2.6.3_rc3-love2 aka "Oops"

Message

steel300 · Post by **steel300** » Wed Feb 18, 2004 4:42 am

There's a new love for your enjoyment. I opted to go against 2.6.3_rc3 again since there's no mm out for 2.6.3_rc4. I have added a cool cflags selection patch from bssteph, the latest evms support, aureal-vortex sound cards, and a fancy boot logo. As usual it can be found here: http://www.linuxmall.us/~lovepatch/love-sources

The broken-out.tar.bz2 file is the current broken-out directory. Use at your own risk, not all patches apply or unapply cleanly. I also put the patch for nforce2 boards on the site, rather than include it in the release. I'm not sure if it will break the nforce2 boards that work. So if you need it, it's there.

steel300 · Post by **steel300** » Wed Feb 18, 2004 5:18 am

I just noticed the nforce2 patch doesn't actually compile. I'm still working on a solution and will keep you informed.

steel300 · Post by **steel300** » Wed Feb 18, 2004 6:06 am

I need to sleep real bad. That nforce patch is already included, enjoy!

monotux · Post by **monotux** » Wed Feb 18, 2004 6:24 am

have you guys ever heard abba?
I read some really strange lyrics from one of their songs....

Code: Select all

so I say thank you for the love-sources,
for bringing it to meeeeeeeeeee

isn't that strange?

steel300 · Post by **steel300** » Wed Feb 18, 2004 6:26 am

retribrute wrote:have you guys ever heard abba?
I read some really strange lyrics from one of their songs....
Code: Select all
so I say thank you for the love-sources,
for bringing it to meeeeeeeeeee
isn't that strange?

Creepy. Remind me never to walk alone at night.

bssteph · Post by **bssteph** » Wed Feb 18, 2004 6:53 am

steel300 wrote:
retribrute wrote:have you guys ever heard abba?
I read some really strange lyrics from one of their songs....
Code: Select all
so I say thank you for the love-sources,
for bringing it to meeeeeeeeeee
isn't that strange? :)
Creepy. Remind me never to walk alone at night.

steel300 sir, never walk alone at night.

iamarug · Post by **iamarug** » Wed Feb 18, 2004 7:05 am

could someone sketch out some possible scenarios of what would happen if he did walk alone at night?

btw, great job! seriously, I wish I had as much drive and energy as you do.

GentooBox · Post by **GentooBox** » Wed Feb 18, 2004 7:20 am

Thanks Steel300 for the new broken-out dir.

_Adik_ · Post by **_Adik_** » Wed Feb 18, 2004 7:53 am

@steel300

you are our god

first love-sources from 2.6.1 stable for me!
its a pizza for you!

MG-Cloud · Post by **MG-Cloud** » Wed Feb 18, 2004 8:00 am

/me hums "the freaks come out at night .... the freaks come out at night!" (please tell me someone else knows this song

)

Great job on another new release. My problem from love1 still stands however (tbh, I wasn't really expecting it to be solved with the given changelog

):

These sources look great - haven't updated since 2.6.1-love5 because I've been insanely busy. However, something's changed since then (I'm willing to bet its the new cfq, unless I messed up my kernel config) - while my kernel boots, it hangs at the same spot. It says:

Using cfq io scheduler
... snipped stuff about my cdrom drives ...
ide1 at 0x170-0x177, 0x376 on irq 15
hda: max request size: 128KiB

Following this, it hangs.

I've tried this with my crazy cflags, then removed them, thinking that was the cause of the issue. Unfortunately, it was not. The ide chipset I selected was the VIA one, which has always worked in the past.

Today (only had about 30 mins), I tried to isolate the error, trying a variety of different options. What I've found was that cfq is not the problem - the noop i/o scheduler gives me the same crash in the same spot. Tomorrow if I have time, I plan on trying mm-sources to see if the crash is love-specific, and if it crashes with mm too, then vanilla-sources to see if the crash is from an mm patch or from the actual kernel upgrade (currently using 2.6.1-lvoe5).

If anyone has time (I saw a few othe rpeople with the same problem in the previous thread), what results do you get performing the tests I listed? Hopefully once the patchset that causes the bug is isolated, the individual patch that causes the problem can then be isolated, and the error fixed or a bug sent upstream.

EDIT: 2.6.3 final and mm1 released

Didn't have time to fully look at changelog but saw this really quickly:
"[PATCH] fix too early probing of default IDE ports

It was broken by IDE modular fixes in 2.6.2."

No idea if that's the problem I was experiencing or not, but it might be.

Pink · Post by **Pink** » Wed Feb 18, 2004 8:09 am

@MG-Cloud: I didn't have time to try anything lastnight as I was playing with nvidia and udev...

I am going to try the new love-sources today and will let you know. Out of interst 2.6.3_rc2-love2 works perfectly, so something in the rc3 would seem to be the culprit.

Anyway, nice one steel

[edit]

Out of interest 2.6.3 (final) is out! link

MG-Cloud · Post by **MG-Cloud** » Wed Feb 18, 2004 8:13 am

Cool, thanks PickledOnion

I will be sure to try that if the newer ones still refuse to work.

BTW, just checked - the change I was hoping would fix the problems I just posted above was made in an earlier rc - its not present in the changelogs from rc3-rc4 nor from rc4-final.

Night all

JeffBlair · Post by **JeffBlair** » Wed Feb 18, 2004 8:26 am

Wow, what is takeing so long to get them out? You were putting 1 out a day.

