headers
Andrey Batyiev | 6 Oct 18:21

APIC misconfiguration

Hello everyone

I'm trying to create touchscreen driver for Wibrain UMPC. I found out that device doesn't emit interrupts
until I switch polarity field in IO_APIC_route_entry. To do this, I need to export ioapic_read_entry and
ioapic_write_entry symbols from arch/x86/kernel/io_apic_{32,64}.c
Is there any other (better) way to do this?

Thanks,
   Andrey
Permalink | Reply |
headers
Cyrill Gorcunov | 6 Oct 18:50
Gravatar

Re: APIC misconfiguration

[Andrey Batyiev - Mon, Oct 06, 2008 at 07:23:08PM +0300]
| Hello everyone
| 
| I'm trying to create touchscreen driver for Wibrain UMPC. I found
| out that device doesn't emit interrupts until I switch polarity
| field in IO_APIC_route_entry. To do this, I need to export
| ioapic_read_entry and ioapic_write_entry symbols from
| arch/x86/kernel/io_apic_{32,64}.c
| Is there any other (better) way to do this?
| 
| Thanks,
|    Andrey
| 

Hi Andrey,

actually it's not a bug of APIC itself but rather misconfigured
device I think. Not sure thru which bus it does work (pci
i suppose) but I think you better are to fix it on more higher
level like in config space of this device. PCI guys could tell
you more about this area I believe.

		- Cyrill -
Permalink | Reply |
headers
Andrey Batyiev | 6 Oct 18:55

Re: APIC misconfiguration

Hi Cyrill,

> actually it's not a bug of APIC itself but rather misconfigured
> device I think. Not sure thru which bus it does work (pci
> i suppose) but I think you better are to fix it on more higher
> level like in config space of this device. PCI guys could tell
> you more about this area I believe.

Device is connected to ISA bus and haven't any autoconfiguration methods.

Thanks,
   Andrey
Permalink | Reply |
headers
Cyrill Gorcunov | 6 Oct 19:17
Gravatar

Re: APIC misconfiguration

[Andrey Batyiev - Mon, Oct 06, 2008 at 07:55:28PM +0300]
| Hi Cyrill,
| 
| > actually it's not a bug of APIC itself but rather misconfigured
| > device I think. Not sure thru which bus it does work (pci
| > i suppose) but I think you better are to fix it on more higher
| > level like in config space of this device. PCI guys could tell
| > you more about this area I believe.
| 
| Device is connected to ISA bus and haven't any autoconfiguration methods.
| 
| Thanks,
|    Andrey
| 

hmm... don't we have any ISA quirks already? If this device support
PnP there should be descriptor for irq line where you could change it
since I don't believe it would be a first device in kernel which behave
not as supposed and developers has to override some configs bits.
Hope it help (i didn't check ISA code though -- too busy now).

		- Cyrill -
Permalink | Reply |
headers
Andrey Batyiev | 6 Oct 19:30

Re: APIC misconfiguration

> hmm... don't we have any ISA quirks already? If this device support
> PnP there should be descriptor for irq line where you could change it
> since I don't believe it would be a first device in kernel which behave
> not as supposed and developers has to override some configs bits.
> Hope it help (i didn't check ISA code though -- too busy now).

Well, it seems device haven't either PnP or isapnp doesn't detect it.

Thanks,
   Andrey
Permalink | Reply |
headers
Cyrill Gorcunov | 6 Oct 19:54
Gravatar

Re: APIC misconfiguration

[Andrey Batyiev - Mon, Oct 06, 2008 at 08:30:42PM +0300]
| > hmm... don't we have any ISA quirks already? If this device support
| > PnP there should be descriptor for irq line where you could change it
| > since I don't believe it would be a first device in kernel which behave
| > not as supposed and developers has to override some configs bits.
| > Hope it help (i didn't check ISA code though -- too busy now).
| 
| Well, it seems device haven't either PnP or isapnp doesn't detect it.
| 
| Thanks,
|    Andrey
| 

Lets better wait for advices for other developers. Writting directly
to io-apic routing table is not a good idea i think. Some quirks
should be involved.

		- Cyrill -
Permalink | Reply |
headers
Matthew Garrett | 7 Oct 13:47
Favicon

Re: APIC misconfiguration

On Mon, Oct 06, 2008 at 08:30:42PM +0300, Andrey Batyiev wrote:
> > hmm... don't we have any ISA quirks already? If this device support
> > PnP there should be descriptor for irq line where you could change it
> > since I don't believe it would be a first device in kernel which behave
> > not as supposed and developers has to override some configs bits.
> > Hope it help (i didn't check ISA code though -- too busy now).
> 
> Well, it seems device haven't either PnP or isapnp doesn't detect it.

It'd be more likely to be acpipnp nowadays. Does it appear anywhere in 
/sys/class/pnp?

--

-- 
Matthew Garrett | mjg59 <at> srcf.ucam.org
Permalink | Reply |

Gmane