Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Love-sources - get them before they go obsolete
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3, 4, 5, 6, 7  Next  
Reply to topic    Gentoo Forums Forum Index Off the Wall
View previous topic :: View next topic  
Author Message
lurid
Guru
Guru


Joined: 12 Mar 2003
Posts: 595
Location: Florida

PostPosted: Tue Sep 09, 2003 4:27 pm    Post subject: Reply with quote

I didn't actually write the patch, nor do I maintain it. I just think its probably one of the best patches to come along in a while. Especially since anyone using an nVidia card is stuck at 60hz for anything console related. The README within the patch itself documents the patch process and editing the vesafb_modes.h file very well. I'm assuming that anyone using a dev kernel, and especially this patchset, are fairly advanced users and should be able to figure it out.

For the record, I did switch monitors while having this patch installed, but it was to a better one so it had no problems displaying my old vesamode other than the picture being off center a bit. The patch is avalible here. Again though, I'm not the auther so you might want to contact spock_ about packaging it.
_________________
Go find a cheerleader and saw her legs off. - Nny
Back to top
View user's profile Send private message
milothurston
Apprentice
Apprentice


Joined: 01 May 2002
Posts: 231
Location: Oxford, England.

PostPosted: Wed Sep 10, 2003 2:13 pm    Post subject: Reply with quote

This kernel compiled and booted nicely, but the machine could not bring itself to run Word under cxoffice when running under this kernel.
If I were not obliged to edit a manuscript containing Endnote references, this would not be a problem :wink:
Milo
Back to top
View user's profile Send private message
milothurston
Apprentice
Apprentice


Joined: 01 May 2002
Posts: 231
Location: Oxford, England.

PostPosted: Wed Sep 10, 2003 2:13 pm    Post subject: Reply with quote

This kernel compiled and booted nicely, but the machine could not bring itself to run Word under cxoffice when running under this kernel.
If I were not obliged to edit a manuscript containing Endnote references, this would not be a problem :wink:
Milo
Back to top
View user's profile Send private message
milothurston
Apprentice
Apprentice


Joined: 01 May 2002
Posts: 231
Location: Oxford, England.

PostPosted: Wed Sep 10, 2003 2:19 pm    Post subject: Reply with quote

This kernel compiled and booted nicely, but the machine could not bring itself to run Word under cxoffice when running under this kernel.
If I were not obliged to edit a manuscript containing Endnote references, this would not be a problem :wink:
Milo
Back to top
View user's profile Send private message
Lovechild
Advocate
Advocate


Joined: 17 May 2002
Posts: 2858
Location: Århus, Denmark

PostPosted: Wed Sep 10, 2003 2:23 pm    Post subject: Reply with quote

I don't own crossover office so I'm unable to confirm this - does that program work with any other 2.6 kernel ?
Back to top
View user's profile Send private message
milothurston
Apprentice
Apprentice


Joined: 01 May 2002
Posts: 231
Location: Oxford, England.

PostPosted: Wed Sep 10, 2003 2:27 pm    Post subject: Reply with quote

Lovechild wrote:
I don't own crossover office so I'm unable to confirm this - does that program work with any other 2.6 kernel ?


Yes - I'm currently using it with test4-mm4, and I've been using it since the 2.4s came out.
Word* would simply hang, with no apparent CPU load. If I get time, I'll try without the CFQ scheduler to see if that did it.
Milo.

* I don't use this through personal choice, of course.
Back to top
View user's profile Send private message
Lovechild
Advocate
Advocate


Joined: 17 May 2002
Posts: 2858
Location: Århus, Denmark

PostPosted: Wed Sep 10, 2003 2:31 pm    Post subject: Reply with quote

if removing the CFQ scheduler it might be a good idea to test with vanilla test5 since Linus might have included something nasty.
Back to top
View user's profile Send private message
lurid
Guru
Guru


Joined: 12 Mar 2003
Posts: 595
Location: Florida

PostPosted: Wed Sep 10, 2003 3:21 pm    Post subject: Reply with quote

Ok, I was having a few problems with XMMS with this kernel. At first, about 5 seconds into a song it would start getting garbled. Switching from OSS emulation to the actual Alsa driver fixed that. Not a kernel issue. However, after that I was getting really small clicks with internet streams. Mp3s off the hd were fine, but I listen to a lot of Live365 stations and so this was starting to annoy me. Not nearly as bad as the browser/xmms skips in the early dev phase, but still noticable.

