img to ramdisk.image.gz and put it where BootX can see
it.

2. Fill supp.img into a floppy using
   dd if=supp.img of=/dev/fd0 bs=1440k

3. If you want to install from a HD, put your RedHat folder on a UNIX
partition, not on a HFS one, otherwise the installer can't see it (IMHO
this is a bug, or a feature of a not quite finished product...)

4. Run BootX with "use Ramdisk" checked. After a couple of preliminary
questions, the installer will ask you to change your floppy. Put the
floppy with the supp.img into the drive. The installation continues then
normally.

> 
>  I was waiting R5 to put on my new PBG3, didn't want to mess with an R4 -> R5
> upgrade later.
> 
> BTW, should I go ahead and put it on my father's PBG3 as well, knowing that
> will be his first professional steps in linuxppc (and linux at all) so I don't
> want to mess him with a too much experimental system... would be 4 or 4.1 or
> 4.7 with recent kernel for PBG3s (2.1.130 I believe ?) better ??
> 

It is still very "beta". For me, the X server is not yet usable.
I am on a PM6400 with the built-in video (valkyrie), so I don't know
about PBG3s.

If I use XPmac, then it works in principle, but the colors are all
messed up, most of the time nothing is readable. I say 'most of the
time', because the weird colors are not always the same, they depend on
the number of the console from which X is started. Also, if you touch
the title bar of a window, the X server dies. This has been mentioned
here, and it seems that someone already has a fix.

If I use XF68_FBDev, then the colors are OK, but the keyboard and the
mouse do not work correctly: The option (alt) keys do not work, and
there is no simulation for mouse buttons 2 and 3.
This is a problem that has been mentioned to the author (Geert U.)
already some time ago, but since he doesn't have ADB devices any more,
it is not clear whether this will be fixed soon.

So for people with standard Apple keyboards and mice, there remains a
serious problem.

On the other hand, here is something positive: The compilers are working
well with the new glibc. One example: Since I am using dhcp for the
network connection (like you probably), I fetched the sources for
dhcpcd-1.3.16 (the dhcpcd-0.70-2.ppc.rpm which comes with the R5beta is
completely useless for linuxppc 2.1.1xx kernels). In linuxppc R4, the
compilation used to be somewhat tricky, due to glibc problems. Here it
works now without any problems; just one line with some CFLAGS in the
Makefile had to be adapted. And the best thing is: it works! In my R4
system, since a week or so, I always get timeouts from dhcp, so I had to
use either an old 2.1.24 kernel or a fixed IP address. Now I can even
chroot from the R4 system to the R5beta and run dhcpcd there and it
connects! I suppose this has also something to do with the more recent
networking packages, net-tools etc.

--
Martin

[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to  Cc linuxppc-dev  if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]

Re: pre-R5&miscs

[Tom Rini]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: TEXT/PLAIN; charset=X-UNKNOWN, Size: 3256 bytes --]


On Sun, 6 Dec 1998, Martin Costabel wrote:

> "G. Laurès" wrote:
> > 
> > Hi guys, I just discovered the pre-R5 directory @ dev.linuxppc.org.
> > 
> >  It does include the latest version of Bootx, but not ramdisk.image.gz as in
> > previous distributions (R4.1 for example). Note that this is also the case for R4.7.
> 
> What and where is R4.7?

Just another symmlink on dev.linuxppc.org pointing back to
/pub/users/fray/RedHat or there abouts.

> If I use XPmac, then it works in principle, but the colors are all
> messed up, most of the time nothing is readable. I say 'most of the
> time', because the weird colors are not always the same, they depend on
> the number of the console from which X is started. Also, if you touch
> the title bar of a window, the X server dies. This has been mentioned
> here, and it seems that someone already has a fix.

I believe this is similar to the problem of using Xpmac compat mode
normally, isn't it?

> If I use XF68_FBDev, then the colors are OK, but the keyboard and the
> mouse do not work correctly: The option (alt) keys do not work, and
> there is no simulation for mouse buttons 2 and 3.
> This is a problem that has been mentioned to the author (Geert U.)
> already some time ago, but since he doesn't have ADB devices any more,
> it is not clear whether this will be fixed soon.
>
> So for people with standard Apple keyboards and mice, there remains a
> serious problem.

