Old posts from the morph-sources thread

[yardbird]

Yay, i got Win4Lin working in morph5 =)!

That's great Some win4lin experimental patches have appeared and the softwaresuspend folks are probably going to integrate them in the next releases. I'll integrate their work into -morph as soon as it comes out.

Edit. the network driver works... edit

Good I won't integrate it into morph for the moment though, I think it's better to wait for an official update.

Why dosent the azx driver show up in the kernel?? I have to emerge the alsa-drivers damm :/ and the sound sucks

This happens because the alsa-patch I used is from november, and it seems like it does not include the azx driver (which is indeed in the alsa-driver-1.0.7 package). I'm investigating a way to integrate latest alsa drivers into morph, but don't know if this can be done easily. That's because in kernel 2.6.9 a lot of suspend/resume pci code changed, affecting heavily the sound drivers. That's also the source of the problems that some people are having with nitro and alsa.

I just tried to compile an smp kernel using the morph5 for the x86_64 plateform.
I got the following error during the compilation.

CC arch/x86_64/lib/usercopy.o
AR arch/x86_64/lib/lib.a
GEN .version
CHK include/linux/compile.h
UPD include/linux/compile.h
CC init/version.o
LD init/built-in.o
LD .tmp_vmlinux1
kernel/built-in.o(.text+0x21166): In function `smp_suspend':
: undefined reference to `__smp_suspend_lowlevel'
make: *** [.tmp_vmlinux1] Error 1

I started from scratch using make mkproper.
Any idea on what went wrong ?

I'm investigating this. I'll report back when I've found a solution. Thanks for the info

[yardbird]

I got the following error during the compilation.

CC arch/x86_64/lib/usercopy.o
AR arch/x86_64/lib/lib.a
GEN .version
CHK include/linux/compile.h
UPD include/linux/compile.h
CC init/version.o
LD init/built-in.o
LD .tmp_vmlinux1
kernel/built-in.o(.text+0x21166): In function `smp_suspend':
: undefined reference to `__smp_suspend_lowlevel'
make: *** [.tmp_vmlinux1] Error 1

Mmhhh... would you please save and take a look at the output of kernel compilation? I need to understand in which file the function `__smp_suspend_lowlevel' appears. You should see something like "Warning: implicit declaration of function `smp_suspend'". I suspect it is in kernel/power/process.c, can you confirm this? What compiler are you using?

[TheNull]

Hi,
is the lowmem patch applied in morph?

I just installed 2.6.10-rc1-morph5 and torsmo is still showing my maximum memory as 884M.

Or is there something special I have to do to activate the patch?

[yardbird]

Hi,
is the lowmem patch applied in morph?

I just installed 2.6.10-rc1-morph5 and torsmo is still showing my maximum memory as 884M.

Or is there something special I have to do to activate the patch?

It is in ck, so it is also in morph. You just have to enable it in Processor type and features --> 1GB Low memory support.

Cheers!

[TheNull]

Thanks very much =)

[seringen]

Hi,
is the lowmem patch applied in morph?

I just installed 2.6.10-rc1-morph5 and torsmo is still showing my maximum memory as 884M.

Or is there something special I have to do to activate the patch?

yeah, you have to make sure lowmem is turned on. I don't have the luxury of having a gig of ram, but it should work.

BTW it's CONFIG_1LOWMEM in your kernel config

[rommel]

well thanks for the kernel... it seems to be working very well for me... i just added mouse polling , that was all that wasnt in it that i wanted.

[bollucks]

Hi,
is the lowmem patch applied in morph?

I just installed 2.6.10-rc1-morph5 and torsmo is still showing my maximum memory as 884M.

Or is there something special I have to do to activate the patch?

yeah, you have to make sure lowmem is turned on.
BTW it's CONFIG_1LOWMEM in your kernel config

By "lowmem is turned on" he means "highmem is turned off"

[Robin79]

But nitro uses alsa 1.0.7? and i cant find the azx driver there either... Also that driver sucks for azx so i have compiled the latest development 8.rc1 and on that one i can use mixers couldnt that with 1.0.7 o when i get home i will hope i have sound so i can skip wintendo xp...

