Skip to contentSkip to conversation

3.19: device name associates with IRQ's for ahci controllers operating with a single IRQ changed from "ahci?" to "<BDF>"

DISCUSSION Sander Eikelenboom 4 participants 11 messages Feb 9, 2015 – Mar 2, 2015

lore
Share & tools
lore
#0 lore raw

Hi.

In 3.19 the device name associates with IRQ's for ahci controllers operating with a single IRQ changed from "ahci?" to "<BDF>", was this intentional ?

It's probably commit 18dcf433f3ded61eb140a55e7048ec2fef79e723 (or another one in that series).

--
Sander

/proc/interrupts of an ahci controller with a single irq:
52: 13529 0 0 0 xen-pirq-msi 0000:00:1f.2

/proc/interrupts of an ahci controller with multiple irq's:

 114:     412535          0          0          0          0          0  xen-pirq-msi       ahci0
 115:          0          0          0          0          0          0  xen-pirq-msi       ahci1
 116:      16717          0          0          0          0          0  xen-pirq-msi       ahci2
 117:          0          0          0          0          0          0  xen-pirq-msi       ahci3
 118:          0          0          0          0          0          0  xen-pirq-msi       ahci4
 119:          0          0          0          0          0          0  xen-pirq-msi       ahci5
#1 lore raw
Quoted 7 lines · Sander Eikelenboom

Hi.

In 3.19 the device name associates with IRQ's for ahci controllers operating with a single IRQ changed from "ahci?" to "<BDF>", was this intentional ?

It's probably commit 18dcf433f3ded61eb140a55e7048ec2fef79e723 (or another one in that series).

Oh, looking at that commit that might make sense. In ahci_host_activate_single_irq it now uses dev_driver_name instead of dev_name as it did before (from ata_host_activate). The description sounds like before the driver is registered this will return the bus. And registering an interrupt would likely be before the driver is fully registered... But you probably saw that, too. Whether that really was intentional is still the question. :)

-Stefan

Quoted 17 lines

--
Sander

/proc/interrupts of an ahci controller with a single irq:
52: 13529 0 0 0 xen-pirq-msi 0000:00:1f.2

/proc/interrupts of an ahci controller with multiple irq's:

 114:     412535          0          0          0          0          0  xen-pirq-msi       ahci0
 115:          0          0          0          0          0          0  xen-pirq-msi       ahci1
 116:      16717          0          0          0          0          0  xen-pirq-msi       ahci2
 117:          0          0          0          0          0          0  xen-pirq-msi       ahci3
 118:          0          0          0          0          0          0  xen-pirq-msi       ahci4
 119:          0          0          0          0          0          0  xen-pirq-msi       ahci5
#2 lore raw

Hello,

Quoted 9 lines · Stefan Bader
Quoted 7 lines · Sander Eikelenboom

Hi.

In 3.19 the device name associates with IRQ's for ahci controllers operating with a single IRQ changed from "ahci?" to "<BDF>", was this intentional ?

It's probably commit 18dcf433f3ded61eb140a55e7048ec2fef79e723 (or another one in that series).

Quoted 6 lines

Oh, looking at that commit that might make sense. In ahci_host_activate_single_irq it now uses dev_driver_name instead of dev_name as it did before (from ata_host_activate). The description sounds like before the driver is registered this will return the bus. And registering an interrupt would likely be before the driver is fully registered... But you probably saw that, too. Whether that really was intentional is still the question. :)

Urgh... Alexander, can you please send a patch to fix up the names?

Thanks.

Signature · 1 line
tejun
#3 lore raw
Quoted 12 lines · Tejun Heo

Hello,

Quoted 9 lines · Stefan Bader
Quoted 7 lines · Sander Eikelenboom

Hi.

In 3.19 the device name associates with IRQ's for ahci controllers operating with a single IRQ changed from "ahci?" to "<BDF>", was this intentional ?

It's probably commit 18dcf433f3ded61eb140a55e7048ec2fef79e723 (or another one in that series).

Quoted 8 lines
Quoted 6 lines

Oh, looking at that commit that might make sense. In ahci_host_activate_single_irq it now uses dev_driver_name instead of dev_name as it did before (from ata_host_activate). The description sounds like before the driver is registered this will return the bus. And registering an interrupt would likely be before the driver is fully registered... But you probably saw that, too. Whether that really was intentional is still the question. :)

Urgh... Alexander, can you please send a patch to fix up the names?

Will try to send over the weekend.

Quoted 4 lines

Thanks.

Signature · 1 line
tejun
Signature · 3 lines
Regards,
Alexander Gordeev
[email protected]
#4 lore raw
In reply to Tejun Heo
Quoted 12 lines · Tejun Heo

Hello,

Quoted 9 lines · Stefan Bader
Quoted 7 lines · Sander Eikelenboom

Hi.

