Serengeti dts

[Myles Watson]

I'm running with a very minimalized dts (attached) to try to get to the
point where a device gets matched correctly and disabled.  I think devices
that should be disabled are causing me grief in SimNOW.

This is a snippet of the log file that has me wondering what I'm doing
wrong:

PCI: pci_scan_bus pci_probe_dev returns dev 0x00000000(None (not found))
PCI: devfn 0x8
pci_scan_get_dev: list is 0x000ce8bc, *list is 0x00000000
PCI: pci_scan_bus pci_scan_get_dev returns dev None (no dev in tree yet)
new_device: devcnt 14
find_device_operations: check all_device_operations[i] 0x0000d7a0
find_device_operations: cons 0x0000d7a0, cons id PCI: 1022:0001
find_device_operations: check all_device_operations[i] 0x0000d800
find_device_operations: cons 0x0000d800, cons id APIC_CLUSTER: 1022:1100
find_device_operations: check all_device_operations[i] 0x0000d860
find_device_operations: cons 0x0000d860, cons id PCI: 1022:1100
find_device_operations: check all_device_operations[i] 0x0000dc80
find_device_operations: cons 0x0000dc80, cons id PCI: 1022:7460
find_device_operations: check all_device_operations[i] 0x0000dc20
find_device_operations: cons 0x0000dc20, cons id PCI: 1022:7462
find_device_operations: match
PCI: 01:01.0 id PCI: 1022:7462 ops
PCI: 01:01.0 [PCI: 1022:7462] enabled
PCI: pci_scan_bus pci_probe_dev returns dev 0x00012c70(dynamic PCI: 01:01.0)
Not a multi function device, or the device is not present. Skip to next
device.

Even though there's a match, a new device got created (dynamic PCI: 01:01.0)

Because of that, the enabled flag gets lost.

Thanks,
Myles

[Carl-Daniel Hailfinger]

On 29.10.2008 22:23, Myles Watson wrote:
> I'm running with a very minimalized dts (attached) to try to get to the
> point where a device gets matched correctly and disabled.  I think devices
> that should be disabled are causing me grief in SimNOW.
>
> This is a snippet of the log file that has me wondering what I'm doing
> wrong:
>
> PCI: pci_scan_bus pci_probe_dev returns dev 0x00000000(None (not found))
> PCI: devfn 0x8
> pci_scan_get_dev: list is 0x000ce8bc, *list is 0x00000000
> PCI: pci_scan_bus pci_scan_get_dev returns dev None (no dev in tree yet)
> new_device: devcnt 14
> find_device_operations: check all_device_operations[i] 0x0000d7a0
> find_device_operations: cons 0x0000d7a0, cons id PCI: 1022:0001
> find_device_operations: check all_device_operations[i] 0x0000d800
> find_device_operations: cons 0x0000d800, cons id APIC_CLUSTER: 1022:1100
> find_device_operations: check all_device_operations[i] 0x0000d860
> find_device_operations: cons 0x0000d860, cons id PCI: 1022:1100
> find_device_operations: check all_device_operations[i] 0x0000dc80
> find_device_operations: cons 0x0000dc80, cons id PCI: 1022:7460
> find_device_operations: check all_device_operations[i] 0x0000dc20
> find_device_operations: cons 0x0000dc20, cons id PCI: 1022:7462
> find_device_operations: match
> PCI: 01:01.0 id PCI: 1022:7462 ops
> PCI: 01:01.0 [PCI: 1022:7462] enabled
> PCI: pci_scan_bus pci_probe_dev returns dev 0x00012c70(dynamic PCI: 01:01.0)
> Not a multi function device, or the device is not present. Skip to next
> device.
>
> Even though there's a match, a new device got created (dynamic PCI: 01:01.0)
>
> Because of that, the enabled flag gets lost.

You can't fix that in the dts. It's a code bug. I had a fix for the old
device tree code, but forward porting will takes some time. I hope to
create a working version over the weekend.

