On 09/12/11 17:31, Greg KH wrote:

...

First, it requires acpi=noirq and i915.semaphores=1 to make it pass the loader screen , otherwise it freezes completely...

In fact I did the install from an USB stick also.. -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On 09/12/11 17:47, Greg KH wrote:

On Fri, Dec 09, 2011 at 05:39:36PM -0300, Cristian Rodríguez wrote:

...

On 09/12/11 17:31, Greg KH wrote:

...

...

First, it requires acpi=noirq and i915.semaphores=1 to make it pass the loader screen , otherwise it freezes completely...

In fact I did the install from an USB stick also..

That's the BIOS doing the USB storage emulation, it's not that USB isn't working properly, what that option does is change the way ACPI handles the interrupts (i.e. it doesn't do it.) So, without any interrupts, your PCI USB controller can't be talked to by Linux.

So I would really try to not pass that option. What happens if you do not add that option?

The USB port doesnt do anything and Boot sequence freezes with message : [FIRMWARE BUG] ACPI(PEGP) defines _DOD but not _DOS There are no BIOS updates available, in fact the box comes with a newer BIOS than the one mentioned as the latest in the support pages... This is with kernel 3.2 RC3 from Kernel:HEAD (I chrooted into the installation to upgrade) -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On Friday 09 December 2011 23:40:30 Greg KH wrote:

On Fri, Dec 09, 2011 at 06:45:13PM -0300, Cristian Rodríguez wrote:

...

On 09/12/11 17:47, Greg KH wrote:

...

On Fri, Dec 09, 2011 at 05:39:36PM -0300, Cristian Rodríguez wrote:

...

On 09/12/11 17:31, Greg KH wrote:

...

...

First, it requires acpi=noirq and i915.semaphores=1 to make it pass the loader screen , otherwise it freezes completely...

In fact I did the install from an USB stick also..

That's the BIOS doing the USB storage emulation, it's not that USB isn't working properly, what that option does is change the way ACPI handles the interrupts (i.e. it doesn't do it.) So, without any interrupts, your PCI USB controller can't be talked to by Linux.

So I would really try to not pass that option. What happens if you do not add that option?

The USB port doesnt do anything and Boot sequence freezes with message :

[FIRMWARE BUG] ACPI(PEGP) defines _DOD but not _DOS That's unrelated.

Maybe unlikely, but if you see this line earlier in boot log: IOAPIC[1]: Invalid reference to IRQ 0 then this one (showing up in 12.1 sooner or later): https://lkml.org/lkml/2011/11/22/53 should help. Otherwise you may want to open a bug and attach acpidump (boot params do not matter) and assign it to me. Thomas -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On 09/12/11 17:47, Greg KH wrote:

So I would really try to not pass that option. What happens if you do not add that option?

