vectorlinux-devel — Vector Linux developer's list

Re: [Vectorlinux-devel] Bad checkinstall slack-depends

[John B <joh...@gm...>]

My suggestion is that we investigate what pieces, if any, of the vlapt
package that was supposed to have been deprecated, are still needed.
Include the /sbin/makeslapt script in the checkinstall package. IIRC,
there is a script that goes into /etc/profile.d called vlapt.sh. That
script puts the "useful information" on the console window. Move that
script into the bootscripts.

That's my $0.02 for now.

Cheers,
John

On 10/11/06, Jose J. Rodriguez <jo...@gm...> wrote:
> During a cleanup on Dynamite 5.9.9.1, I nuked vlapt, thinking it was
> the package installer that got deprecated in favour of gslapt. After
> the big spring cleaning, checkinstall fails because there's no
> /sbin/makeslapt. Of course there wasn't, because makeslapt comes with
> the vlapt package, but it took me about an hour to track this down...
> :(. I just checked the dependencies in the checkinstall package and it
> only mentions bash.
>
> Regards,
> Joe1962
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Vectorlinux-devel mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vectorlinux-devel
>

[Vectorlinux-devel] Bad checkinstall slack-depends

[Jose J. R. <jo...@gm...>]

During a cleanup on Dynamite 5.9.9.1, I nuked vlapt, thinking it was
the package installer that got deprecated in favour of gslapt. After
the big spring cleaning, checkinstall fails because there's no
/sbin/makeslapt. Of course there wasn't, because makeslapt comes with
the vlapt package, but it took me about an hour to track this down...
:(. I just checked the dependencies in the checkinstall package and it
only mentions bash.

Regards,
Joe1962

Re: [Vectorlinux-devel] Cannot login into the forum2

[John B <joh...@gm...>]

Kocil,

You were on the list of members waiting to be activated. I've
activated your account. Before you go into the VL forum, clear your
browser's cache and clear out any cookies related to the VL forum just
to be on the safe side. ;-)

Thanks,
John

On 10/8/06, Kocil <k0...@ya...> wrote:
> Hi admin.
>
> I have problem logging in into the new forum.
> It always says
> "Your email address needs to be validated before you
> can login. Need another activation email?"
>
> But when I asked another activation email, it says
> "Error, your account has been activated".
>
> Please help.
> Kocil.
>
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Vectorlinux-devel mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vectorlinux-devel
>

[Vectorlinux-devel] Cannot login into the forum2

[Kocil <k0...@ya...>]

Hi admin.

I have problem logging in into the new forum.
It always says
"Your email address needs to be validated before you
can login. Need another activation email?"

But when I asked another activation email, it says
"Error, your account has been activated".

Please help.
Kocil.


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Re: [Vectorlinux-devel] Repo suggestions for discussion

[Jose J. R. <jo...@gm...>]

On 10/5/06, John B <joh...@gm...> wrote:
> Here are my $2.00 on the matter.
>
> On 10/5/06, Jose J. Rodriguez <jo...@gm...> wrote:
> > During the repo overhaul I was doing today I came up with some
> > suggestions and questions about possible changes to the repo
> > structure. I'd like to discuss them here:
> >
> > - rename testing/ to unstable/ (this should be for development (cvs,
> > alpha, beta) versions of apps, libs, etc.)
> > - create a new, possibly unstructured testing/ category, but only for
> > temp storage of packages to be checked for correctness.
>
> Maybe just keep testing as is and add an unstable directory for cvs stuff, etc.
>

Could be too, but sounds more logical the other way around, IMHO.

>
> > - unclassified/ should be holding place till decided whether to put in
> > unstable or extras.
> > - extras/unclassified/: temp holding place (try not to abuse this).
> > - unstable/unclassified/: temp holding place (try not to abuse this).
> > - packages/unclassified/: temp holding place (try not to abuse this).
>
> The only issue here is that the makeinfo-slapt and makeinfo-slack
> scripts traverse the directory tree from the packages level. If we're
> going to have an "unclassified" directory, I think it would be best to
> keep it at the same level as the "packages" directory. That way, those
> unclassified items aren't made available via slapt-get or gslapt until
> decisions are made.
>

Agreed, I was leaning in this direction myself.