Regards,
Carl-Daniel

[Myles Watson]

> -----Original Message-----
> From: Carl-Daniel Hailfinger [mailto:[e-mail]]
> Sent: Thursday, October 30, 2008 6:55 AM
> To: Myles Watson
> Cc: Coreboot
> Subject: Re: [coreboot] Serengeti dts
>
> On 29.10.2008 22:23, Myles Watson wrote:
> > I'm running with a very minimalized dts (attached) to try to get to the
> > point where a device gets matched correctly and disabled.  I think
>
> devices
> > that should be disabled are causing me grief in SimNOW.
> >
> > This is a snippet of the log file that has me wondering what I'm doing
> > wrong:
> >
> > PCI: pci_scan_bus pci_probe_dev returns dev 0x00000000(None (not found))
> > PCI: devfn 0x8
> > pci_scan_get_dev: list is 0x000ce8bc, *list is 0x00000000
> > PCI: pci_scan_bus pci_scan_get_dev returns dev None (no dev in tree yet)
> > new_device: devcnt 14
> > find_device_operations: check all_device_operations[i] 0x0000d7a0
> > find_device_operations: cons 0x0000d7a0, cons id PCI: 1022:0001
> > find_device_operations: check all_device_operations[i] 0x0000d800
> > find_device_operations: cons 0x0000d800, cons id APIC_CLUSTER: 1022:1100
> > find_device_operations: check all_device_operations[i] 0x0000d860
> > find_device_operations: cons 0x0000d860, cons id PCI: 1022:1100
> > find_device_operations: check all_device_operations[i] 0x0000dc80
> > find_device_operations: cons 0x0000dc80, cons id PCI: 1022:7460
> > find_device_operations: check all_device_operations[i] 0x0000dc20
> > find_device_operations: cons 0x0000dc20, cons id PCI: 1022:7462
> > find_device_operations: match
> > PCI: 01:01.0 id PCI: 1022:7462 ops
> > PCI: 01:01.0 [PCI: 1022:7462] enabled
> > PCI: pci_scan_bus pci_probe_dev returns dev 0x00012c70(dynamic PCI:
>
> 01:01.0)
> > Not a multi function device, or the device is not present. Skip to next
> > device.
> >
> > Even though there's a match, a new device got created (dynamic PCI:
>
> 01:01.0)
> > Because of that, the enabled flag gets lost.
>
> You can't fix that in the dts. It's a code bug. I had a fix for the old
> device tree code, but forward porting will takes some time. I hope to
> create a working version over the weekend.

I'd love to hear the general idea, even before I see the patch.  I'm
surprised that you think the problem is in the device tree code.

Thanks,
Myles

[Carl-Daniel Hailfinger]

On 30.10.2008 15:47, Myles Watson wrote:
> > -----Original Message-----
> > From: Carl-Daniel Hailfinger [mailto:[e-mail]]
> > Sent: Thursday, October 30, 2008 6:55 AM
> > To: Myles Watson
> > Cc: Coreboot
> > Subject: Re: [coreboot] Serengeti dts
> >
> > On 29.10.2008 22:23, Myles Watson wrote:
> > > I'm running with a very minimalized dts (attached) to try to get to the
> > > point where a device gets matched correctly and disabled.  I think
> >
> > devices
> > > that should be disabled are causing me grief in SimNOW.
> > >
> > > This is a snippet of the log file that has me wondering what I'm doing
> > > wrong:
> > >
> > > PCI: pci_scan_bus pci_probe_dev returns dev 0x00000000(None (not found))
> > > PCI: devfn 0x8
> > > pci_scan_get_dev: list is 0x000ce8bc, *list is 0x00000000
> > > PCI: pci_scan_bus pci_scan_get_dev returns dev None (no dev in tree yet)
> > > new_device: devcnt 14
> > > find_device_operations: check all_device_operations[i] 0x0000d7a0
> > > find_device_operations: cons 0x0000d7a0, cons id PCI: 1022:0001
> > > find_device_operations: check all_device_operations[i] 0x0000d800
> > > find_device_operations: cons 0x0000d800, cons id APIC_CLUSTER: 1022:1100
> > > find_device_operations: check all_device_operations[i] 0x0000d860
> > > find_device_operations: cons 0x0000d860, cons id PCI: 1022:1100
> > > find_device_operations: check all_device_operations[i] 0x0000dc80
> > > find_device_operations: cons 0x0000dc80, cons id PCI: 1022:7460
> > > find_device_operations: check all_device_operations[i] 0x0000dc20
> > > find_device_operations: cons 0x0000dc20, cons id PCI: 1022:7462
> > > find_device_operations: match
> > > PCI: 01:01.0 id PCI: 1022:7462 ops
> > > PCI: 01:01.0 [PCI: 1022:7462] enabled
> > > PCI: pci_scan_bus pci_probe_dev returns dev 0x00012c70(dynamic PCI:
> >
> > 01:01.0)
> > > Not a multi function device, or the device is not present. Skip to next
> > > device.
> > >
> > > Even though there's a match, a new device got created (dynamic PCI:
> >
> > 01:01.0)
> > > Because of that, the enabled flag gets lost.
> >
> > You can't fix that in the dts. It's a code bug. I had a fix for the old
> > device tree code, but forward porting will takes some time. I hope to
> > create a working version over the weekend.
>
> I'd love to hear the general idea, even before I see the patch.  I'm
> surprised that you think the problem is in the device tree code.