Hmm.  Haven't used the XF68_FBDev's that i've built for a while, but they
did work fine here.  Right now I'm using Geert's server from 10something
run throught fix961212.  Make sure you have Xkb stuff turned off in the
XF86Config file.
 
> On the other hand, here is something positive: The compilers are working
> well with the new glibc. One example: Since I am using dhcp for the
> network connection (like you probably), I fetched the sources for
> dhcpcd-1.3.16 (the dhcpcd-0.70-2.ppc.rpm which comes with the R5beta is
> completely useless for linuxppc 2.1.1xx kernels). In linuxppc R4, the
> compilation used to be somewhat tricky, due to glibc problems. Here it
> works now without any problems; just one line with some CFLAGS in the
> Makefile had to be adapted. And the best thing is: it works! In my R4
> system, since a week or so, I always get timeouts from dhcp, so I had to
> use either an old 2.1.24 kernel or a fixed IP address. Now I can even
> chroot from the R4 system to the R5beta and run dhcpcd there and it
> connects! I suppose this has also something to do with the more recent
> networking packages, net-tools etc.

You can blame RedHat for shipping RawHide with a 2.1.1xx kernel and too
old a dhcpcd :)  Same thing w/ gtk+ there now.  It's real nice to have
more official type tools.  Most of the patches needed to port all this
stuff over was packages being too messily done before for either egcs or
glibc 2.1, and those tend to be minor fixes.

---
Tom Rini (TR1265)
http://dobbstown.yeti.edu/


[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to  Cc linuxppc-dev  if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]

Re: pre-R5&miscs

[Martin Costabel]

Tom Rini <tmrini@ntplx.net> wrote on Sun, 6 Dec 1998:

> On Sun, 6 Dec 1998, Martin Costabel wrote:

>> If I use XPmac, then it works in principle, but the colors are all
>> messed up, most of the time nothing is readable. I say 'most of the
>> time', because the weird colors are not always the same, they depend on
>> the number of the console from which X is started. Also, if you touch
>> the title bar of a window, the X server dies. This has been mentioned
>> here, and it seems that someone already has a fix.
>
> I believe this is similar to the problem of using Xpmac compat mode
> normally, isn't it?

If by 'normally', you mean linuxppc R4, then no, this is not similar. For me,
Xpmac on R4 works perfectly (I helped writing the valkyriefb driver, after all,
to make sure that this is so). On pre-R5 *with the same kernel* and the same
boot procedure, the colors are crazy. 
And they depend not only on the number of the console from which X
is started, but also on the time of the day or whatever. Almost every time I
startx, I get different colors. The windows that should be black on pale yellow
in my config, for example, are sometimes yellow on green, sometimes dark blue onvery dark blue, sometimes green on black and so on. You get the picture.

I tried to run the Xpmac from R4, fixed by fix961212, but it dies saying it 
cannot define the default font path and doesn't find the font 'fixed'. But
this is exactly the same font path that Xpmac from pre-R5 finds without
problems. Weird. 

>> If I use XF68_FBDev, then the colors are OK, but the keyboard and the
>> mouse do not work correctly: The option (alt) keys do not work, and
>> there is no simulation for mouse buttons 2 and 3.
>> This is a problem that has been mentioned to the author (Geert U.)
>> already some time ago, but since he doesn't have ADB devices any more,
>> it is not clear whether this will be fixed soon.
>>
>> So for people with standard Apple keyboards and mice, there remains a
>> serious problem.
>
> Hmm.  Haven't used the XF68_FBDev's that i've built for a while, but they
> did work fine here.  Right now I'm using Geert's server from 10something
> run throught fix961212.  Make sure you have Xkb stuff turned off in the
> XF86Config file.
 
The 2 servers from Geert's that I tried didn't want to run on pre-R5. I don't 
quite know why, because they dump 2 pages of messages on the screen, only the
second of which is visible. This is a help file. I guess the real errors are
mentioned on the first, invisible, page. Not very helpful.

This problem with the non-functioning option keys seems to be the same in
R4 and pre-R5. And I am pretty sure that the problem is relatively recent.
I think it was not there at a time when kernel versions were around 
2.1.115-121 or so. I don't know how to test this.