>
> > - rename packages to stable ???
>
> This might not matter too much.
>

Maybe release or a similar name is better, see comment below regarding
dist-upgrade for more on this.

>
> > - rename base/ to libs/ (because stable/ already means the package is
> > in the iso and will be used in distro-upgrade, also base and base-apps
> > would be misleading in extra/ and unstable/) ???
> > - rename base-apps/ to apps/ (same reason as above) ???
>
> The major problem with making those two changes is that I have patched
> slapt-get for VL Dynamite so that the dist-upgrade mechanism will
> work. I've used base$ as the regex for slapt-get's searches. (For the
> slack repos, a$ is the search regex, which would search both the a and
> ap directories.)
>
<snip>
>
> As Jason explained to me, slapt-get needs to parse the a and ap
> directories for Slackware (analogous to our base and base-apps
> directories) so that the packages that are the real base of the distro
> are upgraded correctly. Of course this also means we would need to
> have all of the packages that are used on the ISO in the repository.
>

As I understood it in earlier discussions, the packages repo (or
stable, or whatever) was only for those apps included in the distro
and that would be subject to a distro-upgrade. Everything else would
be in extras. So maybe a name like "release" (or any other that helps
to bring that idea across) would be more understandable. Anyway, if a
and ap (base and base-apps) are not analogous to the cli libs and apps
released with the iso, then I understand your point. We also need to
decide if this repo category will include just the packages that make
up the "bulks", or also the optionally selectable packages that are
listed later on during install.

Regards,
Joe1962

Re: [Vectorlinux-devel] Repo suggestions for discussion

[John B <joh...@gm...>]

Here are my $2.00 on the matter.

On 10/5/06, Jose J. Rodriguez <jo...@gm...> wrote:
> During the repo overhaul I was doing today I came up with some
> suggestions and questions about possible changes to the repo
> structure. I'd like to discuss them here:
>
> Suggestions:
>
> - symlink cleanold, makeinfo-slack, makeinfo-slapt from each category
> to centralize fixes.

This makes sense. I would suggest that the master scripts be kept in
the /home/vectorlinux directory rather than in the repo proper.

> - rename testing/ to unstable/ (this should be for development (cvs,
> alpha, beta) versions of apps, libs, etc.)
> - create a new, possibly unstructured testing/ category, but only for
> temp storage of packages to be checked for correctness.

Maybe just keep testing as is and add an unstable directory for cvs stuff, etc.

> - rename kdei/ to kdei18/

No problem with that.

> - unclassified/ should be holding place till decided whether to put in
> unstable or extras.
> - extras/unclassified/: temp holding place (try not to abuse this).
> - unstable/unclassified/: temp holding place (try not to abuse this).
> - packages/unclassified/: temp holding place (try not to abuse this).

The only issue here is that the makeinfo-slapt and makeinfo-slack
scripts traverse the directory tree from the packages level. If we're
going to have an "unclassified" directory, I think it would be best to
keep it at the same level as the "packages" directory. That way, those
unclassified items aren't made available via slapt-get or gslapt until
decisions are made.

> - move the iso directories outside the package repos, to a
> version-separated structure inside a top-level iso/ category.
>

I have mixed thoughts on this one, so I'll pass for now.

>
> Questions:
>
> - rename packages to stable ???

This might not matter too much.

> - rename base/ to libs/ (because stable/ already means the package is
> in the iso and will be used in distro-upgrade, also base and base-apps
> would be misleading in extra/ and unstable/) ???
> - rename base-apps/ to apps/ (same reason as above) ???

