MSI error when reloading iwlagn module

MSI error when reloading iwlagn module

[Jeremy Fitzhardinge]

Hi,

With a fairly current kernel + xen, I'm seeing this if I rmmod iwlagn
and try to reload it:

[51230.646678] Intel(R) Wireless WiFi Link AGN driver for Linux, in-tree:
[51230.646685] Copyright(c) 2003-2011 Intel Corporation
[51230.646760] xen: registering gsi 17 triggering 0 polarity 1
[51230.646773] xen_map_pirq_gsi: returning irq 17 for gsi 17
[51230.646777] xen: --> pirq=17 -> irq=17 (gsi=17)
[51230.646781] Already setup the GSI :17
[51230.646789] iwlagn 0000:03:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
[51230.646814] iwlagn 0000:03:00.0: setting latency timer to 64
[51230.646935] iwlagn 0000:03:00.0: pci_resource_len = 0x00002000
[51230.646941] iwlagn 0000:03:00.0: pci_resource_base = ffffc9000671c000
[51230.646945] iwlagn 0000:03:00.0: HW Revision ID = 0x35
[51230.647075] iwlagn 0000:03:00.0: xen map irq failed -22 for 32752 domain
[51230.647081] iwlagn 0000:03:00.0: pci_enable_msi failed
[51230.647113] iwlagn 0000:03:00.0: PCI INT A disabled
[51230.647126] iwlagn: probe of 0000:03:00.0 failed with error -22

with this on the Xen console

(XEN) physdev.c:139: dom0: can't create irq for msi!


I'm running Xen as of a422e2a4451e, which your MSI changes in them,
which I suspect of having caused a regression (since I don't remember
having problems reloading msi-using drivers before).

Thanks,
    J

Re: MSI error when reloading iwlagn module

[Jan Beulich]

>>> On 27.09.11 at 21:57, Jeremy Fitzhardinge <jeremy@goop.org> wrote:
> Hi,
> 
> With a fairly current kernel + xen, I'm seeing this if I rmmod iwlagn
> and try to reload it:
> 
> [51230.646678] Intel(R) Wireless WiFi Link AGN driver for Linux, in-tree:
> [51230.646685] Copyright(c) 2003-2011 Intel Corporation
> [51230.646760] xen: registering gsi 17 triggering 0 polarity 1
> [51230.646773] xen_map_pirq_gsi: returning irq 17 for gsi 17
> [51230.646777] xen: --> pirq=17 -> irq=17 (gsi=17)
> [51230.646781] Already setup the GSI :17
> [51230.646789] iwlagn 0000:03:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
> [51230.646814] iwlagn 0000:03:00.0: setting latency timer to 64
> [51230.646935] iwlagn 0000:03:00.0: pci_resource_len = 0x00002000
> [51230.646941] iwlagn 0000:03:00.0: pci_resource_base = ffffc9000671c000
> [51230.646945] iwlagn 0000:03:00.0: HW Revision ID = 0x35
> [51230.647075] iwlagn 0000:03:00.0: xen map irq failed -22 for 32752 domain
> [51230.647081] iwlagn 0000:03:00.0: pci_enable_msi failed
> [51230.647113] iwlagn 0000:03:00.0: PCI INT A disabled
> [51230.647126] iwlagn: probe of 0000:03:00.0 failed with error -22
> 
> with this on the Xen console
> 
> (XEN) physdev.c:139: dom0: can't create irq for msi!
> 
> 
> I'm running Xen as of a422e2a4451e, which your MSI changes in them,
> which I suspect of having caused a regression (since I don't remember
> having problems reloading msi-using drivers before).

Are you certain (i.e. did you try reverting the top 1 to 3 commits from
there)? Alternatively, do you know what c/s last worked for you? Is
this one the first removal, or after several of them?

The message solely indicates a failure of create_irq(), with either
find_unassigned_irq() or __assign_irq_vector() failing being the cause,
none of which I touched recently. Instead I wonder whether
23812:32814ad7458d wouldn't be a more likely candidate (with
23815:9fa77d26a813 and 23816:7f357e1ef60a being less likely ones).

I'll nevertheless see whether this reproduces with one of the MSI
drivers that my machines have in use and allow easy removal/reload.