Anyway, I tracked it down to the CFQ elevator. I had 'elevator=cfq' in my grub.conf, when I removed it the XMMS clicks completly went away. What gets me is that just now when I went to make this post I noticed in Lovechilds original post that it says CFQ is used by default. 0_o So really, I have no idea which elevator is being used now, but either way specifying CFQ in grub.conf makes my XMMS skip.

EDIT: Feh. Spoke too soon. The clicks have reared their ugly head again. I wonder if they messed with any ALSA stuff in test4. This is my first test4 kernel so I'm more apt to blame that than this patchset. There is a love patch for test5 now, so I supposed I'll upgrade to that (even though I just installed test4 yesterday :P) and see if that fixes the problem.
_________________
Go find a cheerleader and saw her legs off. - Nny
Back to top
View user's profile Send private message
shm
Advocate
Advocate


Joined: 09 Dec 2002
Posts: 2380
Location: Atlanta, Universe

PostPosted: Wed Sep 10, 2003 4:08 pm    Post subject: Reply with quote

for a second, I thought you were distributing the old caldera/SCO linux kernel, named after Ransom Love. :twisted:

thanks for the kernel though, compiling now.
_________________
what up
Back to top
View user's profile Send private message
Lovechild
Advocate
Advocate


Joined: 17 May 2002
Posts: 2858
Location: Århus, Denmark

PostPosted: Wed Sep 10, 2003 4:16 pm    Post subject: Reply with quote

shm wrote:
for a second, I thought you were distributing the old caldera/SCO linux kernel, named after Ransom Love. :twisted:

thanks for the kernel though, compiling now.


OH SHIT !!!

SCO had a kernel named love... I'm so screwed now.. who will chip in for the legal costs ?
Back to top
View user's profile Send private message
bssteph
l33t
l33t


Joined: 26 Feb 2003
Posts: 651
Location: Wisconsin

PostPosted: Wed Sep 10, 2003 5:05 pm    Post subject: Reply with quote

YOU NAME THIEF! You're doubly in trouble now.

Yeah, anyway. Well done on the patchset again, Lovechild. My laptop is liking test5-love1
One regression though..
Code:
[drm:radeon_cp_init] *ERROR* radeon_cp_init called without lock held
[drm:radeon_unlock] *ERROR* Process 3639 using kernel context 0

My Radeon Mobility with the DRM drivers in the kernel done broked. Any suggestions? (forums search turned up a couple similar problems but no concrete fix. Nothing on LKML that i could see aside from radeonfb stuff.)
It's a bit crusty but DRM worked fine in my last kernel on here, test3-mm3.

In the middle of building other kernels to narrow it down but I'm back and forth with uni work.

[EDIT: errors in vanilla test5 and test5-mm1 as well, will look at it some more]
Back to top
View user's profile Send private message
Lovechild
Advocate
Advocate


Joined: 17 May 2002
Posts: 2858
Location: Århus, Denmark

PostPosted: Wed Sep 10, 2003 7:42 pm    Post subject: Reply with quote

Well, I guess this would be an issue with DRM which I haven't touched - so unless it's getting attention in mainstream I would think the best option would be to report a bug to lkml on this.
Back to top
View user's profile Send private message
Cyrana
n00b
n00b


Joined: 12 Aug 2003
Posts: 21

PostPosted: Wed Sep 10, 2003 8:15 pm    Post subject: Reply with quote

Which Radeon Mobility do you have? And this is the kernel drm right, not the ati-drivers?

Just asking since on Test5 (no love patches on this box yet) it works fine for me. Mobility M7 (Radeon 7500 Mobility).

bssteph wrote:
YOU NAME THIEF! You're doubly in trouble now.

Yeah, anyway. Well done on the patchset again, Lovechild. My laptop is liking test5-love1
One regression though..
Code:
[drm:radeon_cp_init] *ERROR* radeon_cp_init called without lock held
[drm:radeon_unlock] *ERROR* Process 3639 using kernel context 0

My Radeon Mobility with the DRM drivers in the kernel done broked. Any suggestions? (forums search turned up a couple similar problems but no concrete fix. Nothing on LKML that i could see aside from radeonfb stuff.)
It's a bit crusty but DRM worked fine in my last kernel on here, test3-mm3.