The major problem with making those two changes is that I have patched
slapt-get for VL Dynamite so that the dist-upgrade mechanism will
work. I've used base$ as the regex for slapt-get's searches. (For the
slack repos, a$ is the search regex, which would search both the a and
ap directories.) Also see the following from the slapt-get FAQ located
at http://software.jaos.org/BUILD/slapt-get/FAQ.html#slgFAQ9 -

9.  Can I "dist-upgrade" to a newer Slackware release?

 Yes. See also http://slackwiki.org/Upgrade_Using_Slapt-get.

 Change your source location within /etc/slapt-get/slapt-getrc to point to
 the newer release directory.  --update your local package cache, then
 --dist-upgrade to the newer release.  You will also want to disable any
 third party package sources while upgrading to a newer release due to the fact
 that third party packages might not be as rigorously tested as official
 packages, or may cause conflicts with files provided by official packages.

 For instance, you have the following line as your source:

   SOURCE=ftp://distro.ibiblio.org/pub/Linux/distributions/slackware/slackware-9.0/

 Simply change the URL to point to current, like so:

   SOURCE=ftp://distro.ibiblio.org/pub/Linux/distributions/slackware/slackware-current/

 CAVEAT: Follow the instructions in UPGRADE.TXT.

 Now you --dist-upgrade to retrieve the package data for that release.
 What dist-upgrade does is basically make sure that any missing packages
 from the base disk set are installed, as well as upgrade to any newer
 versions of the currently installed packages.  All of this happens in
 one transaction.

 dist-upgrade'ing involves doing the following:

   ### upgrade your local package cache
   $ slapt-get --update

   ### first, download all of the packages prior to upgrading
   $ slapt-get --dist-upgrade --download-only

   ### second, install (not upgrade!) the newest kernel
   $ slapt-get --install kernel --no-upgrade

   ### then, upgrade the most important
   #(this installs any newer versions)
   $ slapt-get --install glibc-solibs pkgtools sed

 In newer versions of slapt-get, dist-upgrade automatically handles
 installing/upgrading glibc-solibs, pkgtools, and sed prior to any
 other changes.

   ### finally, let slapt-get upgrade the rest
   $ slapt-get --dist-upgrade

  You can specify --remove-obsolete to remove all depreciated packages.
  CAUTION: this can only take into consideration packages that are available
  from your current sources.  Thus self made packages will be considered
  obsolete as well (they can be excluded within slapt-getrc, however).  Only
  use this if you have been using Slackware packages exclusively.  You can
  also use this option for information purposes, observing the obsolete
  packages, then passing all the ones you are sure of to --remove (which can
  accept multiple packages at once).
   ### dist upgrade and remove all depreciated/obsolete packages
   $ slapt-get --dist-upgrade --remove-obsolete

 Then follow the rest of the directions in UPGRADE.TXT.

 You should make sure any new packages within the disk sets you are using are
 also installed, as they will not be detected during --dist-upgrade, as only
 packages already installed will be upgraded.  To accomplish this, you can
 specify slapt-get to install a disk set like so:

   ### install all X and X application packages
   $ slapt-get --install-set x xap

 Do the same for all your installed disk sets, such as gnome, x, xap, l, n, etc.

As Jason explained to me, slapt-get needs to parse the a and ap
directories for Slackware (analogous to our base and base-apps
directories) so that the packages that are the real base of the distro
are upgraded correctly. Of course this also means we would need to
have all of the packages that are used on the ISO in the repository.


>
> Regards,
> Joe1962

Again, just my $2.00,
JohnB

Re: [Vectorlinux-devel] Repo suggestions for discussion

[Jose J. R. <jo...@gm...>]

On 10/5/06, Jose J. Rodriguez <jo...@gm...> wrote:
> During the repo overhaul I was doing today I came up with some
> suggestions and questions about possible changes to the repo
> structure. I'd like to discuss them here:
>
> Questions:
>
> - rename base/ to libs/

Obviously, this should have been move, not rename, because libs/ already exists.