Jan

Re: Re: MSI error when reloading iwlagn module

[George Dunlap]

[-- Attachment #1: Type: text/plain, Size: 2497 bytes --]

Jeremy, can you try the attached patch (which reverts some of the
changes from c/s 23786:3a05da2dc7c0)?
 -George

On Wed, Sep 28, 2011 at 9:33 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>> On 27.09.11 at 21:57, Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>> Hi,
>>
>> With a fairly current kernel + xen, I'm seeing this if I rmmod iwlagn
>> and try to reload it:
>>
>> [51230.646678] Intel(R) Wireless WiFi Link AGN driver for Linux, in-tree:
>> [51230.646685] Copyright(c) 2003-2011 Intel Corporation
>> [51230.646760] xen: registering gsi 17 triggering 0 polarity 1
>> [51230.646773] xen_map_pirq_gsi: returning irq 17 for gsi 17
>> [51230.646777] xen: --> pirq=17 -> irq=17 (gsi=17)
>> [51230.646781] Already setup the GSI :17
>> [51230.646789] iwlagn 0000:03:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
>> [51230.646814] iwlagn 0000:03:00.0: setting latency timer to 64
>> [51230.646935] iwlagn 0000:03:00.0: pci_resource_len = 0x00002000
>> [51230.646941] iwlagn 0000:03:00.0: pci_resource_base = ffffc9000671c000
>> [51230.646945] iwlagn 0000:03:00.0: HW Revision ID = 0x35
>> [51230.647075] iwlagn 0000:03:00.0: xen map irq failed -22 for 32752 domain
>> [51230.647081] iwlagn 0000:03:00.0: pci_enable_msi failed
>> [51230.647113] iwlagn 0000:03:00.0: PCI INT A disabled
>> [51230.647126] iwlagn: probe of 0000:03:00.0 failed with error -22
>>
>> with this on the Xen console
>>
>> (XEN) physdev.c:139: dom0: can't create irq for msi!
>>
>>
>> I'm running Xen as of a422e2a4451e, which your MSI changes in them,
>> which I suspect of having caused a regression (since I don't remember
>> having problems reloading msi-using drivers before).
>
> Are you certain (i.e. did you try reverting the top 1 to 3 commits from
> there)? Alternatively, do you know what c/s last worked for you? Is
> this one the first removal, or after several of them?
>
> The message solely indicates a failure of create_irq(), with either
> find_unassigned_irq() or __assign_irq_vector() failing being the cause,
> none of which I touched recently. Instead I wonder whether
> 23812:32814ad7458d wouldn't be a more likely candidate (with
> 23815:9fa77d26a813 and 23816:7f357e1ef60a being less likely ones).
>
> I'll nevertheless see whether this reproduces with one of the MSI
> drivers that my machines have in use and allow easy removal/reload.
>
> Jan
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>

[-- Attachment #2: revert-23786-partial.diff --]
[-- Type: text/plain, Size: 887 bytes --]

diff -r cc339ab1d917 -r fd4fd4868325 xen/arch/x86/irq.c
--- a/xen/arch/x86/irq.c	Thu Sep 22 18:37:06 2011 +0100
+++ b/xen/arch/x86/irq.c	Wed Sep 28 11:16:37 2011 +0100
@@ -225,6 +225,9 @@ static void __clear_irq_vector(int irq)
     for_each_cpu_mask(cpu, tmp_mask)
         per_cpu(vector_irq, cpu)[vector] = -1;
 
+    if ( cfg->used_vectors )
+        clear_bit(vector, cfg->used_vectors);
+
     cfg->vector = IRQ_VECTOR_UNASSIGNED;
     cpus_clear(cfg->cpu_mask);
     cfg->used = IRQ_UNUSED;
@@ -239,12 +242,6 @@ static void __clear_irq_vector(int irq)
         per_cpu(vector_irq, cpu)[cfg->old_vector] = -1;
      }
 
-    if ( cfg->used_vectors )
-    {
-        ASSERT(test_bit(vector, cfg->used_vectors));
-        clear_bit(vector, cfg->used_vectors);
-    }
-
     cfg->move_in_progress = 0;
     cfg->old_vector = IRQ_VECTOR_UNASSIGNED;
     cpus_clear(cfg->old_cpu_mask);

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Re: MSI error when reloading iwlagn module

[Jan Beulich]

