Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Pappy's Kernel Seeds Part III<Closed. Please use new thread>
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3, 4 ... 22, 23, 24  Next  
This topic is locked: you cannot edit posts or make replies.    Gentoo Forums Forum Index Unsupported Software
View previous topic :: View next topic  
Author Message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Thu Jul 15, 2010 6:36 pm    Post subject: Reply with quote

At this point, I'm going to keep my eyes open for when they finally get xorg-server-1.8.x stabilized. Since things are working fine with xorg-server-1.7.7, I'm not in any rush to move ahead.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
stuNNed2
n00b
n00b


Joined: 17 Jul 2009
Posts: 32
Location: New Orleans, LA, U.S.

PostPosted: Sat Jul 17, 2010 12:51 am    Post subject: question about upgrading kernels Reply with quote

thanks for the seeds pappy!

i'm currently running a stable tuxonice kernel 2.6.30-r11. it's the only stable kernel listed that doesn't affect me with a bug that messes up keyboard input to the point of no use. (please see https://bugs.gentoo.org/show_bug.cgi?id=320071 and my comments (Lance Lassetter) in reference to https://bugs.launchpad.net/gentoo/+bug/124406 )

what is the best way to upgrade a kernel with your seeds? just do a make oldconfig with an old seed or use the new seed and start fresh?
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Sat Jul 17, 2010 5:57 am    Post subject: Reply with quote

When I update, I usually start fresh with a new seed. At this point, I am so intimate with my machines, I don't even have to look at the old .config to make sure the devices are set properly.

That said, if you have a kernel that is working the way you wish, make oldconfig will work. Be advised that it is wise to make sure that things are in their proper place. Kernel setups change from major release to major release, and there are times that make oldconfig can get a bit mixed up. A quick once-over on the important stuff (HDD drivers, file system setups) and you should be good to go.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
iamben
Apprentice
Apprentice


Joined: 10 May 2004
Posts: 275

PostPosted: Wed Jul 21, 2010 6:42 pm    Post subject: Reply with quote

couple of questions about the "working with kernel seeds" page:

why is make xconfig still being recommended? it uses qt3 which is long gone for gentoo users

why are you still railing on kernel modesetting? not only is it currently recommended for intel video users, but user mode setting is no longer even an option starting in xf86-video-intel-2.10, which will surely be stable soon. radeon should not be far behind, and nouveau is getting there.

while i agree that it's generally a good idea to build wifi drivers as modules, in case something goes wrong, i know of none that MUST be built as modules to function. where are you getting this?

have you actually ever used the intel hardware RNG? i was under the impression that ICH mobos could not use it.

thanks for all your work, and your help here
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Wed Jul 21, 2010 8:50 pm    Post subject: Reply with quote

iamben wrote:
couple of questions about the "working with kernel seeds" page:

why is make xconfig still being recommended? it uses qt3 which is long gone for gentoo users


You are assuming that *I* no longer use qt or KDE-3, which is incorrect. Also, make xconfig has been "updated" in some sources to use qt4. It isn't going away anytime in the near future...no matter what the prattle on line may say.

Quote:
why are you still railing on kernel modesetting? not only is it currently recommended for intel video users, but user mode setting is no longer even an option starting in xf86-video-intel-2.10, which will surely be stable soon. radeon should not be far behind, and nouveau is getting there.


Stable soon? I've been reading that since KMS came out. Ha! So, I should wait for "soon" until I use this machine again? Really?

The ABORTION that is KMS has been in vogue for years, and it is STILL COMPLETELY UNSTABLE! That instability translates to an unusable computer. I'm sorry if you expect me to rally 'round the new crap, even when it sucks.

I don't do that. While the assumption may be that I don't really use my computer for anything more than troubleshooting the bombs that come down the portage pipeline, the truth is Gentoo is my primary OS, and I use my computer to do work. I don't just work on my computer.

KMS prevents me from working WITH my computer, and forces me to work ON it. Thanks, but no thanks.

Quote:
while i agree that it's generally a good idea to build wifi drivers as modules, in case something goes wrong, i know of none that MUST be built as modules to function. where are you getting this?