Regards,
Joe1962

[Vectorlinux-devel] Repo suggestions for discussion

[Jose J. R. <jo...@gm...>]

During the repo overhaul I was doing today I came up with some
suggestions and questions about possible changes to the repo
structure. I'd like to discuss them here:

Suggestions:

- symlink cleanold, makeinfo-slack, makeinfo-slapt from each category
to centralize fixes.
- rename testing/ to unstable/ (this should be for development (cvs,
alpha, beta) versions of apps, libs, etc.)
- create a new, possibly unstructured testing/ category, but only for
temp storage of packages to be checked for correctness.
- rename kdei/ to kdei18/
- unclassified/ should be holding place till decided whether to put in
unstable or extras.
- extras/unclassified/: temp holding place (try not to abuse this).
- unstable/unclassified/: temp holding place (try not to abuse this).
- packages/unclassified/: temp holding place (try not to abuse this).
- move the iso directories outside the package repos, to a
version-separated structure inside a top-level iso/ category.


Questions:

- rename packages to stable ???
- rename base/ to libs/ (because stable/ already means the package is
in the iso and will be used in distro-upgrade, also base and base-apps
would be misleading in extra/ and unstable/) ???
- rename base-apps/ to apps/ (same reason as above) ???

Regards,
Joe1962

[Vectorlinux-devel] Repo overhaul

[Jose J. R. <jo...@gm...>]

I've been doing some extensive overhauling of the vectorlinux-current
repo, in preparation for going to Standard beta:

- created a skeleton/ directory, to contain the empty directory tree
that we should use in each repo category, this allowed me to easily
copy the structure over to the others.

- copied the skeleton structure to each of the working repo
categories: packages/, extra/, testing/.

- created packages/, extra/, and testing/ directories inside the
source/ category and reorganized the source packages accordingly.

- copied the skeleton structure to each of the directories inside the
source/ category: source/packages/, source/extra/, source/testing/.

- created an unclassified/ category as temporary holding place for
packages if not sure where to classify them. Also created
unclassified/ directories inside packages/, extra/ and testing/.

- renamed packages in extra/ and testing/ that were still "vl6" to "vl58".

- moved gambas2 packages to testing/x-dev/ as this is an unstable
development version.

After al that, I ran both update scripts (with the "all" option)
inside each repo category and then the trigger script.

Regards,
Joe1962

[Vectorlinux-devel] ATI cards testing summary.

[<nig...@ju...>]

Just to summarize for now: After numerous installs on my test system
these results have been consistent:

Both the old Radeon 9500 Pro and a new 9250 lock up when running smp +
ati or fglrx.

The 9250 was a disappointment. 3D accel using "ati" gives corrupted
display in Google Earth (reason for wanting 3D). With "fglrx" 8.28.8 it
does not run at all, just hangs on splash screen. I tried downgrading
to 8.26.18, but this one gave me a long list of "API ERROR: could not
register entrypoint for *********.." The result was the same between VL
5.1 SOHO and VLD 5.9.9.x. This card does not have a fan and runs very
hot with 3D enabled.

The 9500 with nonsmp and fglrx does fine. Produces a lot of heat, but
the fan is enough to keep it stable.

That's about it for now. For further testing of VLD 5.9.9.1 I plan on
using an nV GF6200, which works very well when using smp and 3D.


________________________________________________________________________=

Try Juno Platinum for Free! Then, only $9.95/month!
Unlimited Internet Access with 1GB of Email Storage.
Visit http://www.juno.com/value to sign up today!