In the middle of building other kernels to narrow it down but I'm back and forth with uni work.

[EDIT: errors in vanilla test5 and test5-mm1 as well, will look at it some more]
Back to top
View user's profile Send private message
lurid
Guru
Guru


Joined: 12 Mar 2003
Posts: 595
Location: Florida

PostPosted: Wed Sep 10, 2003 8:53 pm    Post subject: Reply with quote

test5-love1 fixes the skip problem. I'm going to chalk it up to test4 doing something screwy with alsa.
_________________
Go find a cheerleader and saw her legs off. - Nny
Back to top
View user's profile Send private message
Lovechild
Advocate
Advocate


Joined: 17 May 2002
Posts: 2858
Location: Århus, Denmark

PostPosted: Wed Sep 10, 2003 9:07 pm    Post subject: Reply with quote

lurid wrote:
test5-love1 fixes the skip problem. I'm going to chalk it up to test4 doing something screwy with alsa.


Rather than worry about why this issue was there - I'll close my eyes and be happy it's gone now..

Ignorance is bliss....
Back to top
View user's profile Send private message
bssteph
l33t
l33t


Joined: 26 Feb 2003
Posts: 651
Location: Wisconsin

PostPosted: Wed Sep 10, 2003 10:22 pm    Post subject: Reply with quote

Oh, don't worry about me, either I'm an idiot or oldconfig hates me. In my class w/out network jacks I noticed that AGPGART was dropped from my config. :oops:

(odds are that I'm an idiot, say my chums) :lol:
Back to top
View user's profile Send private message
zypher
Retired Dev
Retired Dev


Joined: 10 Jun 2002
Posts: 416
Location: Cologne, ger.

PostPosted: Thu Sep 11, 2003 7:22 am    Post subject: Reply with quote

lovechild: I get an error-message when patching the kernel-tree.
It says something like "unexpected end of patchfile"
I downloaded your patch twice. It ends with:

Code:

+}
+
 /**
  * invalidate_mapping_pages - Invalidate all the unlocked pages of one inode
  * @mapping: the address_space which holds the pages to invalidate
 


Is something wrong?
Thanks for your help,
zypher
_________________
linux user 65882
Back to top
View user's profile Send private message
ribo
n00b
n00b


Joined: 14 Aug 2002
Posts: 53

PostPosted: Thu Sep 11, 2003 9:12 am    Post subject: Reply with quote

using 2.6.0-test5-love1

anyone have probles with `df` output saying df:`/mnt/reiser': Value too large for defined data type

?
Back to top
View user's profile Send private message
Lovechild
Advocate
Advocate


Joined: 17 May 2002
Posts: 2858
Location: Århus, Denmark

PostPosted: Thu Sep 11, 2003 9:17 am    Post subject: Reply with quote

zypher wrote:
lovechild: I get an error-message when patching the kernel-tree.
It says something like "unexpected end of patchfile"
I downloaded your patch twice. It ends with:

Code:

+}
+
 /**
  * invalidate_mapping_pages - Invalidate all the unlocked pages of one inode
  * @mapping: the address_space which holds the pages to invalidate
 


Is something wrong?
Thanks for your help,
zypher


Hrmm, I'll look into that once I get a spare moment - I think it would have to do with the way I diff. I never noticed it personally because I run the tree I merge in - so I never applied the patch myself (foolish I know).

Anyone know a better way to diff than diff -Naur orig modified > file.patch ? - Come to think of it, this way would diff any .rej files and crap along with it.. I better go read some man pages.
Back to top
View user's profile Send private message
milothurston
Apprentice
Apprentice


Joined: 01 May 2002
Posts: 231
Location: Oxford, England.