From reality! I've troubleshot a LOT of wireless connections, using all kinds of adapters. I have pages and pages of proof that wireless adapters work best (or work at all) only with modules. That's after two years plus of research.

Quote:
have you actually ever used the intel hardware RNG? i was under the impression that ICH mobos could not use it.


Well, I don't know what to say on that one as it's set up on this machine right now. I think it would be logical to assume that since it works on this machine (and the others I have with Intel-based mobos), your assumptions are incorrect.

Quote:
thanks for all your work, and your help here


No problem. You're most welcome.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
iamben
Apprentice
Apprentice


Joined: 10 May 2004
Posts: 275

PostPosted: Wed Jul 21, 2010 10:38 pm    Post subject: Reply with quote

Not that we have to keep going back & forth on these, but I may as well reply per-item--

QT3-- you may love it and still use it, but to assume that the new users reading your guide will have it... doesn't really make sense to me.

KMS-- When I speak of "stable soon", I'm talking gentoo stable. It's coming. If it's been a few versions (on xf86-video-intel, or kernel) since you last tried it, I recommend trying it again.

Wifi-- To say that "most" wireless drivers simply will not function built-in-- I'm not sure where you're getting that. I've only used a half dozen different drivers, but they all work just fine built-in.

RNG-- You don't know if it works, but somehow it's logical to assume that my assumptions are incorrect? I've tried it on various Core2/ICH platforms and it doesn't work. It's for i8xx chipsets only.

I'd really love for this guide to be the go-to resource for new users' kernel builds-- not just a historical document for how things used to be. I'm just trying to help here, and it seems like all I'm getting is "I know what I'm doing, you don't. Trust me on this." Or maybe I'm just reading things wrong.

-Ben
Back to top
View user's profile Send private message
dufeu
l33t
l33t


Joined: 30 Aug 2002
Posts: 924
Location: US-FL-EST

PostPosted: Wed Jul 21, 2010 11:18 pm    Post subject: Reply with quote

pappy_mcfae wrote:
iamben wrote:
couple of questions about the "working with kernel seeds" page:

why are you still railing on kernel modesetting? not only is it currently recommended for intel video users, but user mode setting is no longer even an option starting in xf86-video-intel-2.10, which will surely be stable soon. radeon should not be far behind, and nouveau is getting there.

Stable soon? I've been reading that since KMS came out. ... even when it sucks. ... KMS prevents me from working WITH my computer, and forces me to work ON it.

I have to agree with Pappy on this one. I get more systems with feelings of instability with KMS than without. In particular, on large memory systems {4 gig+} use of my swap partition seems borked. I work with fairly large image files {personal art images} on a regular basis. Typically, these can run as much as 2 gigs in memory. Proper working of swap is rather important to me.

The other place where I believe KMS may introduce some instability is in the functioning of 3rd party javascripts. This also seems related to swap functionality.

Yet another place where I question the wisdom of KMS doesn't have any direct issue with KMS itself. The real questions I have are the complexities introduced regarding motherboards with built in GPUs {i.e. IGP chipsets} where the graphics memory is taken as a chunk of main memory. Especially when the amount of memory set aside is 'auto'. KMS is a great concept for conserving memory and for using memory more efficiently. However, the kernel ends up putting even more reliance on the correctness of how the BIOS behaves. Load 4 gigs of RAM into a motherboard with this kind of setup and then watch memtest86+ run on it. The memory breaks and ranges may not be where you think they are!

In purest speculation with absolutely no evidence whatsoever, I would guess there may be a memory leak under some very complicated circumstances depending on motherboard, BIOS, amount of memory installed, associated graphics drivers, interaction with swap ...

It's not an easy subject and fraught with potential pitfalls.

For what it's worth, I perform package and kernel upgrades often. In the process, I occasionally try out {state of the } KMS set to 'ON' from time to time. If I end up with everything grinding to a halt, I reboot and remove KMS. I don't report anything upstream because I don't know how to gather meaningful information for them. At the present time, I consider KMS to be less than 100% reliable. When my actual use of KMS feels more stable during my KMS trial periods, then I'll use it as part of my personal standard. For now, I take the very conservative approach and don't use it.