An update.. after looking at the logs, I noticed the following messages, when acpi=noirq is used Dec 9 19:25:08 linux-lu80 kernel: [ 0.533331] pci 0000:00:1c.1: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:25:08 linux-lu80 kernel: [ 0.533592] pci 0000:00:1c.5: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:25:08 linux-lu80 kernel: [ 20.551883] i801_smbus 0000:00:1f.3: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:25:08 linux-lu80 kernel: [ 20.586975] atl1c 0000:06:00.0: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:25:08 linux-lu80 kernel: [ 20.677009] iwlwifi 0000:03:00.0: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:27:31 linux-lu80 kernel: [ 0.534407] pci 0000:00:1c.1: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:27:31 linux-lu80 kernel: [ 0.534639] pci 0000:00:1c.5: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:27:31 linux-lu80 kernel: [ 18.161298] i801_smbus 0000:00:1f.3: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:27:31 linux-lu80 kernel: [ 18.162244] iwlwifi 0000:03:00.0: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask Dec 9 19:27:31 linux-lu80 kernel: [ 18.163329] atl1c 0000:06:00.0: IRQ 9 doesn't match PIRQ mask 0x8c80; try pci=usepirqmask So I tried acpi=noirq pci=usepirqmask but the wifi card fails after a few minutes Dec 9 19:36:33 linux-lu80 kernel: [ 116.401607] Pid: 59, comm: kworker/u:4 Tainted: G C 3.2.0-rc3-4-desktop #1 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401608] Call Trace: Dec 9 19:36:33 linux-lu80 kernel: [ 116.401617] [<ffffffff810043da>] dump_trace+0xaa/0x2b0 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401622] [<ffffffff815a4eb9>] dump_stack+0x69/0x6f Dec 9 19:36:33 linux-lu80 kernel: [ 116.401626] [<ffffffff81055bfb>] warn_slowpath_common+0x7b/0xc0 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401633] [<ffffffffa04e6668>] iwl_trans_pcie_tx+0x888/0x8e0 [iwlwifi] Dec 9 19:36:33 linux-lu80 kernel: [ 116.401682] [<ffffffffa04c8108>] iwlagn_tx_skb+0x298/0x630 [iwlwifi] Dec 9 19:36:33 linux-lu80 kernel: [ 116.401698] [<ffffffffa04bf7ff>] iwlagn_mac_tx+0x10f/0x1c0 [iwlwifi] Dec 9 19:36:33 linux-lu80 kernel: [ 116.401721] [<ffffffffa046227c>] __ieee80211_tx+0x6c/0x230 [mac80211] Dec 9 19:36:33 linux-lu80 kernel: [ 116.401759] [<ffffffffa046414e>] ieee80211_tx+0xae/0xe0 [mac80211] Dec 9 19:36:33 linux-lu80 kernel: [ 116.401813] [<ffffffffa046513d>] ieee80211_tx_pending+0x7d/0x250 [mac80211] Dec 9 19:36:33 linux-lu80 kernel: [ 116.401852] [<ffffffff8105cb00>] tasklet_action+0x80/0x170 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401857] [<ffffffff8105d1aa>] __do_softirq+0xaa/0x280 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401861] [<ffffffff815c8b2c>] call_softirq+0x1c/0x30 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401865] [<ffffffff81004245>] do_softirq+0x65/0xa0 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401868] [<ffffffff8105cdfa>] local_bh_enable_ip+0xaa/0xc0 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401875] [<ffffffffa044e13e>] ieee80211_process_addba_resp+0x12e/0x130 [mac80211] Dec 9 19:36:33 linux-lu80 kernel: [ 116.401888] [<ffffffffa04571e9>] ieee80211_iface_work+0x319/0x380 [mac80211] Dec 9 19:36:33 linux-lu80 kernel: [ 116.401907] [<ffffffff81072c31>] process_one_work+0x111/0x4d0 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401909] [<ffffffff810733b2>] worker_thread+0x152/0x340 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401913] [<ffffffff8107785e>] kthread+0x7e/0x90 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401916] [<ffffffff815c8a34>] kernel_thread_helper+0x4/0x10 Dec 9 19:36:33 linux-lu80 kernel: [ 116.401918] ---[ end trace ab42b9c994f859f8 ]--- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On 09/12/11 17:31, Greg KH wrote:

On Fri, Dec 09, 2011 at 05:12:25PM -0300, Cristian Rodríguez wrote:

...

Hi:

I got this new laptop with openSUSE 12.1 and so far cannot successfully boot it :(

First, it requires acpi=noirq and i915.semaphores=1 to make it pass the loader screen , otherwise it freezes completely...

After poking around kernel boot options yesterday finally isolated two options that work: either booting with: - "acpi=noirq" alone - or.. just "noapic" using "nolapic" also works, but axes SMP capabilities. -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On Sunday 11 December 2011 21:17:10 Cristian Rodríguez wrote:

On 09/12/11 17:31, Greg KH wrote:

...

On Fri, Dec 09, 2011 at 05:12:25PM -0300, Cristian Rodríguez wrote:

...

Hi:

I got this new laptop with openSUSE 12.1 and so far cannot successfully boot it :(

First, it requires acpi=noirq and i915.semaphores=1 to make it pass the loader screen , otherwise it freezes completely...

The "noirq" option is suspicious, that shouldn't happen on any "new" hardware these days unless the bios is really broken.

