Official thread: "zen-sources" - Part II

[xveilsidex]

i re-emerged zen-sources but this problem dont solved!

[loeb-it]

With zen-sources-2.6.24-r1 tuxonice does not work for me. With nvidia-drivers I get a black screen right after atomic copy and the system hangs. No output visible...

With kamikaze-sources I had the same problem before, but then I got it to work starting glxgears before suspending. This does not help any more. Does anyone else help the problem.

Perhaps I can manage to attach a serial console the next days, to see some output...

-------------------------------------

Addon: Tried different configurations. worked once, but crashed on resume. several tries with same config afterwards failed. Seems to be somehow random...

[loeb-it]

When trying to bring up the rtap interface it does not work:

galadriel mithrandir # ifconfig rtap0 up
SIOCSIFFLAGS: Das Argument ist ungültig

This occurs with ipwraw driver and also with ipw3945 driver, so it seems to be a more general issue.

It also worked with kamikaze-sources before.

Anyone else tried?

Addon:
------------------------------------------------------
In 2.6.24 kernels an interface can only be brought up with a mac address set. Created patches to fix this for ipw3945 and ipwraw drivers.

ipw3945-2.6.24-rtap_up.patch:

Code: Select all

*** linux_bak/drivers/net/wireless/ipw3945.c    2008-02-06 19:20:41.000000000 +0100
--- linux/drivers/net/wireless/ipw3945.c        2008-02-07 00:43:10.000000000 +0100
***************
*** 15131,15134 ****
--- 15131,15142 ----
        priv->config |= CFG_CUSTOM_MAC;
        memcpy(priv->mac_addr, addr->sa_data, ETH_ALEN);
+ 
+       // hack to get rtap device up, since in 2.6.24 kernels the mac of an iface must be set
+       // setting to mac of the wifi interface
+       
+       if (rtap_iface) memcpy(priv->prom_net_dev->dev_addr, addr->sa_data, ETH_ALEN);  
+ 
+       // hack end
+ 
        printk(KERN_INFO "%s: Setting MAC to %s\n",
               priv->net_dev->name, print_mac(mac, priv->mac_addr));
***************
*** 16284,16287 ****
--- 16292,16305 ----
        priv->prom_priv->ieee->iw_mode = IW_MODE_MONITOR;
  
+       // ToDo: find real mac here
+ 
+       // Hack to get rtap device up, since in 2.6.24 kernels the mac of an iface must be set
+       // Setting to a dummy mac since I don't know where to get the real one.
+       // Since no packets are being sent to this interface, this is not really important.
+       
+       strcpy(priv->prom_net_dev->dev_addr,"0018DE95031A");  
+ 
+       // hack end
+ 
        rc = register_netdev(priv->prom_net_dev);
        if (rc) {

ipwraw-2.6.24-rtap_up.patch:

Code: Select all

*** linux_bak/drivers/net/wireless/ipwraw.c     2008-02-06 23:41:16.000000000 +0100
--- linux/drivers/net/wireless/ipwraw.c 2008-02-07 00:43:06.000000000 +0100
***************
*** 7613,7616 ****
--- 7613,7624 ----
        priv->config |= CFG_CUSTOM_MAC;
        memcpy(priv->mac_addr, addr->sa_data, ETH_ALEN);
+ 
+       // hack to get rtap device up, since in 2.6.24 kernels the mac of an iface must be set
+       // setting to mac of the wifi interface
+       
+       if (param_rtap_iface) memcpy(priv->prom_net_dev->dev_addr, addr->sa_data, ETH_ALEN);  
+ 
+       // hack end
+ 
        printk(KERN_INFO "%s: Setting MAC to %s\n",
               priv->net_dev->name, print_mac(mac, priv->mac_addr));
***************
*** 8343,8346 ****
--- 8351,8364 ----
        priv->prom_net_dev->hard_start_xmit = ipw_prom_hard_start_xmit;
  