Quote:
Quote:
while i agree that it's generally a good idea to build wifi drivers as modules, in case something goes wrong, i know of none that MUST be built as modules to function. where are you getting this?

From reality! I've troubleshot a LOT of wireless connections, using all kinds of adapters. I have pages and pages of proof that wireless adapters work best (or work at all) only with modules. That's after two years plus of research.

Many of the kernel help sections I've read of the various NIC and wireless NIC drivers recommend compiling as modules. For example, see:
Code:
  Say Y here if you have a Realtek 8169 PCI Gigabit Ethernet adapter. 
 
 To compile this driver as a module, choose M here: the module
 will be called r8169.  This is recommended.

In the Gigabyte NIC driver section, just like Realtek: IP1000, Yellowfin, SiS, SysKonnect, all the Broadcom and several others are recommended by the responsible kernel maintainers to be compiled as 'modules'. That's nearly half of the NIC drivers in that section alone which recommend compiling as modules.

In kernel help, I've never seen any text to the effect of recommending a given NIC not be compiled as a module. On the other hand, I know for a fact {from personal experience} that some of the motherboards I currently own will not initiate their Realtek NICs properly if I compile their driver directly into the kernel.

The implied rule-of-thumb would be that all NIC drivers of any type can be compiled as modules and loaded as needed. Indeed, all my various NIC drivers {wired and wireless} have initialized their respective NICs when the drivers were in 'module' form. That's not to say I didn't have other problems with driver breakage ... after all, compiled form doesn't matter if the driver itself is broken.

When you have a situation where there are multiple NICs of various types installed, then it becomes more important to take the conservative approach and be sure all the NIC drivers you need are compiled as modules. This is even more true if you have several identical NICs. I have run into situations where some NIC drivers will load one copy of themselves for each identical NIC installed. I suspect, but do not know, this may be a factor especially for those drivers which are 'module' recommended.

As an aside, my observation is that some drivers really do require multiple copies in the kernel if you have multiple identical NICs. In some cases, rather than a whole copy of the driver, a copy of the data structures for the driver will be created. In both these types of instances, having the driver be loaded as a module is a requirement. i.e. A compiled in NIC driver, in many cases, cannot simultaneously drive multiple identical NICs because the compiled in data structures are not expandable. I'm not a kernel programmer so I have no idea if this is actually the case. But I have observed behavior and done some reading which leads me believe this is what's going on. i.e. Compiled in a NIC driver, have it work for one card but not have it work for two installed identical cards, re-compiled it as module and seen two copies loaded into the kernel with both cards working. And I really have seen other cards where only one copy of the driver is loaded but /proc tells me there are two data structures, one for each physical card.

For me, the kernel seeds represent an attempt at reliability first, efficiency second. i.e. Make sure it works first then perform all the known solid safe tweaks second.

I make conscious decisions where I take risk in the kernel options I choose. For this reason, I have KMS turned off and all my wired and wireless NICs are compiled as modules. That's me being very conservative.

Yet I do take 'risks'. Right now, I'm running a single 'device', small btrfs based partition on my primary PC. It contains either static or low update activity data {mostly images, movies and download directories}. I'm also running a file server at my mother's house where all the data files are housed in a 5 device btrfs partition. This results in a single 3 terrabyte file system. {woot!}

The point is to be conservative and control {as much as reasonably possible} the number of places than can blow up unexpectedly. It certainly makes finding the cause of an issue much simpler if there are only a few 'experimental' places you're playing with at any one time.

Of course - all the standard dislaimers apply: YMMV, IANAP, I Know Nothing!
_________________
People whom think M$ is mediocre, don't know the half of it.


Last edited by dufeu on Thu Jul 22, 2010 6:32 am; edited 3 times in total
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Wed Jul 21, 2010 11:39 pm    Post subject: Reply with quote

iamben, you are making points that are, as far as I'm concerned, invalid.

This is my project. I have taken much input on making it better, and in some cases, I have implemented the ideas given. In other cases, I have decided that my way was working better. But still, it remains my baby, so to speak.