Reduced even more, box works with only nox2apic

Suspect messages at boot

ec 10 16:21:08 linux-lu80 kernel: [ 3.235138] APIC error on CPU0: 00(80) : Illegal register address Dec 10 16:21:08 linux-lu80 kernel: [ 3.236549] APIC error on CPU0: 80(80) : Illegal register address That's likely mainline commit: 41750d31fc9599fd81763e685a6b7b42d298c4f8 which came in 3.2.-rc1

Theoretically you could double check whether BIOS requests to not enable x2apic even the OS found one by checking ACPI tables (this is what the patch does/checks for, not enabling x2apic if BIOS requests it). I now added the patch to 12.1 branch. Please wait until a new kernel of the day is built and exported (should be in some hours) and check that this is included in the changelog (rpm -qp --changelog kernel-desktop.rpm |less): commit c1c0cfd66da3cd5dd779d0fb24384fa497488ea7 Author: Thomas Renninger Date: Mon Dec 12 02:34:32 2011 +0100 x86, x2apic: Enable the bios request for x2apic optout. Thomas -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On Sunday, December 11, 2011 06:58:33 AM Cristian Rodríguez wrote:

On 11/12/11 22:46, Thomas Renninger wrote:

...

On Sunday 11 December 2011 21:17:10 Cristian Rodríguez wrote:

...

On 09/12/11 17:31, Greg KH wrote:

...

On Fri, Dec 09, 2011 at 05:12:25PM -0300, Cristian Rodríguez wrote:

...

Hi:

I got this new laptop with openSUSE 12.1 and so far cannot successfully boot it :(

First, it requires acpi=noirq and i915.semaphores=1 to make it pass the loader screen , otherwise it freezes completely...

The "noirq" option is suspicious, that shouldn't happen on any "new" hardware these days unless the bios is really broken.

Reduced even more, box works /with only nox2apic

In case this is of any use or provides more info, I have packaged/upgraded the Linux firmware test kit, in the OBS project "hardware".. and run it as well.

attached is the results on this failing laptop, I see in the results that there are errors in the SSDT and DSDT, even If I already fixed some of them, I cannot load the corrected tables because apparently adding the files to ACPI_DSDT in /etc/sysconfig/kernel and running mkinitrd no longer have any effect :-| This may work again soon. I've sent out these: https://lkml.org/lkml/2011/11/3/440 One patch touches ACPICA code, but those guys said they are going to pick it up, but are/were busy with spec 5.0 parts. And no complaints from Linux people... Anyway, chances are really good that these will show up in 3.3 and I pushed them into 12.1 already.

You would have to do it "manually", but that's easy: Compile your modified table: iasl -sa DSDT.dsl copy the result to the /boot dir: cp DSDT.aml /boot/initrd_instrumented concatenate the original initrd on top: cat /boot/initrd-3.0-xxx >>/boot/initrd_instrumented and best duplicate a boot entry in /boot/grub/menu.lst and let it point (use the) modified initrd. You find examples or more details in the Documentation parts of the patches. Thomas -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On Monday, December 12, 2011 02:50:49 AM Cristian Rodríguez wrote:

On 11/12/11 22:46, Thomas Renninger wrote:

...

commit c1c0cfd66da3cd5dd779d0fb24384fa497488ea7 Author: Thomas Renninger Date: Mon Dec 12 02:34:32 2011 +0100

x86, x2apic: Enable the bios request for x2apic optout.

Thomas

Ok, thank you :-) but after some extra testing, only when either noapic or acpi=noirq is used the machine *works* reliable, it boots correctly with nox2apic but after a few minutes the wifi adapter hangs with backtrace posted earlier.