PostPosted: Thu Sep 11, 2003 10:34 am    Post subject: Reply with quote

[quote="Lovechild"]if removing the CFQ scheduler it might be a good idea to test with vanilla test5 since Linus might have included something nasty.[/quote]

Not quite what you asked for, but I booted up test5-mm1 with cfq (by mistake) and found that cxoffice worked.
Milo.
Back to top
View user's profile Send private message
Lovechild
Advocate
Advocate


Joined: 17 May 2002
Posts: 2858
Location: Århus, Denmark

PostPosted: Thu Sep 11, 2003 8:50 pm    Post subject: Reply with quote

I'll just do a quick -test5-love2 release:

Updated to Nick v15 which fixes a possible oops and adds a bit of interactivity tweaking (Nick is talking about high res timers for v16 btw.)
Supermount-ng updated to the official 2.02 release which has proper support for Jens Axboes changes in test5 - changes I hacked around in -test5-love2, I assume this release has some bugfixes as well - I didn't see a changelog.
And I included the vesafb patch - the README file in Documentation/README.vesafb - and I will offer NO SUPPORT WHAT SO EVER on that feature - instead find the original thread, I'm merely offering -love as a testing ground of that patch.

The main reason for this release is fixing the possible oops, and I really wanted to upgrade Reiser4 but Namesys has yet to provide the promised snapshot - as it's filesystem code and thus can destroy a lot of data if it has a flaw I won't merge with their bitkeeper repo.
So Reiser4 will have to wait.

And of course I have a patch in here directly from the AMD labs to address the Instability on certain AMD chips when using prefetch. This should hopefully enabled prefetching on none broken AMD chips and disable it on those that are broken - instead of completely disabling it as we do now. I hope this will add stability on the affected setups and speed on those not affected... hope is the operative word here.

So because of the AMD errata and the oops fix here goes yet another release:

http://userportal.iha.dk/~01876/2.6.0-test5-love2.tar.bz2

(just when Lin-matt got the bmg ebuild up to date - hehe)

Oh and feedback on the performance of the scheduler would be really nice, any kind of feedback would help Nick make the scheduler even better.
Back to top
View user's profile Send private message
bssteph
l33t
l33t


Joined: 26 Feb 2003
Posts: 651
Location: Wisconsin

PostPosted: Thu Sep 11, 2003 9:24 pm    Post subject: Reply with quote

I was just in the middle of collecting a bit of "this feels like..." data but it was interactivity related, so i'll try love2 and have another go at it.
Back to top
View user's profile Send private message
bssteph
l33t
l33t


Joined: 26 Feb 2003
Posts: 651
Location: Wisconsin

PostPosted: Thu Sep 11, 2003 9:38 pm    Post subject: Reply with quote

Erm, is your diff backwards?
Code:
skuld linux-2.6.0-test5 # bzcat 2.6.0-test5-love2.tar.bz2 | patch -p1
The next patch would delete the file Documentation/README.vesafb,
which does not exist!  Assume -R? [n] n
Apply anyway? [n] n
Skipping patch.
1 out of 1 hunk ignored
patching file Documentation/filesystems/00-INDEX
Reversed (or previously applied) patch detected!  Assume -R? [n] n
Apply anyway? [n] n
Skipping patch.
1 out of 1 hunk ignored -- saving rejects to file Documentation/filesystems/00-INDEX.rej
The next patch would delete the file Documentation/filesystems/supermount.txt,
which does not exist!  Assume -R? [n] n
Apply anyway? [n] n
Skipping patch.
1 out of 1 hunk ignored
patching file Makefile
Reversed (or previously applied) patch detected!  Assume -R? [n]


Looked at the patch in the tarball and it is trying to remove all the lines in README.vesafb..

Code:
diff -Naur linux-2.6.0-test5-love2/Documentation/README.vesafb linux-2.6.0-test5/Documentation/README.vesafb
--- linux-2.6.0-test5-love2/Documentation/README.vesafb 2003-09-11 18:25:39.000000000 +0200
+++ linux-2.6.0-test5/Documentation/README.vesafb       1970-01-01 01:00:00.000000000 +0100
@@ -1,163 +0,0 @@
-
- Linux 2.4.x VESAFB Refresh Rate Change Patch v0.1.5
- ---------------------------------------------------


EDIT:
Should have mentioned (if anyone's being turned away by this) that if you add -R to the patch arguments it will reverse them, making it all work as it should.


Last edited by bssteph on Fri Sep 12, 2003 12:10 am; edited 1 time in total
Back to top
View user's profile Send private message
bssteph
l33t
l33t


Joined: 26 Feb 2003
Posts: 651
Location: Wisconsin

PostPosted: Thu Sep 11, 2003 11:41 pm    Post subject: Reply with quote

Here's my report on the scheduling (with -test5-love1).