Is supermount going to be included??

Pink · Post by **Pink** » Wed Feb 18, 2004 8:55 am

@MG-Cloud: Just tried these new sources and no love for me.

Same problem in the same place. I have to go to lectures now (damn university!) and am back with 2.6.3_rc2-love2.

Mayhap someone has ideas - it's seems not to be the scheduler and definitelty in the rc3 release. Maybe the final 2.6.3 will be ok.

Talking of which; Steel it's been out 4 hours, where's the love?

(I'd send you a pizza but it would be more than cold by the time it got you from here in the UK...)

Evil Dark Archon · Post by **Evil Dark Archon** » Wed Feb 18, 2004 9:13 am

do you think that this could be the shortest lived release ever for love-sources? Presuming that steel plans to release a patch based on 2.6.3-final within the next few hours, i'm going to skip doing an evil-sources release based on this patch and wait until 2.6.3-love1 comes out, no rushes btw.

superjaded · Post by **superjaded** » Wed Feb 18, 2004 9:36 am

Hmm.. this 2.6.3-rc3 series of kernels doesn't really seem to be liking me.
Both love1 and love2 lock up without any apparent reason. No oops, no "cannot load module," no nothing. Just a hard freeze.

The only thing I can really say is that it seems to be locking up during the first time it tries to detect (or whatever it does) my floppy drive. There are literally maybe 15 lines on my output kernel log before the thing locks up, so it's rather quick about it.

FWIW, here's my config.
Same config used for both love1 and love2. The only difference is the 8139too was changed from builtin to a module.
It's basically the same config I've used since I reinstalled gentoo a month or two ago.

Initially I thought it might have had something to do with my nic driver (8139too) because that seems to be the next step the kernel takes after the flopping drive thing in my working kernels. So after love1 wouldn't compile, I tried compiling 8139too as a module. Still does the same thing.

Here's my specs:
Duron 900mhz
512mb SDRAM PC133
SB Live! (using the alsa emu10k1 driver from the kernel)
Realtek 8139 card
20gb and 80gb hds, both maxtors

I also have the framebuffer enabled
not sure what else may be relevant

Also, on a less serious note, I've been wondering if people felt the Piggens scheduler was a little less responsive after 2.6.1-love2? While I was compiling 2.6.3-rc3-love2 not too long ago, I was also running gdesklets, epiphany and mplayer playing back a MPEG4 video.. all without any noticeable burps in anything while I was browsing the web. I've tried a few in the 2.6.2 series (just rc1-love1 and rc3-love1, admittedly) and it's MUCH worse in its quickness to adapt. There's a noticeable delay in scrolling and other widget response and if I do any real amount of scrolling, the video will skip.

FWIW, there was a point in time when the scheduler was recommended to be niced to a -10 value because that makes things a little bit easier to predict. Is this still needed/wanted? I guess I'm not search savvy enough for lkml to be of much use to me. -_- But with or without nice the result seems to be about the same.

I probably shouldn't get to mucking around in kernels too much at 4:30 in the morning, but I'll probably try out the "vanilla" mm-sources to see if the lockups persist. Just thought I'd chime in just in case someone might have an idea.

But regardless of possible regressions, I guess that's what makes love-sources fun.

Evil Dark Archon · Post by **Evil Dark Archon** » Wed Feb 18, 2004 9:42 am

a. what motherboard (and chipset) do you have, that probably has more to do with the problem than any of the info you specified.

b. the nice -10 thing is afaik obsolete, that was back when we used the old version of nick's scheduler

superjaded · Post by **superjaded** » Wed Feb 18, 2004 9:54 am

a) I'll have to do some research if you want specific mobo info, but it's an Iwill of some sort, I believe.

IDE controller driver I use is the ALI15x3 so I can get DMA on my hds.
I believe this is my mobo.

b) all right. I figured so.

charlieg · Post by **charlieg** » Wed Feb 18, 2004 10:26 am

OI! 2.6.3-MM1 IS OU!

steel300 · Post by **steel300** » Wed Feb 18, 2004 1:21 pm

Why can't mm release while I'm awake. Every time I go to bed, Andrew releases an new one. Well, off ot building the new love.

izomorfix · Post by **izomorfix** » Wed Feb 18, 2004 2:26 pm

Hi steel300,
I found a bug in my patch:
If you compile alsa only with a vortex card enabled, you get unknown symbols.
You could enable another sound card to fix this problem.
I made a patch, and I hope you will include it in the "2.6.3-love*" series

neenee · Post by **neenee** » Wed Feb 18, 2004 3:07 pm

i love the new logo

snekiepete · Post by **snekiepete** » Wed Feb 18, 2004 3:32 pm

I need some help here with the cfq scheduler. When I run the initrd from grub, my boot stops at this point.

Using cfq io scheduler

and then nothing......deadness......despair......etc.....

luqas · Post by **luqas** » Wed Feb 18, 2004 3:35 pm

BTW steel, what is this nforce2 patch?

Can't wait to apply this today...

zerojay · Post by **zerojay** » Wed Feb 18, 2004 3:37 pm

I also get frozen when it detects the floppy drive. The line before it is the cfq scheduler.. but I think it's the floppy drive. Ironically, I don't even have a floppy drive attached to my box nor do I have the floppy driver compiled into the kernel or as a module, so I don't know why it would detect a floppy drive at all. Problem persists even when the module is compiled (included in the kernel or not) and the floppy is attached. (This is with both the most recent love and evil sources.)