Re: [Vectorlinux-devel] VL5.8 revised docs

[Sriram D. <sri...@gm...>]

Larry,
 i would love to review them for you if you think that is necessary before
letting them loose in the wild.
cheers
ram

On 9/12/06, Larry Gagnon <lgg...@un...> wrote:
>
> LLL and I have worked part-time on the VL5 documentation files over the
> past few months. I think they are ready for inclusion in VL5.8 Standard.
> There has been a great improvement in a number of the sections,
> particularly re slapt-get and gslapt. Also I have updated the mirror
> sites, amongst many other things.
>
> Where shall I upload the docs for inclusion in 5.8???
>
> lagagnon
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Vectorlinux-devel mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vectorlinux-devel
>

[Vectorlinux-devel] VL5.8 revised docs

[Larry G. <lgg...@un...>]

LLL and I have worked part-time on the VL5 documentation files over the 
past few months. I think they are ready for inclusion in VL5.8 Standard. 
There has been a great improvement in a number of the sections, 
particularly re slapt-get and gslapt. Also I have updated the mirror 
sites, amongst many other things.

Where shall I upload the docs for inclusion in 5.8???

lagagnon

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[<nig...@ju...>]

I will try ver. 8.26.18.

-And, to clarify.. when trying to install the new video card, I did
also try hardware autodetection. Also tried a GF6200, same result.

Roy

-- "Jose J. Rodriguez" <jo...@gm...> wrote:
> Well, I uninstalled ATI driver 8.27.10 and installed 8.26.18 and the
> end session hard lockup problem seems to be gone.


________________________________________________________________________=

Try Juno Platinum for Free! Then, only $9.95/month!
Unlimited Internet Access with 1GB of Email Storage.
Visit http://www.juno.com/value to sign up today!

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[Jose J. R. <jo...@gm...>]

On 9/8/06, nig...@ju... <nig...@ju...> wrote:
> I was using 8.28.8
>

That's a bit big (double the size as the previous release). Haven't
downloaded it.

>
> However, the reason for wanting hardware acceleration, Google Earth,
> has problems. Using the open source "ati" driver, it runs, but with
> corrupted display (ghosting of previous images and streaking). With
> "fglrx" it will not run at all, doesn't get past the splash screen.
> Other 3D apps like ppracer and scorched3d ran very well.
>

Google Earth works fine here wth both fglrx drivers I've used.

> - On a side note, neither VLD nor SOHO would set up the new video card
> using vasm > xwindow > xconf. It just quickly flashed and dropped back.
> On fresh install both picked it up. Is there a way to totally reset the
> hardware data or force a deeper probe?
>

Not sure why it works differently after install.

>
>
> -- "Jose J. Rodriguez" <jo...@gm...> wrote:
> Well, I uninstalled ATI driver 8.27.10 and installed 8.26.18 and the
> end session hard lockup problem seems to be gone. I sort of figured
> the problem started when I upgraded to 8.27.10. What driver version
> have you tested nightflier?
>
> Regards,
> Joe1962
>

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[<nig...@ju...>]

I was using 8.28.8

For further testing, I got another card with the older architecture, a
Radeon 9250, and tested on clean installs of both VLD 5.9.9 and VL SOHO
5.1.

With this one I saw no lockups, "ati" or "fglrx".

However, the reason for wanting hardware acceleration, Google Earth,
has problems. Using the open source "ati" driver, it runs, but with
corrupted display (ghosting of previous images and streaking). With
"fglrx" it will not run at all, doesn't get past the splash screen.
Other 3D apps like ppracer and scorched3d ran very well.

- On a side note, neither VLD nor SOHO would set up the new video card
using vasm > xwindow > xconf. It just quickly flashed and dropped back.
On fresh install both picked it up. Is there a way to totally reset the
hardware data or force a deeper probe?

Roy


-- "Jose J. Rodriguez" <jo...@gm...> wrote:
Well, I uninstalled ATI driver 8.27.10 and installed 8.26.18 and the
end session hard lockup problem seems to be gone. I sort of figured
the problem started when I upgraded to 8.27.10. What driver version
have you tested nightflier?

Regards,
Joe1962


________________________________________________________________________=

Try Juno Platinum for Free! Then, only $9.95/month!
Unlimited Internet Access with 1GB of Email Storage.
Visit http://www.juno.com/value to sign up today!