>>> On 28.09.11 at 10:33, "Jan Beulich" <JBeulich@suse.com> wrote:
>>>> On 27.09.11 at 21:57, Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>> Hi,
>> 
>> With a fairly current kernel + xen, I'm seeing this if I rmmod iwlagn
>> and try to reload it:
>> 
>> [51230.646678] Intel(R) Wireless WiFi Link AGN driver for Linux, in-tree:
>> [51230.646685] Copyright(c) 2003-2011 Intel Corporation
>> [51230.646760] xen: registering gsi 17 triggering 0 polarity 1
>> [51230.646773] xen_map_pirq_gsi: returning irq 17 for gsi 17
>> [51230.646777] xen: --> pirq=17 -> irq=17 (gsi=17)
>> [51230.646781] Already setup the GSI :17
>> [51230.646789] iwlagn 0000:03:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
>> [51230.646814] iwlagn 0000:03:00.0: setting latency timer to 64
>> [51230.646935] iwlagn 0000:03:00.0: pci_resource_len = 0x00002000
>> [51230.646941] iwlagn 0000:03:00.0: pci_resource_base = ffffc9000671c000
>> [51230.646945] iwlagn 0000:03:00.0: HW Revision ID = 0x35
>> [51230.647075] iwlagn 0000:03:00.0: xen map irq failed -22 for 32752 domain
>> [51230.647081] iwlagn 0000:03:00.0: pci_enable_msi failed
>> [51230.647113] iwlagn 0000:03:00.0: PCI INT A disabled
>> [51230.647126] iwlagn: probe of 0000:03:00.0 failed with error -22
>> 
>> with this on the Xen console
>> 
>> (XEN) physdev.c:139: dom0: can't create irq for msi!
>> 
>> 
>> I'm running Xen as of a422e2a4451e, which your MSI changes in them,
>> which I suspect of having caused a regression (since I don't remember
>> having problems reloading msi-using drivers before).
> 
> Are you certain (i.e. did you try reverting the top 1 to 3 commits from
> there)? Alternatively, do you know what c/s last worked for you? Is
> this one the first removal, or after several of them?
> 
> The message solely indicates a failure of create_irq(), with either
> find_unassigned_irq() or __assign_irq_vector() failing being the cause,
> none of which I touched recently. Instead I wonder whether
> 23812:32814ad7458d wouldn't be a more likely candidate (with
> 23815:9fa77d26a813 and 23816:7f357e1ef60a being less likely ones).
> 
> I'll nevertheless see whether this reproduces with one of the MSI
> drivers that my machines have in use and allow easy removal/reload.

A simple remove/load of mptsas doesn't reproduce this for me, so if
George's suggested change doesn't help it would be good if you had
more detail.

Jan

Re: Re: MSI error when reloading iwlagn module

[George Dunlap]

Actually, I think the patch I sent shouldn't have any effect on Intel
systems, unless you've expicitly enabled one of the irq vector map
options on your Xen command line.
 -George

On Wed, Sep 28, 2011 at 11:27 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>> On 28.09.11 at 10:33, "Jan Beulich" <JBeulich@suse.com> wrote:
>>>>> On 27.09.11 at 21:57, Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>>> Hi,
>>>
>>> With a fairly current kernel + xen, I'm seeing this if I rmmod iwlagn
>>> and try to reload it:
>>>
>>> [51230.646678] Intel(R) Wireless WiFi Link AGN driver for Linux, in-tree:
>>> [51230.646685] Copyright(c) 2003-2011 Intel Corporation
>>> [51230.646760] xen: registering gsi 17 triggering 0 polarity 1
>>> [51230.646773] xen_map_pirq_gsi: returning irq 17 for gsi 17
>>> [51230.646777] xen: --> pirq=17 -> irq=17 (gsi=17)
>>> [51230.646781] Already setup the GSI :17
>>> [51230.646789] iwlagn 0000:03:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
>>> [51230.646814] iwlagn 0000:03:00.0: setting latency timer to 64
>>> [51230.646935] iwlagn 0000:03:00.0: pci_resource_len = 0x00002000
>>> [51230.646941] iwlagn 0000:03:00.0: pci_resource_base = ffffc9000671c000
>>> [51230.646945] iwlagn 0000:03:00.0: HW Revision ID = 0x35
>>> [51230.647075] iwlagn 0000:03:00.0: xen map irq failed -22 for 32752 domain
>>> [51230.647081] iwlagn 0000:03:00.0: pci_enable_msi failed
>>> [51230.647113] iwlagn 0000:03:00.0: PCI INT A disabled
>>> [51230.647126] iwlagn: probe of 0000:03:00.0 failed with error -22
>>>
>>> with this on the Xen console
>>>
>>> (XEN) physdev.c:139: dom0: can't create irq for msi!
>>>
>>>
>>> I'm running Xen as of a422e2a4451e, which your MSI changes in them,
>>> which I suspect of having caused a regression (since I don't remember
>>> having problems reloading msi-using drivers before).
>>
>> Are you certain (i.e. did you try reverting the top 1 to 3 commits from
>> there)? Alternatively, do you know what c/s last worked for you? Is
>> this one the first removal, or after several of them?
>>
>> The message solely indicates a failure of create_irq(), with either
>> find_unassigned_irq() or __assign_irq_vector() failing being the cause,
>> none of which I touched recently. Instead I wonder whether
>> 23812:32814ad7458d wouldn't be a more likely candidate (with
>> 23815:9fa77d26a813 and 23816:7f357e1ef60a being less likely ones).
>>
>> I'll nevertheless see whether this reproduces with one of the MSI
>> drivers that my machines have in use and allow easy removal/reload.
>
> A simple remove/load of mptsas doesn't reproduce this for me, so if
> George's suggested change doesn't help it would be good if you had
> more detail.
>
> Jan
>
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>