A seed is a basic thing. It is nothing more than a modified .config file. The information I've turned up for the site pages is from personal experimental research, and reading about what works and what doesn't. The way I have it set up now works for most people. That setup is going to change, as I'm still in process of writing the pages.

Since this is my project, it is going to carry some of my opinions forward. In my opinion, KMS is complete and utter garbage, and has been since it was first introduced to the kernel. I have tried it on EVERY kernel version since it came out...and guess what? NOTHING BUT AN UNSTABLE MESS! This remains the case up to 2.6.34.

I am NOT going to promote something I think is BROKEN!!! I don't know how else to say it. I am not a bandwagon jumper-on-er. When KMS works, then I'll change my mind. Until then, you can argue this point all you wish. I will not change my mind, nor the page, nor the seeds to suit your desires. Nor will I follow the latest trends because popular opinion deems it necessary.

This may be a non-profit business, but it is still my most successful business to date. I think that says a lot. Remember, I've read everything Ayn Rand has ever written. I'm just a stubborn egoist enough to follow my own vision, not the vision others want me to have.

As for the wireless module argument, if you would but take the time to look, you will note that at least one of the drivers tells you outright to set it up as a module...that it will not work otherwise. I don't recall which, as that page was "finished" some months ago. Then there are ndiswrapper and broadcom-sta, neither which can't exist as anything but a module.

As for the other things, you are free to do what you wish with the seed once you have it on your machine. The settings as they exist now have brought many to the Linux table that might have otherwise lived in fear of the thought of penguins on their screens. The fact that there exist three threads for this discussion tells me that something is really right in what I'm doing.

Once the page is done, the seed configuration is going to be changing...a lot. Then page rewrite time comes. So, keep an eye out to see what comes from the next round of updates. They may disappoint you, but you can bet they will work.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
iamben
Apprentice
Apprentice


Joined: 10 May 2004
Posts: 275

PostPosted: Thu Jul 22, 2010 12:12 am    Post subject: Reply with quote

Edit: I should not have said that. Carry on.c

Edit2: I mean this post itself. First 2 stand =)
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Sun Jul 25, 2010 9:26 am    Post subject: Reply with quote

Two more hardened-sources .configs are ready for your perusal. I've just uploaded the .configs for 2.6.32-hardened-r11 and 2.6.34-hardened-r1 in both x86 and x86_64 flavors. If all goes well, I should have the next information page up some time tomorrow afternoon or evening.

Cheers,
Pappy

PS, NeddySeagoon, your server appears to be down at the moment.
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Mon Jul 26, 2010 10:52 am    Post subject: Reply with quote

At long last, settings page ten has made it to the site. I've also updated all the other setting pages to contain their proper copyright information as well as the page genesis and update information. Enjoy!

My thanks to my new anonymous assistant who helped me with some of the more techie stuff.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
Killerchronic
Tux's lil' helper
Tux's lil' helper


Joined: 24 Apr 2007
Posts: 91
Location: UK

PostPosted: Mon Jul 26, 2010 11:36 am    Post subject: Reply with quote

Hey, Firstly i love your kernel seeds, always found they make a nice base to start from, so thank you for the time you put in :)

However i have a bit of a problem, i've noticed while using the 2.6.34-gentoo-r1-x86-07.config that it seems to be missing some parts that are in the default config. The problem is i need one of these parts, the one below imparticular. The thing is i tried just pasting this in the relevant section but it still doesnt't show up after doing make menuconfig. How would i go about getting this to show up? I'm guessing it needs more than the section i've found to show up. I'm no kernel guru sadly so this goes a bit beyond me.