[Vectorlinux-devel] Announcing new cpu-freq utility

[Jose J. R. <jo...@gm...>]

Check out the new cpu-freq utility made in Gambas:

http://jjrweb.webhop.net/vl/dev/cpu-freq/

You can run it on VL 5.1 using the gambas-runtime-full-1.0.15 package
available in the VL repo, or on the latest Dynamite which should
include (I think) the new gambas-runtime-1.0.17 package.

Regards,
Joe1962

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[Jose J. R. <jo...@gm...>]

Well, I uninstalled ATI driver 8.27.10 and installed 8.26.18 and the
end session hard lockup problem seems to be gone. I sort of figured
the problem started when I upgraded to 8.27.10. What driver version
have you tested nightflier?

Regards,
Joe1962

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[uel a. <ue...@gm...>]

i put a post to the forum looking for similar hardware so we can see if this
problem can be duplicated
also I have built a non-smp kernel for the next dynamite hopefully we can
figure this out

Uel

On 9/6/06, Jose J. Rodriguez <jo...@gm...> wrote:
>
> On 9/6/06, Jose J. Rodriguez <jo...@gm...> wrote:
> > On 9/6/06, nig...@ju... <nig...@ju...> wrote:
> > > Not sure if I'm doing this right.. In lilo.conf, I changed the line:
> > > append = ""
> > > to this:
> > > append = "nosmp"
> > > or:
> > > append = "maxcpus=0"
> > >
> > > and then ran "lilo"
> > >
> > > In both cases, I get some kind of IDE errors:
> > > --- snip--
> > > ide1 at 0x170-0x177, 0x376 on irq 15
> > > hda: max request size: 128 KiB
> > > hda: lost interrupt             <------- here is where it slows down--
> > > hda: lost interrupt             <----- no response to keyboard---
> > > hda: lost interrupt
> > > hda: lost interrupt
> > > hda: (xxxxxxx sectors, size, cache, UDMA status...)
> > > hda: cache flush supported
> > > hda:<4>hda: dma_timer_expiry: dma status == 0x661
> > > hda: lost interrupt
> > >  hda1:
> > > hda: lost interrupt
> > > .......
> > >
> > > It repeats this for hdb and hdc, after around 5 minutes, about halfway
> > > through hdc the screen goes blank and nothing more happens.
> > >
> > > Are there variances in the syntax I should try?
> > >
> >
> > Sounds right to me. Anyway, just found another one specifically for HT:
> >
> > "The `noht' Argument:
> > This will disable hyper-threading on intel processors that have this
> feature."
> >
>
> After some testing and more searching, "noht" seems to be gone from
> current kernels and "nosmp" gives me the same problem as you get.
>
> Funny thing is, I've been getting a hard lockup lately when closing
> the KDE session. KDM starts to show up, but it will not get as far as
> showing any text or user avatars. Only a hard switch-off will get me
> out of that. I'm not sure when it started, but I suspect after
> upgrading the propietary ATI driver from one version to the next.
> Can't be quite sure though. A search shows this is something that
> happens now and again in various KDE versions, but I couldn't find a
> definite explanation for any case. Anyway, this happens the same with
> HT enabled or disabled in the BIOS. I'll have to experiment with my
> previous ATI driver, or the new one that was just released.
>
> I still don't see the kind of X unresponsiveness you get though. I'll
> have to experiment extensively on the next Dynamite, as it seems to be
> an xorg 6.9 problem and I have 6.8.2 on my SOHO.
>
> Regards,
> Joe1962
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Vectorlinux-devel mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vectorlinux-devel
>

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[Jose J. R. <jo...@gm...>]