[silverter]

Great kernel patch so far. My applications start faster than with my last kernel (mm-sources). But I've got a couple of issues with this patchset:

When emerge ist running or something is compiling in the background, my system is almost unusable. I can't use the command line because the responsiveness ist so slow, and mplayer or rhythmbox stutters like hell. I'm using the same .config as with my last kernel (mm-sources) and with mm-sources I didn't have these issues. I might habe missed something in my config. Can someone provide me with some help on this.

tnx and regards,

[bollucks]

Great kernel patch so far. My applications start faster than with my last kernel (mm-sources). But I've got a couple of issues with this patchset:

When emerge ist running or something is compiling in the background, my system is almost unusable. I can't use the command line because the responsiveness ist so slow, and mplayer or rhythmbox stutters like hell. I'm using the same .config as with my last kernel (mm-sources) and with mm-sources I didn't have these issues. I might habe missed something in my config. Can someone provide me with some help on this.

tnx and regards,

-ck recommends generous use of +nice values for background tasks, or even batch scheduling.

[RogerWilco]

This is to confirm, that the morph5 patch works fine on a slackware-current box. Win4Lin is running nicely (worth noting, that the patch from netraverse did not apply cleanly, but no problem with morph5)

I seem to miss two patches that I applied manually - lufs (needed if you want captive-ntfs working) and cdfs. Any chance of including those in the next release?

http://lufs.sourceforge.net/lufs/

http://www.elis.UGent.be/~ronsse/cdfs/

[blaster999]

And what about the message saying "Warning: the latest version of cdfs will crash if you read audio CDs with kernels>2.6.3" on the cdfs website?

[JBerro]

Well, well now the kernel is compiling without trouble on my 32 bits gentoo.
So just give me a couple of minutes to try another compilation of the kernel on my 64 bit gentoo.
So I reboot and I will be back with the result.

See you.

[RogerWilco]

And what about the message saying "Warning: the latest version of cdfs will crash if you read audio CDs with kernels>2.6.3" on the cdfs website?

I don't know, have'nt tried on audio CDs. If true, than it's a no-go, of course.

[JBerro]

So to be short:
Compile ok with x86 platform.
Compile ok with x86_64 platform if monoprocessor
Compile ko if x86_64 platform and SMP.

I trapped the errror messages as follow.

[i]
kernel/intermodule.c:179: warning: `inter_module_register' is deprecated (declared at kernel/intermodule.c:38)
kernel/intermodule.c:180: warning: `inter_module_unregister' is deprecated (declared at kernel/intermodule.c:79)
kernel/intermodule.c:183: warning: `inter_module_put' is deprecated (declared at kernel/intermodule.c:160)
fs/reiser4/debug.c: In function `print_clog':
fs/reiser4/debug.c:701: warning: cast from pointer to integer of different size
fs/reiser4/debug.c:701: warning: cast from pointer to integer of different size
drivers/char/agp/backend.c: In function `agp_add_bridge':
drivers/char/agp/backend.c:281: warning: `inter_module_register' is deprecated (declared at include/linux/module.h:577)
drivers/char/agp/backend.c: In function `agp_remove_bridge':
drivers/char/agp/backend.c:301: warning: `inter_module_unregister' is deprecated (declared at include/linux/module.h:578)
In file included from drivers/net/wireless/orinoco.c:448:
drivers/scsi/aacraid/commctrl.c: In function `aac_send_raw_srb':
drivers/scsi/aacraid/commctrl.c:562: warning: cast to pointer from integer of different size
drivers/scsi/aacraid/commsup.c: In function `aac_handle_aif':
drivers/scsi/aacraid/commsup.c:783: warning: unused variable `hw_fib'
drivers/scsi/aic7xxx/aic79xx_osm.c: In function `ahd_linux_register_host':
drivers/scsi/aic7xxx/aic79xx_osm.c:2106: warning: ignoring return value of `scsi_add_host', declared with attribute warn_unused_result
drivers/scsi/aic7xxx/aic7xxx_osm.c: In function `ahc_linux_register_host':
drivers/scsi/aic7xxx/aic7xxx_osm.c:1744: warning: ignoring return value of `scsi_add_host', declared with attribute warn_unused_result