Quick googling revealed: New Dell XPS 15z (L511z) Wireless Card Issues http://en.community.dell.com/support-forums/laptop/f/3518/p/19382502/2000021... and: https://wiki.ubuntu.com/HardwareSupport/Machines/Laptops/Dell/XPS/15z#Wirele... there they state wireless working, maybe it's the additional pcie_aspm param: pcie_aspm=force acpi=noirq i915.semaphores=1 To be honest: I would send it back if you still can. Get an HP, best pre-loaded with SLE11 SP1 (soon with SP2 and a recent 3.0 kernel, afaik latest models can get upgraded and still have support, but I don't know for sure). You would not have to pay the Windows license if you don't need it and you would have a well tested system. If something should really not work, you have full support and HP generally cares about making their BIOS work with Linux. I have one and every single piece of HW works like a charme out of the box. Here another example where lacking Linux vendor support, again Dell, "is not nice" (I better avoid too harsh words on a public list): It's about not being able to reboot and the blacklist is full of Dell machines: http://www.gossamer-threads.com/lists/linux/kernel/1450309 Let me paste the important parts of the thread:

...

...

...

It's a bug in the SMM code on these machines, and it goes away if you disable VT-D. Let's try to actually fix it How do you intend to fix that? From description above it sounds as if this problem is not fixable from OS side.

Tear down VT-D state before reboot. But doing this for all X86 machines because of a handful of Dells with broken SMM code sounds wrong as well.

Doesn't Dell ship with Ubuntu pre-loaded? Can't someone point them to this to get this fixed in BIOS?

I've been trying. ... Thomas -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On 11/12/11 22:46, Thomas Renninger wrote: -qp --changelog kernel-desktop.rpm |less): BTW.. is there any way to "slow down" or a hidden "pager" for kernel logs so I can see and take note on the very early ACPI messages that happend before hanging ? I will attempt to use netconsole ..not sure if that will ever work... This is not the first time I get this problems and Im getting pretty annoyed by the fact I have no way to retrieve debug information, can anyone recommend a device that can be bought to get this messages on a system that has no serial port ? -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On 09/12/11 17:31, Greg KH wrote:

...

On Fri, Dec 09, 2011 at 05:12:25PM -0300, Cristian Rodríguez wrote:

...

Hi:

I got this new laptop with openSUSE 12.1 and so far cannot successfully boot it :(

First, it requires acpi=noirq and i915.semaphores=1 to make it pass the loader screen , otherwise it freezes completely...

The "noirq" option is suspicious, that shouldn't happen on any "new" hardware these days unless the bios is really broken.

That's what is causing your USB devices to not show up, that needs to be fixed to get USB to work properly.

Does any older release boot on this box properly?

thanks,

greg k-h

Ok, After a fight with mkinitrd (netconsole support is broken) I got this attached boot sequence Is the machine really freezing there or is there also a backtrace on

On Monday, December 12, 2011 11:51:07 PM Cristian Rodríguez wrote: the screen (best with vga=normal instead of vga=xxx)? If there is a backtrace, you could take a picture. Does sysrq still work (sysrq_always_enabled boot param and then hit SysRq - h). I can't see anything bad on a quick look. Maybe comparing this one with a acpi=noirq reveals something. Are there virtualization settings in BIOS that can be switched off? Why are the hpet boot params? nohz=off boot param has probably also tried already by someone. Sigh, looks like this machine has a dozen different and hard to find/solve issues... Unfortunately bugzilla.kernel.org does not exist anymore, best is you open a bug on bugzilla.novell.com and attach all you have. Thomas -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On 09/12/11 17:31, Greg KH wrote:

That's what is causing your USB devices to not show up, that needs to be fixed to get USB to work properly.

Nope Greg, I just figured it out why the lack of USB during the boot sequence, apparently there is a bug (?) in mkinitrd... module "xhci-hcd" is not picked up by the "usb" feature of mkinitrd.. adding xhci-hcd to INITRD_MODULES solves the problem. -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On 23/12/11 01:39, Greg KH wrote: I wonder why the

installer didn't get this right for you in the beginning...

I do not wonder, here is the fix :)

On 04/01/12 05:52, Jean Delvare wrote:

On Friday 23 December 2011 05:54:09 am Cristian Rodríguez wrote:

...

On 23/12/11 01:39, Greg KH wrote: I wonder why the

...

installer didn't get this right for you in the beginning...

I do not wonder, here is the fix :)

Good catch. Was this fix applied by now?

No, at least it is not yet in GIT. -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org