--
Martin


[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to  Cc linuxppc-dev  if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]

Re: pre-R5&miscs

[Tom Rini]

On Mon, 7 Dec 1998, Martin Costabel wrote:

> Tom Rini <tmrini@ntplx.net> wrote on Sun, 6 Dec 1998:
> 
> > On Sun, 6 Dec 1998, Martin Costabel wrote:
> 
> >> If I use XPmac, then it works in principle, but the colors are all
> >> messed up, most of the time nothing is readable. I say 'most of the
> >> time', because the weird colors are not always the same, they depend on
> >> the number of the console from which X is started. Also, if you touch
> >> the title bar of a window, the X server dies. This has been mentioned
> >> here, and it seems that someone already has a fix.
> >
> > I believe this is similar to the problem of using Xpmac compat mode
> > normally, isn't it?
> 
> If by 'normally', you mean linuxppc R4, then no, this is not similar. For me,
> Xpmac on R4 works perfectly (I helped writing the valkyriefb driver, after all,
> to make sure that this is so). On pre-R5 *with the same kernel* and the same
> boot procedure, the colors are crazy. 
> And they depend not only on the number of the console from which X
> is started, but also on the time of the day or whatever. Almost every time I
> startx, I get different colors. The windows that should be black on pale yellow
> in my config, for example, are sometimes yellow on green, sometimes dark blue onvery dark blue, sometimes green on black and so on. You get the picture.

Hmmm.  What XFree86 rpm is up there?  I've only got 3.3.2.23-22 installed
right now, but I'm also grabbing 3.3.3-10b.  Could you try 10b (Xpmac and
FBDev) when they show up?

> I tried to run the Xpmac from R4, fixed by fix961212, but it dies saying it 
> cannot define the default font path and doesn't find the font 'fixed'. But
> this is exactly the same font path that Xpmac from pre-R5 finds without
> problems. Weird. 

Nope, expected.  There's some oddity with X servers and X11R6<->XFree86 it
seems.  I -think-.  Might just be problems trying to read the fonts of
course..

> >> If I use XF68_FBDev, then the colors are OK, but the keyboard and the
> >> mouse do not work correctly: The option (alt) keys do not work, and
> >> there is no simulation for mouse buttons 2 and 3.
> >> This is a problem that has been mentioned to the author (Geert U.)
> >> already some time ago, but since he doesn't have ADB devices any more,
> >> it is not clear whether this will be fixed soon.
> >>
> >> So for people with standard Apple keyboards and mice, there remains a
> >> serious problem.
> >
> > Hmm.  Haven't used the XF68_FBDev's that i've built for a while, but they
> > did work fine here.  Right now I'm using Geert's server from 10something
> > run throught fix961212.  Make sure you have Xkb stuff turned off in the
> > XF86Config file.
>  
> The 2 servers from Geert's that I tried didn't want to run on pre-R5. I don't 
> quite know why, because they dump 2 pages of messages on the screen, only the
> second of which is visible. This is a help file. I guess the real errors are
> mentioned on the first, invisible, page. Not very helpful.

Hmm.  Possibly a problem running it on valkfb?  Geert?

> This problem with the non-functioning option keys seems to be the same in
> R4 and pre-R5. And I am pretty sure that the problem is relatively recent.
> I think it was not there at a time when kernel versions were around 
> 2.1.115-121 or so. I don't know how to test this.

Well, I'm outta ideas here.   I've compiled a few FBDevs and they've
worked fine under pre-R5.   I'm using 2.1.131 from kernel.org right now,
but have had no problem with 2.1.121 or greater or so from vger.  Does
anyone else see this problem?

---
Tom Rini (TR1265)
http://dobbstown.yeti.edu/


[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to  Cc linuxppc-dev  if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]

Re: pre-R5&miscs

[Brad Ackerman]

On Sun, 6 Dec 1998, Martin Costabel wrote:

> This is a new installer (Redhat 5.1, I guess). The old one doesn't work
> with the files they put into the base folder. It is so new that
> apparently they have not yet had the time to write instructions. The
> following worked for me:
[^O]

This may work for hard drives, but it doesn't work for install off CD-ROM.
The installer quits after I select the CD-ROM installation option with a
second-stage bootloader "file not found" error.