+       // ToDo: find real mac here
+ 
+       // Hack to get rtap device up, since in 2.6.24 kernels the mac of an iface must be set
+       // Setting to a dummy mac since I don't know where to get the real one.
+       // Since no packets are being sent to this interface, this is not really important.
+       
+       strcpy(priv->prom_net_dev->dev_addr,"0018DE95031A"); 
+ 
+       // hack end
+ 
        rc = register_netdev(priv->prom_net_dev);
        if (rc) {

[broch]

getting fatal error while loading e100 during boot up

dmesg output:
PCI: Firmware left 0000:08:08.0 e100 interrupts enabled, disabling
e100: Unknown symbol mii_ethtool_sset
e100: Unknown symbol mii_link_ok
e100: Unknown symbol mii_check_link
e100: Unknown symbol mii_nway_restart
e100: Unknown symbol generic_mii_ioctl
e100: Unknown symbol mii_ethtool_gset
e100: Intel(R) PRO/100 Network Driver, 3.5.23-k4-NAPI
e100: Copyright(c) 1999-2006 Intel Corporation

I am wondering if this is related to rcu?
switched to classec rcu to get proper poweroff and this error appeared.

Thanks for any input

[termite]

Alsa is better now. It loads, at least

However, alsamixer doesn't work (it dies with 'function snd_mixer_load failed: no such file or directory'), and when alsasound is stopped, it can't store the mixer values.

Testing TOI next...

edit: suspend to ram works fine, though weirdly, it throws an error about the powersave daemon, then just works...

[Waninkoko]

getting fatal error while loading e100 during boot up

dmesg output:
PCI: Firmware left 0000:08:08.0 e100 interrupts enabled, disabling
e100: Unknown symbol mii_ethtool_sset
e100: Unknown symbol mii_link_ok
e100: Unknown symbol mii_check_link
e100: Unknown symbol mii_nway_restart
e100: Unknown symbol generic_mii_ioctl
e100: Unknown symbol mii_ethtool_gset
e100: Intel(R) PRO/100 Network Driver, 3.5.23-k4-NAPI
e100: Copyright(c) 1999-2006 Intel Corporation

I am wondering if this is related to rcu?
switched to classec rcu to get proper poweroff and this error appeared.

Thanks for any input

I have the same ethernet device and works fine here o_O

[forceflow2]

I had this problem with e100 too, but it turned out to be because I wasn't paying attention and didn't mount /boot before copying over the new kernel and rebooting. Just a thought.

[termite]

Here's a Mandriva bug about the alsa problem: http://qa.mandriva.com/show_bug.cgi?id=37253

[termite]

Edit: sorry, I booted into the wrong kernel. No fix or workaround for the alsa problem as of yet...status is the same as the previous post.

[forceflow2]

Getting this when trying to build now.

Code: Select all

  CC [M]  sound/pci/intel8x0.o
sound/pci/intel8x0.c: In function 'fill_nocache':
sound/pci/intel8x0.c:714: error: implicit declaration of function 'set_pages_uc'
sound/pci/intel8x0.c:716: error: implicit declaration of function 'set_pages_wb'
make[2]: *** [sound/pci/intel8x0.o] Error 1
make[1]: *** [sound/pci] Error 2
make: *** [sound] Error 2

[dodo1122]

Getting this when trying to build now.

Code: Select all

  CC [M]  sound/pci/intel8x0.o
sound/pci/intel8x0.c: In function 'fill_nocache':
sound/pci/intel8x0.c:714: error: implicit declaration of function 'set_pages_uc'
sound/pci/intel8x0.c:716: error: implicit declaration of function 'set_pages_wb'
make[2]: *** [sound/pci/intel8x0.o] Error 1
make[1]: *** [sound/pci] Error 2
make: *** [sound] Error 2

Fixed.


dodo

[broch]

getting fatal error while loading e100 during boot up

dmesg output:
PCI: Firmware left 0000:08:08.0 e100 interrupts enabled, disabling
e100: Unknown symbol mii_ethtool_sset
e100: Unknown symbol mii_link_ok
e100: Unknown symbol mii_check_link
e100: Unknown symbol mii_nway_restart
e100: Unknown symbol generic_mii_ioctl
e100: Unknown symbol mii_ethtool_gset
e100: Intel(R) PRO/100 Network Driver, 3.5.23-k4-NAPI
e100: Copyright(c) 1999-2006 Intel Corporation

I am wondering if this is related to rcu?
switched to classec rcu to get proper poweroff and this error appeared.