On 9/6/06, Jose J. Rodriguez <jo...@gm...> wrote:
> On 9/6/06, nig...@ju... <nig...@ju...> wrote:
> > Not sure if I'm doing this right.. In lilo.conf, I changed the line:
> > append = ""
> > to this:
> > append = "nosmp"
> > or:
> > append = "maxcpus=0"
> >
> > and then ran "lilo"
> >
> > In both cases, I get some kind of IDE errors:
> > --- snip--
> > ide1 at 0x170-0x177, 0x376 on irq 15
> > hda: max request size: 128 KiB
> > hda: lost interrupt             <------- here is where it slows down--
> > hda: lost interrupt             <----- no response to keyboard---
> > hda: lost interrupt
> > hda: lost interrupt
> > hda: (xxxxxxx sectors, size, cache, UDMA status...)
> > hda: cache flush supported
> > hda:<4>hda: dma_timer_expiry: dma status == 0x661
> > hda: lost interrupt
> >  hda1:
> > hda: lost interrupt
> > .......
> >
> > It repeats this for hdb and hdc, after around 5 minutes, about halfway
> > through hdc the screen goes blank and nothing more happens.
> >
> > Are there variances in the syntax I should try?
> >
>
> Sounds right to me. Anyway, just found another one specifically for HT:
>
> "The `noht' Argument:
> This will disable hyper-threading on intel processors that have this feature."
>

After some testing and more searching, "noht" seems to be gone from
current kernels and "nosmp" gives me the same problem as you get.

Funny thing is, I've been getting a hard lockup lately when closing
the KDE session. KDM starts to show up, but it will not get as far as
showing any text or user avatars. Only a hard switch-off will get me
out of that. I'm not sure when it started, but I suspect after
upgrading the propietary ATI driver from one version to the next.
Can't be quite sure though. A search shows this is something that
happens now and again in various KDE versions, but I couldn't find a
definite explanation for any case. Anyway, this happens the same with
HT enabled or disabled in the BIOS. I'll have to experiment with my
previous ATI driver, or the new one that was just released.

I still don't see the kind of X unresponsiveness you get though. I'll
have to experiment extensively on the next Dynamite, as it seems to be
an xorg 6.9 problem and I have 6.8.2 on my SOHO.

Regards,
Joe1962

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[Jose J. R. <jo...@gm...>]

On 9/6/06, nig...@ju... <nig...@ju...> wrote:
> Not sure if I'm doing this right.. In lilo.conf, I changed the line:
> append = ""
> to this:
> append = "nosmp"
> or:
> append = "maxcpus=0"
>
> and then ran "lilo"
>
> In both cases, I get some kind of IDE errors:
> --- snip--
> ide1 at 0x170-0x177, 0x376 on irq 15
> hda: max request size: 128 KiB
> hda: lost interrupt             <------- here is where it slows down--
> hda: lost interrupt             <----- no response to keyboard---
> hda: lost interrupt
> hda: lost interrupt
> hda: (xxxxxxx sectors, size, cache, UDMA status...)
> hda: cache flush supported
> hda:<4>hda: dma_timer_expiry: dma status == 0x661
> hda: lost interrupt
>  hda1:
> hda: lost interrupt
> .......
>
> It repeats this for hdb and hdc, after around 5 minutes, about halfway
> through hdc the screen goes blank and nothing more happens.
>
> Are there variances in the syntax I should try?
>

Sounds right to me. Anyway, just found another one specifically for HT:

"The `noht' Argument:
This will disable hyper-threading on intel processors that have this feature."

HTH.

Joe1962

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[<nig...@ju...>]

Not sure if I'm doing this right.. In lilo.conf, I changed the line:
append =3D "" =

to this:
append =3D "nosmp"
or:
append =3D "maxcpus=3D0"

and then ran "lilo"

In both cases, I get some kind of IDE errors:
--- snip--
ide1 at 0x170-0x177, 0x376 on irq 15
hda: max request size: 128 KiB
hda: lost interrupt             <------- here is where it slows down--
hda: lost interrupt             <----- no response to keyboard---
hda: lost interrupt =

hda: lost interrupt =

hda: (xxxxxxx sectors, size, cache, UDMA status...)
hda: cache flush supported
hda:<4>hda: dma_timer_expiry: dma status =3D=3D 0x661
hda: lost interrupt =

 hda1:
hda: lost interrupt =

=2E......