Code:

# CONFIG_ATH_COMMON is not set
# CONFIG_ATH_DEBUG is not set
# CONFIG_ATH5K is not set
# CONFIG_ATH5K_DEBUG is not set
# CONFIG_ATH9K_HW is not set
# CONFIG_ATH9K_COMMON is not set
# CONFIG_ATH9K is not set


Thanks in advance

Killer
Back to top
View user's profile Send private message
dufeu
l33t
l33t


Joined: 30 Aug 2002
Posts: 924
Location: US-FL-EST

PostPosted: Mon Jul 26, 2010 3:20 pm    Post subject: Reply with quote

Killerchronic wrote:
Hey, Firstly i love your kernel seeds, always found they make a nice base to start from, so thank you for the time you put in :)

However i have a bit of a problem, i've noticed while using the 2.6.34-gentoo-r1-x86-07.config that it seems to be missing some parts that are in the default config. The problem is i need one of these parts, the one below imparticular. The thing is i tried just pasting this in the relevant section but it still doesnt't show up after doing make menuconfig. How would i go about getting this to show up? I'm guessing it needs more than the section i've found to show up. I'm no kernel guru sadly so this goes a bit beyond me.

Code:

# CONFIG_ATH_COMMON is not set
# CONFIG_ATH_DEBUG is not set
# CONFIG_ATH5K is not set
# CONFIG_ATH5K_DEBUG is not set
# CONFIG_ATH9K_HW is not set
# CONFIG_ATH9K_COMMON is not set
# CONFIG_ATH9K is not set


Thanks in advance

Killer

I don't know if those options have been deprecated {probably not} so I'm assuming that they are still available. The only way I know to deal with this situation is to create a default configuration and then use the seed as a guide. First, I recommend you read chapter 4 of the Linux Kernel in a Nutshell written by Greg K. The link takes you to the individual chapter pdfs. If you go up one level there, you can get a zipped file of all the chapter pdfs.

Basically, you want to go to the linux source root where ever you're working on it {i.e. /usr/src/version} and execute:
Code:
make defconfig
or you can execute:
Code:
make allmodconfig

The first will give you a default .config comparable to what Linus uses and the second with all the possible modules turned on.

Then, you can use the configuration mode of your choice {I'm a traditionalist and use 'make menuconfig'} in one window and then open a browser on Pappy's website where he discusses all the options and simply go through them.

Alternatively, if you have a second PC available, you can run the 'make defconfig' on one and install the seed on the other. Then you can open both PCs side by side to see the differences.

I use both these methods.

What happens is that sections of a .config file will 'dissappear' depending on options. So when you're using a 'seed' which is designed to be minimal, then sometimes a section that you need might not be there. Once you get this straightened out, you may want to post where the section resides. A lot of these little sections have flags that control if the sub-options appear. Pappy may want to turn on that flag but leave all the sub-options off. That way the section will remain in the seed but have no effect on the compiled kernel.

BTW - The LKN book pretty much says not to edit .config manually.

There's even a separate tool {make (silent)oldconfig} to copy a working config from the current working kernel to be used as a base for a newer kernel. This is something I just found out this week myself when I finally read all the chapters. Chapter 10 is the companion to chapter 4 with all the possible kernel make targets described.

Usually I'm in chapter 9 which is the penultimate guide to kernel parameters passed at boot time.

However, chapter 8 is ... dated. Especially the discussion on SATA drivers.

Good luck!
_________________
People whom think M$ is mediocre, don't know the half of it.


Last edited by dufeu on Mon Jul 26, 2010 3:48 pm; edited 2 times in total
Back to top
View user's profile Send private message
iamben
Apprentice
Apprentice


Joined: 10 May 2004
Posts: 275

PostPosted: Mon Jul 26, 2010 3:27 pm    Post subject: Reply with quote

Don't edit .config directly, use menuconfig. If something is hidden, it's because you don't have all of its dependencies satisfied. If you're looking for CONFIG_ATH9K, hit "/" in menuconfig and search for ATH9K. It will tell you where it's found, what its deps are, and whether you have each dep satisfied or not.
Back to top
View user's profile Send private message
Killerchronic
Tux's lil' helper
Tux's lil' helper


Joined: 24 Apr 2007
Posts: 91
Location: UK

PostPosted: Mon Jul 26, 2010 4:54 pm    Post subject: Reply with quote

Big thanks to both of you.