[There's no vmlinux given for R5b1, so I'm using the R4 installer's copy. 
The CD from which I am trying to install is in ISO-9660/RockRidge format,
made with mkisofs on an IBM RS/6000 SP[1], with the RedHat directory in
the root.]

[1] Not that I think this part makes much difference [except to my CPU
quota], but you never know.

Brad Ackerman N1MNB      "...faced with the men and women who bring home
bsa3@cornell.edu          the pork, voters almost always re-elect them."
http://skaro.pair.com/                   -- _The Economist_, 31 Oct 1998


[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to  Cc linuxppc-dev  if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]

Re: pre-R5&miscs

[Geert Uytterhoeven]

On Mon, 7 Dec 1998, Tom Rini wrote:
> On Mon, 7 Dec 1998, Martin Costabel wrote:
> > The 2 servers from Geert's that I tried didn't want to run on pre-R5. I don't 
> > quite know why, because they dump 2 pages of messages on the screen, only the
> > second of which is visible. This is a help file. I guess the real errors are
> > mentioned on the first, invisible, page. Not very helpful.
> 
> Hmm.  Possibly a problem running it on valkfb?  Geert?

That's quite reasonable. There were some changes in XF68_FBDev that may make it
break on frame buffer devices that aren't 100% compliant. Guess I need to dump
more info with the -verbose flag.

Martin: what about redirecting the X server output to a file, like

    XF68_FBDev >& /tmp/out

Greetings,

						Geert

--
Geert Uytterhoeven                     Geert.Uytterhoeven@cs.kuleuven.ac.be
Wavelets, Linux/{m68k~Amiga,PPC~CHRP}  http://www.cs.kuleuven.ac.be/~geert/
Department of Computer Science -- Katholieke Universiteit Leuven -- Belgium


[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to  Cc linuxppc-dev  if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]

Re: pre-R5&miscs

[Jeff Rugen]

On Mon, 7 Dec 1998, Tom Rini wrote:

> > I tried to run the Xpmac from R4, fixed by fix961212, but it dies saying it 
> > cannot define the default font path and doesn't find the font 'fixed'. But
> > this is exactly the same font path that Xpmac from pre-R5 finds without
> > problems. Weird. 
> 
> Nope, expected.  There's some oddity with X servers and X11R6<->XFree86 it
> seems.  I -think-.  Might just be problems trying to read the fonts of
> course..
> 

I think the X11R6 servers can't decompress the gzipped (.gz) fonts, but they
can decompress compressed (.Z) fonts.  I remember having this problem some
time about with X11R6 and the XFree fonts.

On a side note, I've rebuilt the Xbh server in the XFree86 Servers heirarchy
and it seems to be running ok (though I added the power management function
from XFree and haven't tested it yet).  This runs with the XFree86 3.3.3
distribution (glibc1.99 only -- I don't have the hard drive space to upgrade
as of yet).  In addition, I changed it to open /dev/tty0 instead of
/dev/console so it works with the later 2.1.X kernels, _and_ it has some
tweaks for 32-bit color to work (but there's a lot of flicker when text
scrolls so I don't know if I have the timing values right -- no bitblt at
that depth).

I haven't talked to Chris Nadig (hope I have the name right -- going from
memory) recently -- he did most of the Xbh stuff originally -- so most of
these tweaks are probably only in my code.  I'll make the server available
if people want it, though I don't know a place to put it yet.

In addition, I've been playing with the Xfree86 mono server just to learn
Xfee86 stuff, and as a break (as if I've been playing with it enough to need
a break) from the clgen framebuffer code.  With my luck I'll have the mono
server working first.  Anyway -- about clgen, I have it so I can boot with
that framebuffer enabled, but I can't open a virtual terminal with it (but
remotely logging in I can change the screenmode with fbmode (or whatever
that program was called).  haven't looked into it recently, but I hope to
have more time now.  When trying to install the framebuffer as a module, I
get kernel panics, but I'm wondering if its because I'm trying to send debug
output (printk) to a non-initialized console.

Any comments on the above appreciated.

----------------------------------------------------------------------------
Jeff Rugen                      jrugen@primenet.com

..Had this been an actual emergency, we would have fled in terror, and you 
would not have been informed.


[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to  Cc linuxppc-dev  if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]