It repeats this for hdb and hdc, after around 5 minutes, about halfway
through hdc the screen goes blank and nothing more happens. =


Are there variances in the syntax I should try?

Roy


>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Found it! Can you try one of these?:

The `nosmp' Argument:
Use of this option will tell a SMP kernel on a SMP machine to operate
single processor.  Typically only used for debugging and determining
if a particular problem is SMP related.

The `maxcpus=3D' Argument:
The number given with this argument limits the maximum number of CPUs
activated in SMP mode.  Using a value of 0 is equivalent to the nosmp
option.

Regards,
Joe1962



________________________________________________________________________=

Try Juno Platinum for Free! Then, only $9.95/month!
Unlimited Internet Access with 1GB of Email Storage.
Visit http://www.juno.com/value to sign up today!

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[Jose J. R. <jo...@gm...>]

On 9/5/06, Jose J. Rodriguez <jo...@gm...> wrote:
> On 9/5/06, nig...@ju... <nig...@ju...> wrote:
> > No problem, Joe. I'm better at testing and complaining than fixing, so
> >  I'm just glad y'all put up with me.
> >
> > I tried the patch that Uel pointed out and saw no improvement.
> >
> > I have yet to notice any difference in speed when running HT on any OS.
> > Turning it off is no sacrifice for me personally, but maybe we should
> > keep a non-smp kernel available in the repository for those cases where
> > it may be of help?
> >
>
> I wonder if a simple kernel parameter given in lilo might do it? I'll
> investigate...
>

Found it! Can you try one of these?:

The `nosmp' Argument:
Use of this option will tell a SMP kernel on a SMP machine to operate
single processor.  Typically only used for debugging and determining
if a particular problem is SMP related.

The `maxcpus=' Argument:
The number given with this argument limits the maximum number of CPUs
activated in SMP mode.  Using a value of 0 is equivalent to the nosmp
option.

Regards,
Joe1962

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[Jose J. R. <jo...@gm...>]

On 9/5/06, nig...@ju... <nig...@ju...> wrote:
> No problem, Joe. I'm better at testing and complaining than fixing, so
>  I'm just glad y'all put up with me.
>
> I tried the patch that Uel pointed out and saw no improvement.
>
> I have yet to notice any difference in speed when running HT on any OS.
> Turning it off is no sacrifice for me personally, but maybe we should
> keep a non-smp kernel available in the repository for those cases where
> it may be of help?
>

I wonder if a simple kernel parameter given in lilo might do it? I'll
investigate...

Regards,
Joe1962

Re: [Vectorlinux-devel] VLdynamite locking up using ATI driver.

[<nig...@ju...>]

No problem, Joe. I'm better at testing and complaining than fixing, so
 I'm just glad y'all put up with me.

I tried the patch that Uel pointed out and saw no improvement.

I have yet to notice any difference in speed when running HT on any OS.
Turning it off is no sacrifice for me personally, but maybe we should
keep a non-smp kernel available in the repository for those cases where
it may be of help?

Roy


-- "Jose J. Rodriguez" <jo...@gm...> wrote:
Oh, I hope I didn't come across as defensive. Never meant to. I was
just pointing out who to blame, lol.

Regards,
Joe1962


________________________________________________________________________=

Try Juno Platinum for Free! Then, only $9.95/month!
Unlimited Internet Access with 1GB of Email Storage.
Visit http://www.juno.com/value to sign up today!

Re: [Vectorlinux-devel] vl-hot 0.3.6 with vl-hot-configurator!

[Jose J. R. <jo...@gm...>]

A question: what is the proper way to get a list of charsets and
codepages? I'm currently getting them from "/lib/modules/" & uname -r
&  "/kernel/fs/nls" but what if they were compiled in, instead of as
modules?

Regards,
Joe1962