sound/core/memory.c: In function `copy_to_user_fromio':
sound/core/memory.c:270: warning: passing arg 2 of `memcpy_fromio' discards qualifiers from pointer target type
net/ipv4/netfilter/ipchains_core.c:1:2: warning: #warning ipchains is obsolete, and will be removed soon.
net/ipv4/netfilter/ipfwadm_core.c:1:2: warning: #warning ipfwadm is obsolete, and will be removed soon.
kernel/built-in.o(.text+0x22186): In function `smp_suspend':
: undefined reference to `__smp_suspend_lowlevel'
make: *** [.tmp_vmlinux1] Error 1
[/i]

Hope this will help, it's really tiny but this the only info i got.

So most likely something wrong with SMP code for x86_64.
Anyway the patchset is really fast and working fine on x86.
Hopefully this issue will bwe solved in the next release.
Again many thanks for the work done.

Cheers

[COiN3D]

What about adding chmp-patch for changing a usb mouse' polling rate ?

[silverter]

Great kernel patch so far. My applications start faster than with my last kernel (mm-sources). But I've got a couple of issues with this patchset:

When emerge ist running or something is compiling in the background, my system is almost unusable. I can't use the command line because the responsiveness ist so slow, and mplayer or rhythmbox stutters like hell. I'm using the same .config as with my last kernel (mm-sources) and with mm-sources I didn't have these issues. I might habe missed something in my config. Can someone provide me with some help on this.

tnx and regards,

-ck recommends generous use of +nice values for background tasks, or even batch scheduling.

Can u point me to where I can read about this batch scheduling stuff? As for +nice values for background task, is there a possibility to automate this?

tnx and regards,

[blaster999]

If I am not mistaken you can set PORTAGE_NICENESS in make.conf to something like +5 or even +10 to make portage and tools called by portage (gcc, make and so on) use less cpu when there are other tasks running.

[bollucks]

If I am not mistaken you can set PORTAGE_NICENESS in make.conf to something like +5 or even +10 to make portage and tools called by portage (gcc, make and so on) use less cpu when there are other tasks running.

Ideally for something as unimportant in the foreground as a compilation you just use the lowest prority possible (+19). To use batch scheduling you need to get schedtools from here:
http://freshmeat.net/projects/schedtool/?topic_id=136
and start a task you want to run batch like this:
schedtool -B -e your_command_here

[yardbird]

For those interested 2.6.10-morph6 is out. There are some news:

- added mouse polling patch,
- added many patches frome gentoo dev-sources (thanks to dsd for the great job in collecting these patches!),
- updates from Con Kolivas: new version of his scheduler and a small security fix.

NOTE: it seems like the way to change mouse polling rate is now to specify a parameter when loading the usbhid kernel module. This means that you will not find an option in the kernel config to set polling rate (like it was before). Follow the link for more info on that.

Regarding the patches from dev-sources I included only those of them that are bugfixes. I left out the massive sparc updates, for example.

well thanks for the kernel... it seems to be working very well for me... i just added mouse polling , that was all that wasnt in it that i wanted.

Added

But nitro uses alsa 1.0.7? and i cant find the azx driver there either... Also that driver sucks for azx so i have compiled the latest development 8.rc1 and on that one i can use mixers couldnt that with 1.0.7 o when i get home i will hope i have sound so i can skip wintendo xp...