Thanks for any input

I have the same ethernet device and works fine here o_O

the error is random, appears not at every boot and it does not seem (until now) to affect NIC functionality.

during a boot e100 complains about "unknown symbol in module or unknown parameter"

this is not a case of 2.6.23.14
I am going to try vanilla 2.6.24

Note added:
o.k. after few reboots of two kernels
classic rcu can poweroff but generates the above e100 module error
preemptive rcu will not power off but also will not generate e100 modules error (this would require more system restarts to confirm)

[nowitzness]

hi,

when i select "Zen Kernel Setups (Low Latency Desktop)" at the Zen options and "Preemption Model (Preemptible Kernel (Low-Latency Desktop))" at standard options, will i get any problems at runtime?

my idea was that both options may interfere negatively.

[Veldrin]

I think they don't...

Simply because "Zen Kernel Setups (Low Latency Desktop)" sets the fine grained behaviour of the scheduler, while "Preemption Model (Preemptible Kernel (Low-Latency Desktop))" only sets the general - preemtpive or not preemptive - behaviour.

[creidiki]

works fine - in fact, its exactly what you want.

[Phlogiston]

works perfect here, with all the extras

[dodo1122]

Latest btrfs (0.12) in master-devel and master-dev-rt.

Enjoy

dodo

[treffer]

Latest btrfs (0.12) in master-devel and master-dev-rt.

Enjoy

dodo

Is master-devel usable again (compiles fine with acpi and similar stuff)? Or should we wait for an official -25-rcX ?

[Waninkoko]

Latest btrfs (0.12) in master-devel and master-dev-rt.

Enjoy

dodo

Is master-devel usable again (compiles fine with acpi and similar stuff)? Or should we wait for an official -25-rcX ?

It's usable, all .25 shit was reverted.

[Spaulding]

Code: Select all

mm/page_alloc.c: In function ‘badram_markpages’:
mm/page_alloc.c:3073: error: implicit declaration of function ‘phys_to_page’
mm/page_alloc.c:3073: warning: initialization makes pointer from integer without a cast
make[1]: *** [mm/page_alloc.o] Błąd 1
make: *** [mm] Błąd 2

please fix it

[keenblade]

I can't shutdown. I force it holding pwr button for 4 seconds.
while shutting down,it stops with this;

Code: Select all

...
Pid:  10471, com: iwconfig Tainted:P 2.6.24-zen1 #2
...
---[ end of trace c60d184c28f803d2 ] ---
* Stopping lo
* Bringing down lo

Also while booting I got this:

Code: Select all

* Starting wlan0
SIOCSIFFLAGS: No such device
* Wireless radio has been killed for interface wlan0
* wpa_supplicant will launch, but not associate until
* wireles radio is re-enabled for interface wlan0
* Starting wlan0
SIOCSIFFLAGS: No such device
Could not set interface 'wlan0' up
ioctl[SIOCSIWAUTH]: Operation not supported
[ ok ]th param 4 value 0x0 -
* Starting wpa_cli on wlan0 ...

Any idea? It seems something with iwlwifi driver. I have wm3645agb wireless on this laptop that is runnin ~amd64.

[soundcheck]

Hi there.

A question about the Zen Custom-Tunables.

There seems to be a parameter inconsistency between the custom setting and the other presets such as gaming.

At least I don't see e.g. the vm parameters and sched_granu..... under custom as I can find them in the help functions on the other presets.

Is there any reason for that?

Cheers

[Phlogiston]

Anyone used virualbox modules with this kernel?
Here the modules fail to install. See: http://pastebin.ca/896382

[hirakendu]

use 1.5.4 ...

[creidiki]

That's not good, but it seems ebuild-related rather than zen-sources related.

I'd post a bug for it on b.g.o.