Re: Re: MSI error when reloading iwlagn module

[Jeremy Fitzhardinge]

On 09/28/2011 04:07 AM, George Dunlap wrote:
> Actually, I think the patch I sent shouldn't have any effect on Intel
> systems, unless you've expicitly enabled one of the irq vector map
> options on your Xen command line.

All I have on my xen command line is "cpufreq=xen iommu=pv".

    J

Re: Re: MSI error when reloading iwlagn module

[Jan Beulich]

Jeremy,

are you still having this problem? If so, did you have time to look into it
at least to some degree? Or alternatively, did you find other ways to
reproduce this (so others - e.g. me - could try to look into it)?

Thanks, Jan

>>> On 28.09.11 at 12:17, George Dunlap <George.Dunlap@eu.citrix.com> wrote:
> Jeremy, can you try the attached patch (which reverts some of the
> changes from c/s 23786:3a05da2dc7c0)?
>  -George
> 
> On Wed, Sep 28, 2011 at 9:33 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>> On 27.09.11 at 21:57, Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>>> Hi,
>>>
>>> With a fairly current kernel + xen, I'm seeing this if I rmmod iwlagn
>>> and try to reload it:
>>>
>>> [51230.646678] Intel(R) Wireless WiFi Link AGN driver for Linux, in-tree:
>>> [51230.646685] Copyright(c) 2003-2011 Intel Corporation
>>> [51230.646760] xen: registering gsi 17 triggering 0 polarity 1
>>> [51230.646773] xen_map_pirq_gsi: returning irq 17 for gsi 17
>>> [51230.646777] xen: --> pirq=17 -> irq=17 (gsi=17)
>>> [51230.646781] Already setup the GSI :17
>>> [51230.646789] iwlagn 0000:03:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
>>> [51230.646814] iwlagn 0000:03:00.0: setting latency timer to 64
>>> [51230.646935] iwlagn 0000:03:00.0: pci_resource_len = 0x00002000
>>> [51230.646941] iwlagn 0000:03:00.0: pci_resource_base = ffffc9000671c000
>>> [51230.646945] iwlagn 0000:03:00.0: HW Revision ID = 0x35
>>> [51230.647075] iwlagn 0000:03:00.0: xen map irq failed -22 for 32752 domain
>>> [51230.647081] iwlagn 0000:03:00.0: pci_enable_msi failed
>>> [51230.647113] iwlagn 0000:03:00.0: PCI INT A disabled
>>> [51230.647126] iwlagn: probe of 0000:03:00.0 failed with error -22
>>>
>>> with this on the Xen console
>>>
>>> (XEN) physdev.c:139: dom0: can't create irq for msi!
>>>
>>>
>>> I'm running Xen as of a422e2a4451e, which your MSI changes in them,
>>> which I suspect of having caused a regression (since I don't remember
>>> having problems reloading msi-using drivers before).
>>
>> Are you certain (i.e. did you try reverting the top 1 to 3 commits from
>> there)? Alternatively, do you know what c/s last worked for you? Is
>> this one the first removal, or after several of them?
>>
>> The message solely indicates a failure of create_irq(), with either
>> find_unassigned_irq() or __assign_irq_vector() failing being the cause,
>> none of which I touched recently. Instead I wonder whether
>> 23812:32814ad7458d wouldn't be a more likely candidate (with
>> 23815:9fa77d26a813 and 23816:7f357e1ef60a being less likely ones).
>>
>> I'll nevertheless see whether this reproduces with one of the MSI
>> drivers that my machines have in use and allow easy removal/reload.
>>
>> Jan
>>
>>
>> _______________________________________________
>> Xen-devel mailing list
>> Xen-devel@lists.xensource.com 
>> http://lists.xensource.com/xen-devel 
>>