This one is a bit toughy. I've dug a bit and it seems like the azx driver is not present in alsa-kernel-cvs, but it is there in the alsa-1.0.7 standalone package. Also the latest alsa 1.0.8rc1 does not include it yet. The driver itself seems more a "concept" driver than a real one, and on the alsa mailing list it seems that it gave quite a few problems (kernel crashes). My suggestion at this point would be to emerge alsa-drivers separately, to see if you can get it this way. Sorry about that

When emerge ist running or something is compiling in the background, my system is almost unusable. I can't use the command line because the responsiveness ist so slow, and mplayer or rhythmbox stutters like hell. I'm using the same .config as with my last kernel (mm-sources) and with mm-sources I didn't have these issues. I might habe missed something in my config. Can someone provide me with some help on this.

Try bollucks' recomendations. Also you could try with the "as" io-scheduler. The default is now "cfq", but in my experience sometimes it seems to cause some lag in responsiveness. To change scheduler append the line

elevator="as"

in the relevant section of your grub.conf. For example:

Code: Select all

title  Gentoo 2.6.9-morph11
root (hd0,1)
kernel (hd0,1)/boot/vmlinuz-2.6.9-morph11 root=/dev/hda6 elevator=as

Verify with "dmesg" that you are using the as elevator (it will say somewhere "elevator: using anticipatory as default io scheduler", or something like that).

I seem to miss two patches that I applied manually - lufs (needed if you want captive-ntfs working) and cdfs. Any chance of including those in the next release?

I modified the lufs patch so that it'll apply cleanly to morph6. However if compiled as module it gives an unresolved symbol message (which disappear if compiled statically in-kernel). I won't include it since the project seems abandoned, but you can download the patch here:

http://ing.unitn.it/~rbiscani/lufs-0.9. ... rph6.patch

Just apply it on top of morph6. Let me know if it works. Regarding cdfs, the message on their homepage is scary

So most likely something wrong with SMP code for x86_64.
Anyway the patchset is really fast and working fine on x86.
Hopefully this issue will bwe solved in the next release.

It's difficult to debug this for me since I don't have an x86-64 machine available But I have an idea, will post a patch tomorrow (I hope!)

Thanks for all the feedback! The patch is getting better and growing fast, so I've sent an application to be hosted on sourceforge. If it is accepted it will be easier to manage all these patches, thanks to the infrastructure of sourceforge. Let's hope for the best

Cheers!

[bollucks]

Try bollucks' recomendations. Also you could try with the "as" io-scheduler. The default is now "cfq", but in my experience sometimes it seems to cause some lag in responsiveness. To change scheduler append the line

elevator="as"

This has recently changed to
elevator=anticipatory

but you no longer need to reboot thanks to Jens' on-the-fly changing modifications. You can change it at any time like this:
echo cfq > /sys/block/hda/queue/scheduler
for example to change hda to cfq

[yardbird]

This has recently changed to
elevator=anticipatory

but you no longer need to reboot thanks to Jens' on-the-fly changing modifications. You can change it at any time like this:
echo cfq > /sys/block/hda/queue/scheduler
for example to change hda to cfq

Great! I didn't know that this feature was introduced into mainline. Thanks for the tip

[JBerro]

Hi yardbird,

The patchset is working fine with performances and features improvement on the x86 architecture.
This my main system I use for my daily home activity.
So as I said it is already very nice to have it running on my main system.
I just tried the morph6 on x86_64: same result.
Feel free to tell me to do some testing if you want, no problem.
Provided of course I undertstand what to do.
I am living in Singapore and I spend a couple of hours every night on my system.
So you can expect a daily feedback plus some additonal time zone delay ( Singapore is 7 hours ahead of Europe right Now.

See you then.

[feld]

OK new morph sources are pretty good except you need to do as they are doing in the new gentoo-dev-sources (r2) that should be released any moment. remove the following patch:

Removed: 1900_umem_catch.patch


because it causes tons of probs. For GL games and it also doesnt allow TeamSpeak to run for me, and for others it crashes after a min or two. Its just messed up....

Thought u should know.


-Feld