I had tried the "/" and search for ATH9K found it was supposed to be there but couldn't find the correct dependency. Then thought i would take a look in the .config and when it didn't show up there i thought it was an issue with that, had no idea that options didn't actually show up in the .config if it's dependencies were not met. Just figured they were all there from the start and didn't show in "make menuconfig" till you had set the correct options.

Learnt something new today :D

Thank you for the link dufeu, some good stuff in there. I have this kernel almost totally configured though except the wifi (LAN is working fine so this was just because im a perfectionist lol) so starting over is something i would rather not do. But it's certainly something i will be reading through in the coming days, i like to understand what im doing rather than just copying :)

And i don't edit the .config manually, i just search through it when im having trouble finding certain things and need to check if there on or off.


Anyway just incase someone has the same prob as me:

Code:
[*] Networking support  --->
      -*-   Wireless  --->
          <M>   cfg80211 - wireless configuration API
          <M>   Generic IEEE 802.11 Networking Stack (mac80211)


Is the dependencies that i needed to get the rest to show up:

Code:
[*] Network device support  --->
    [*]   Wireless LAN  ---> 
    <M>   Atheros Wireless Cards  --->
            [ ]   Atheros wireless debugging
            < >   Atheros 5xxx wireless cards support (NEW)
            < >   Atheros 802.11n wireless cards support (NEW)
            < >   Atheros AR9170 802.11n USB support (NEW)


Thanks again to both of you :D

Edit: typo
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Tue Aug 03, 2010 7:53 am    Post subject: Reply with quote

Today was a busy day in kernel sources land. Lots of new sources means lots of new .configs. To that end, I've added .configs for 2.6.27.49, 2.6.32.17, 2.6.33.7, 2.6.34.2, 2.6.34-gentoo-r3, 2.6.35, and 2.6.35-gentoo in x86 and x86_64 flavors. I also had to fix a few errors with some of the seeds...mostly being improperly named. Enjoy!

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
krinn
Watchman
Watchman


Joined: 02 May 2003
Posts: 7470

PostPosted: Tue Aug 03, 2010 9:12 am    Post subject: Reply with quote

pappy, what you think of kernel 2.6.34 ?

I'm just asking that, because i saw many users having weird issues using it, and a lot of regressions bugs, makes me think a bad branch was used to build that one. And seeing it reach 2.6.34.2 it might be more mature and issues might be gone ?


So : will you recommend it for a 2.6.33 user ?
Back to top
View user's profile Send private message
dufeu
l33t
l33t


Joined: 30 Aug 2002
Posts: 924
Location: US-FL-EST

PostPosted: Tue Aug 03, 2010 2:13 pm    Post subject: Reply with quote

krinn wrote:
pappy, what you think of kernel 2.6.34 ?

I'm just asking that, because i saw many users having weird issues using it, and a lot of regressions bugs, makes me think a bad branch was used to build that one. And seeing it reach 2.6.34.2 it might be more mature and issues might be gone ?


So : will you recommend it for a 2.6.33 user ?

FWIW - 2.6.34 was definitely unstable for me. Currently running 2.6.34.1 and happy with it.

Caveats: Of the three systems I'm currently running 2.6.34.1 on, one of them uses an ATI chipset. If you have similar, you may want to disable "AMD C1E" (if present) in your BIOS.
_________________
People whom think M$ is mediocre, don't know the half of it.
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Tue Aug 03, 2010 7:59 pm    Post subject: Reply with quote

I've had some weird problems with the .34 kernels, mostly related to how it interacts with nvidia-drivers. I've also noticed that at times, the .34 kernel seems to have a lag in it which usually shows up when typing in firefox.

On the plus side of things, the .34 kernels boot very quickly. I shaved at least a second and a half off kernel initialization time, and with baselayout-2/openrc, this system boots significantly quicker than it did with other kernels.

I personally keep on the bleeding edge of the kernels by running the most recent on my machines, if possible. The only time I stray from that path is if a specific kernel family doesn't allow a machine to operate properly. There have been times like that in the past. Since I don't have anything with the intel i810-i865 GPU, I can keep up.