Scan your boot log for the string "error". Don't remember the
capitalization. Basically, we use the wrong device in the tree as basis
for the scan, so we get no matches at all.

I'll post my old patch.

Regards,
Carl-Daniel

[Myles Watson]

> -----Original Message-----
> From: Carl-Daniel Hailfinger [mailto:[e-mail]]
> Sent: Thursday, October 30, 2008 8:49 AM
> To: Myles Watson
> Cc: 'Coreboot'
> Subject: Re: [coreboot] Serengeti dts
>
> On 30.10.2008 15:47, Myles Watson wrote:
> > > -----Original Message-----
> > > From: Carl-Daniel Hailfinger [mailto:[e-mail]]
> > > Sent: Thursday, October 30, 2008 6:55 AM
> > > To: Myles Watson
> > > Cc: Coreboot
> > > Subject: Re: [coreboot] Serengeti dts
> > >
> > > On 29.10.2008 22:23, Myles Watson wrote:
> > > > I'm running with a very minimalized dts (attached) to try to get to
>
> the
> > > > point where a device gets matched correctly and disabled.  I think
> > >
> > > devices
> > > > that should be disabled are causing me grief in SimNOW.
> > > >
> > > > This is a snippet of the log file that has me wondering what I'm doing
> > > > wrong:
> > > >
> > > > PCI: pci_scan_bus pci_probe_dev returns dev 0x00000000(None (not
>
> found))
> > > > PCI: devfn 0x8
> > > > pci_scan_get_dev: list is 0x000ce8bc, *list is 0x00000000
> > > > PCI: pci_scan_bus pci_scan_get_dev returns dev None (no dev in tree
>
> yet)
> > > > new_device: devcnt 14
> > > > find_device_operations: check all_device_operations[i] 0x0000d7a0
> > > > find_device_operations: cons 0x0000d7a0, cons id PCI: 1022:0001
> > > > find_device_operations: check all_device_operations[i] 0x0000d800
> > > > find_device_operations: cons 0x0000d800, cons id APIC_CLUSTER:
>
> 1022:1100
> > > > find_device_operations: check all_device_operations[i] 0x0000d860
> > > > find_device_operations: cons 0x0000d860, cons id PCI: 1022:1100
> > > > find_device_operations: check all_device_operations[i] 0x0000dc80
> > > > find_device_operations: cons 0x0000dc80, cons id PCI: 1022:7460
> > > > find_device_operations: check all_device_operations[i] 0x0000dc20
> > > > find_device_operations: cons 0x0000dc20, cons id PCI: 1022:7462
> > > > find_device_operations: match
> > > > PCI: 01:01.0 id PCI: 1022:7462 ops
> > > > PCI: 01:01.0 [PCI: 1022:7462] enabled
> > > > PCI: pci_scan_bus pci_probe_dev returns dev 0x00012c70(dynamic PCI:
> > >
> > > 01:01.0)
> > > > Not a multi function device, or the device is not present. Skip to
>
> next
> > > > device.
> > > >
> > > > Even though there's a match, a new device got created (dynamic PCI:
> > >
> > > 01:01.0)
> > > > Because of that, the enabled flag gets lost.
> > >
> > > You can't fix that in the dts. It's a code bug. I had a fix for the old
> > > device tree code, but forward porting will takes some time. I hope to
> > > create a working version over the weekend.
> >
> > I'd love to hear the general idea, even before I see the patch.  I'm
> > surprised that you think the problem is in the device tree code.
>
> Scan your boot log for the string "error". Don't remember the
> capitalization. Basically, we use the wrong device in the tree as basis
> for the scan, so we get no matches at all.
>
> I'll post my old patch.

Great.

Thanks,
Myles