Re: Re: MSI error when reloading iwlagn module

[Jeremy Fitzhardinge]

On 11/07/2011 02:08 AM, Jan Beulich wrote:
> Jeremy,
>
> are you still having this problem? If so, did you have time to look into it
> at least to some degree? Or alternatively, did you find other ways to
> reproduce this (so others - e.g. me - could try to look into it)?

I haven't had a chance to re-try it, but I will once I've stabilized F16.

    J

>
> Thanks, Jan
>
>>>> On 28.09.11 at 12:17, George Dunlap <George.Dunlap@eu.citrix.com> wrote:
>> Jeremy, can you try the attached patch (which reverts some of the
>> changes from c/s 23786:3a05da2dc7c0)?
>>  -George
>>
>> On Wed, Sep 28, 2011 at 9:33 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>>> On 27.09.11 at 21:57, Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>>>> Hi,
>>>>
>>>> With a fairly current kernel + xen, I'm seeing this if I rmmod iwlagn
>>>> and try to reload it:
>>>>
>>>> [51230.646678] Intel(R) Wireless WiFi Link AGN driver for Linux, in-tree:
>>>> [51230.646685] Copyright(c) 2003-2011 Intel Corporation
>>>> [51230.646760] xen: registering gsi 17 triggering 0 polarity 1
>>>> [51230.646773] xen_map_pirq_gsi: returning irq 17 for gsi 17
>>>> [51230.646777] xen: --> pirq=17 -> irq=17 (gsi=17)
>>>> [51230.646781] Already setup the GSI :17
>>>> [51230.646789] iwlagn 0000:03:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
>>>> [51230.646814] iwlagn 0000:03:00.0: setting latency timer to 64
>>>> [51230.646935] iwlagn 0000:03:00.0: pci_resource_len = 0x00002000
>>>> [51230.646941] iwlagn 0000:03:00.0: pci_resource_base = ffffc9000671c000
>>>> [51230.646945] iwlagn 0000:03:00.0: HW Revision ID = 0x35
>>>> [51230.647075] iwlagn 0000:03:00.0: xen map irq failed -22 for 32752 domain
>>>> [51230.647081] iwlagn 0000:03:00.0: pci_enable_msi failed
>>>> [51230.647113] iwlagn 0000:03:00.0: PCI INT A disabled
>>>> [51230.647126] iwlagn: probe of 0000:03:00.0 failed with error -22
>>>>
>>>> with this on the Xen console
>>>>
>>>> (XEN) physdev.c:139: dom0: can't create irq for msi!
>>>>
>>>>
>>>> I'm running Xen as of a422e2a4451e, which your MSI changes in them,
>>>> which I suspect of having caused a regression (since I don't remember
>>>> having problems reloading msi-using drivers before).
>>> Are you certain (i.e. did you try reverting the top 1 to 3 commits from
>>> there)? Alternatively, do you know what c/s last worked for you? Is
>>> this one the first removal, or after several of them?
>>>
>>> The message solely indicates a failure of create_irq(), with either
>>> find_unassigned_irq() or __assign_irq_vector() failing being the cause,
>>> none of which I touched recently. Instead I wonder whether
>>> 23812:32814ad7458d wouldn't be a more likely candidate (with
>>> 23815:9fa77d26a813 and 23816:7f357e1ef60a being less likely ones).
>>>
>>> I'll nevertheless see whether this reproduces with one of the MSI
>>> drivers that my machines have in use and allow easy removal/reload.
>>>
>>> Jan
>>>
>>>
>>> _______________________________________________
>>> Xen-devel mailing list
>>> Xen-devel@lists.xensource.com 
>>> http://lists.xensource.com/xen-devel 
>>>
>
>