Laptop/system specs:
1.6 GHz Pentium 4
256 MB RAM
64 (likely shared) MB video RAM on an ATI Radeon Mobility
2.6.0-test5-love1, cfq elevator
partitions reiserfs
gnome 2.4 (BMG version) -- 1 panel + gdesklets (sorry Lovechild), window dragging opaque
X reniced to -10 using the XFree86.wrapper from the forums

== Under a standard emerge system load ==

XMMS occasionally skips. load average 2.29 2.22 2.27 just had a skip while loading a large directory tree in Nautilus (one directory containing 27 directories with 10-200 items in each).

Giving windows the whip-around test eventually balances itself. gnome-terminal with transparency on, two tabs (one emerge, one running top) is choppy on whip at first, eventually (1-2 sec) it seems to get better slices or something and it's much smoother. This is probably expected behavior but I thought it right to mention.

Whipping windows over one another doesn't leave terrible artifacts for a split second in the underlying windows as I had in the Con days, but sometimes slowly moving one over aforementioned gnome-terminal with a fair load average of 3.04 2.36 2.27 will get oddly choppy and starve the redraw on gnome-terminal

== No emerge ==

XMMS still a touch prone to a skip. Starved it for a good 1/2 second by opening my largest directory (music, 30-some folders of artists and ~100 unsorted songs) in Nautilus. (load average 0.64 1.07 1.72 [just cooled off from the emerge]). Three skips (two slight, one ~1 sec) while loading /usr/portage/distfiles (634 files) in Nautilus. load average after: 0.77 0.80 1.45

Screwy thing happened. Dragging that gnome-terminal around (nothing running in it) was doing pretty good until it went too far down, about half of it off the screen, when the entire system just kind of stopped for a second. Nothing redrew and XMMS skipped. Further adventures to no-pixel's-land were fine. load average after: 0.58 0.71 1.36. Maybe this was just a gnome freak out?

XMMS just went skip happy. Skipped a couple times while doing nothing special (not much more than moving the mouse). Currently is 0.36 0.71 1.18. It would skip at regular intervals, did it about six times over two minutes before i went to next song and it did it once before acting normal again. PR was (and still is) 20. Blame seems to be when XMMS reads from disk. It's then that the skips happen. Again time to point out I'm using CFQ, maybe there's some tuning to be done here? I seem to trigger it more when scrolling a frame (mostly text) in Mozilla.

Extended dragging still sometimes lags a bit at first but smooths itself out.

NWN still performs as well as it ever did before on here.

Everything else is smooth.
Back to top
View user's profile Send private message
bssteph
l33t
l33t


Joined: 26 Feb 2003
Posts: 651
Location: Wisconsin

PostPosted: Fri Sep 12, 2003 3:22 am    Post subject: Reply with quote

Another report, this with love2 on a different machine (so take these at face value). I have to make it quick.

Desktop/system specs:
1.4 GHz Athlon Thunderbird
384 MB DDR SDRAM
64 MB video RAM on an Nvidia GeForce 3
2.6.0-test5-love2, cfq elevator
partitions reiserfs
gnome 2.4 (BMG version) -- 1 panel + gdesklets (sorry Lovechild), window dragging opaque
X reniced to -10 using the XFree86.wrapper from the forums

XMMS passes all Nautilus tests, even while under load average: 2.90, 2.16, 1.26

still seem to have the same behavior under load when whipping gnome-terminal around. Starts off choppy and becomes smooth.

No artifacts (except when whipping over Mozilla, but I think Mozilla is to blame for that).

XMMS is fine when under no load, I don't have the screwy gnome-terminal-outside-the-screen problem, and XMMS isn't waiting too long to do a disk read. gnome-terminal didn't seem to be choppy at first when I whipped them around under no load either.

--

General notes. It should be mentioned that the no-emerge testing on the laptop was done after emerging (and my RAM/swap was higher than normal, so maybe there was some going out to disk involved during my tests), whereas on the desktop it was before. And for the most part, I consider the desktop an arbitrarily faster system; at the very least the drives are faster so that could have something to do with XMMS' oddity too.

Hope this helps, Lovechild. Will test love2 on the laptop tomorrow for some hopefully more-useful results w/that.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Off the Wall All times are GMT
Goto page Previous  1, 2, 3, 4, 5, 6, 7  Next
Page 2 of 7

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum