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.

For installation I had to pass "nolapic" in order to boot the media.. and during the installation the "yubikey" which stores the password, worked just fine (otherwise I would not have not been able to enter that bigass password :-D)

Does any older release boot on this box properly?

The laptop has just arrived in my door today, I just started the task of getting it up, will check if it boot with an old OS) -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

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? thanks, greg k-h -- 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 17:47, 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?

booting *without* acpi=noirq: If I pass: "nolapic_timer" --> Kernel panic watchdog detected hard CPU lockup ..." If I use "nolapic" at least the keyboard gets backs to life and shows an OOPS in usb_hcd_irq... -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On 09/12/11 19:40, Greg KH wrote:

Can you contact the manufacturer? This does look like a firmware bug, you can probably confirm it with the Linux Firmware kit, right?

How I explain to them this when windows 7 worked just fine ? I added a new lukskey to the encrypted partition to boot without USB device provided pasword..either passing pci=noacpi or acpi=noirq makes the system run,(but no USB available at boot time) though I have to remove the i915.semaphores=1 line otherwise a few minutes after successful boot the video goes belly up, not even REISUB works. -- 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 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 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 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 Thanks for your help.

On 13/12/11 07:31, Thomas Renninger wrote:

Is the machine really freezing there or is there also a backtrace on the screen (best with vga=normal instead of vga=xxx)?

It really freezes there, however, it is not "consistent" sometimes it freezes right during device-mapper setup or simply after the FIRMARE BUG: (PEGP) mentioned early on this thread.

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 also tried that.. no luck.. maybe I didn't find the right combination of keys in the laptop keyboard.. will try again later

Maybe comparing this one with a acpi=noirq reveals something.

I will try to produce a diff..

Are there virtualization settings in BIOS that can be switched off?

Yes, they are, option "Virtualization" currently in state "Enabled" I tried disabled, same result.

Why are the hpet boot params?

Well, unless I am in a misunderstading using HPET clock source will allow the CPU to go into deeper C states while idle, thus battery life will improve..

nohz=off boot param has probably also tried already by someone.

I have not tried nohz.. will see :-) -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

On Fri, Dec 23, 2011 at 01:22:39AM -0300, Cristian Rodríguez wrote:

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.

Wonderful, thanks for working to resolve this. I wonder why the installer didn't get this right for you in the beginning... greg k-h -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org

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? -- Jean Delvare Suse L3 -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org