I am going to give a try to the .35 kernel later on this afternoon. I'll report back on my initial impressions when I get it going.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
krinn
Watchman
Watchman


Joined: 02 May 2003
Posts: 7470

PostPosted: Tue Aug 03, 2010 8:49 pm    Post subject: Reply with quote

thank you, will wait .35 report, i think i will just pass on the .34 for this time
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Tue Aug 03, 2010 9:53 pm    Post subject: Reply with quote

.35 and ndiswrapper DO NOT get along. I'll probably be filing a bug report on that later this afternoon. broadcom-sta on the other hand works as well as it ever has. This kernel is definitely a slow-mo booter, but it seems to be otherwise ok...checking ALSA...Ah...that's a relief. No ALSA screw-ups this round.

It seems that some of the more basic settings in SATA got hidden under another layer of clicking. Also, they added a new fresh crop of gamepad drivers, which silently enabled themselves. Shame on them!

But no matter. I usually end up having to redo the first generation seed in a particular family. The .35 family is no different. I've uploaded corrected .configs for 2.6.35 and 2.6.35-gentoo in both x86 and x86_64 flavors.

I'm going to set up the 64 bit version and see how that works. That will be one sure way to see if nvidia-drivers is ready for the update...or not. Once again, I'll be reporting back once I've got things working on that machine.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
krinn
Watchman
Watchman


Joined: 02 May 2003
Posts: 7470

PostPosted: Tue Aug 03, 2010 11:56 pm    Post subject: Reply with quote

thank you dufeu (just saw you message, i'm getting old)
i don't have an ati card, but still i'll just stay away from .34 and look for .35 instead (lmao, look, never said: use)
but the bios hint will save many users life i'm sure, can tell what's wrong with AMD C1E option ? (i'm too curious i know, but this kind of stuff can follow many versions)
Back to top
View user's profile Send private message
dufeu
l33t
l33t


Joined: 30 Aug 2002
Posts: 924
Location: US-FL-EST

PostPosted: Wed Aug 04, 2010 1:43 am    Post subject: Reply with quote

krinn wrote:
... bios hint will save many users life i'm sure, can tell what's wrong with AMD C1E option ? (i'm too curious i know, but this kind of stuff can follow many versions)

I encountered a problem with one of my systems {primary desktop} over two months ago and couldn't figure it out. The thread {SOLVED!} was titled: Delays during boot and shutdown

I haven't made a suggestion yet to pappy regarding the issue. I believe, though I don't know for sure yet, that the section of the kernel dealing with CPU frequency scaling {throttling} may have issues with AMD {ATI} 780 family chipsets. Apparently, the timing function is "wreckage". See the LWN article linked in my post in that thread.

I haven't had a chance yet to play with various settings nor have I had a chance to grep the kernel to see where the AMD C1E function is dealt with. For this reason, I haven't yet made any recommendation to pappy regarding associated kernel settings.

I suspect, but don't know for sure because my memory is hazy, the issue may be a regression dating back to 2.6.31.x.
_________________
People whom think M$ is mediocre, don't know the half of it.
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Wed Aug 04, 2010 5:29 am    Post subject: Reply with quote

Interesting article. It should at least merit a warning. I don't know how popular those chips were, but I'm sure someone has one who is reading this.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
pappy_mcfae
Watchman
Watchman


Joined: 27 Dec 2007
Posts: 5999
Location: Pomona, California.

PostPosted: Wed Aug 04, 2010 5:34 am    Post subject: Reply with quote

On the 64 bit machine, things seem to be normal. I haven't noted any issues as of yet, but I don't usually use that machine for multimedia anymore as the latest ALSA version doesn't like envy24 boards.

There, I'm sure that the ALSA regression *published or not* for this driver remains as virulent and choppy as ever. One of these days, I might just add my email address to that bug.

Anyway, I'd say that 2.6.35 works fairly well, as kernels go. Check it out to see if your favorite regression bug lives on in this version.

Cheers,
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.
Back to top
View user's profile Send private message
Display posts from previous:   
This topic is locked: you cannot edit posts or make replies.    Gentoo Forums Forum Index Unsupported Software All times are GMT
Goto page Previous  1, 2, 3, 4 ... 22, 23, 24  Next
Page 3 of 24

 
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