In 3.19 the device name associates with IRQ's for ahci controllers operating with a single IRQ changed from "ahci?" to "<BDF>", was this intentional ?

It's probably commit 18dcf433f3ded61eb140a55e7048ec2fef79e723 (or another one in that series).

Quoted 8 lines
Quoted 6 lines

Oh, looking at that commit that might make sense. In ahci_host_activate_single_irq it now uses dev_driver_name instead of dev_name as it did before (from ata_host_activate). The description sounds like before the driver is registered this will return the bus. And registering an interrupt would likely be before the driver is fully registered... But you probably saw that, too. Whether that really was intentional is still the question. :)

Urgh... Alexander, can you please send a patch to fix up the names?

The problem dates back to commit 4f37b50 ("libata: Use dev_name() for request_irq() to distinguish devices"). If the commit is considered valid then <BDF> still might be legitimate as well. Not sure what is the best approach here.

Quoted 4 lines

Thanks.

Signature · 1 line
tejun
Signature · 3 lines
Regards,
Alexander Gordeev
[email protected]
#5 lore raw

Hello,

Quoted 4 lines · Alexander Gordeev

The problem dates back to commit 4f37b50 ("libata: Use dev_name() for request_irq() to distinguish devices"). If the commit is considered valid then <BDF> still might be legitimate as well. Not sure what is the best approach here.

What does <BDF> indicate? Does it successfully identify the device in question?

Thanks.

Signature · 1 line
tejun
#6 lore raw
Quoted 10 lines · Tejun Heo

Hello,

Quoted 4 lines · Alexander Gordeev

The problem dates back to commit 4f37b50 ("libata: Use dev_name() for request_irq() to distinguish devices"). If the commit is considered valid then <BDF> still might be legitimate as well. Not sure what is the best approach here.

What does <BDF> indicate? Does it successfully identify the device in question?

I suppose, PCI bus-device-function.
Yes, it does.

Quoted 4 lines

Thanks.

Signature · 1 line
tejun
Signature · 3 lines
Regards,
Alexander Gordeev
[email protected]
#7 lore raw
Quoted 9 lines · Alexander Gordeev
Quoted 8 lines · Tejun Heo

Hello,

Quoted 4 lines · Alexander Gordeev

The problem dates back to commit 4f37b50 ("libata: Use dev_name() for request_irq() to distinguish devices"). If the commit is considered valid then <BDF> still might be legitimate as well. Not sure what is the best approach here.

Quoted 5 lines
Quoted 2 lines

What does <BDF> indicate? Does it successfully identify the device in question?

I suppose, PCI bus-device-function.
Yes, it does.

Hmmm... it does stand out like a sore thumb and we probably should use a more descriptive identifier including the driver name. 4f37b50 ("libata: Use dev_name() was fine for the specific platform device as the device identifier included ide but not generally. :(

How about driver name followed by device ID?

Thanks.

Signature · 1 line
tejun
#8 lore raw
Quoted 6 lines · Tejun Heo

Hmmm... it does stand out like a sore thumb and we probably should use a more descriptive identifier including the driver name. 4f37b50 ("libata: Use dev_name() was fine for the specific platform device as the device identifier included ide but not generally. :(

How about driver name followed by device ID?

Or we can move print_id allocation to host allocation time and do

ata%d-%d:DRIVER_NAME

Signature · 1 line
tejun
#9 lore raw
In reply to Alexander Gordeev
Quoted 9 lines · you
Quoted 8 lines · Tejun Heo

Hello,

Quoted 4 lines · Alexander Gordeev

The problem dates back to commit 4f37b50 ("libata: Use dev_name() for request_irq() to distinguish devices"). If the commit is considered valid then <BDF> still might be legitimate as well. Not sure what is the best approach here.

Quoted 5 lines
Quoted 2 lines

What does <BDF> indicate? Does it successfully identify the device in question?

I suppose, PCI bus-device-function.
Yes, it does.

But at least it should be the same for single irq and not single irq, which at present doesn't seem to be the case.

Apart from that you could also combine the devicetype with the BDF, like xen-pciback does that looks like this in /proc/interrupts:
22: 226798 0 0 0 0 0 xen-pirq-ioapic-level xen-pciback[0000:03:06.0]

--
Sander

Quoted 4 lines
Quoted 4 lines

Thanks.

Signature · 1 line
tejun
#10 lore raw
Quoted 2 lines · Sander Eikelenboom

But at least it should be the same for single irq and not single irq, which at present doesn't seem to be the case.

Yeah, I agree the current situation is broken.

Thanks.

Signature · 1 line
tejun