[1/2] r8169: Don't disable ASPM in the driver

Enable or disable ASPM should be done in PCI core instead of in the
device driver.

Commit ba04c7c93bbc ("r8169: disable ASPM") uses
pci_disable_link_state() to disable ASPM. This is incorrect, if the
device really needs to disable ASPM, we should use a quirk in PCI core
to prevent the PCI core from setting ASPM altogether.

Let's remove pci_disable_link_state() for now. Use PCI core quirks if
any regression happens.

Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
---
v2:
- Remove module parameter.
- Remove pci_disable_link_state().

 drivers/net/ethernet/realtek/r8169.c | 5 -----
 1 file changed, 5 deletions(-)

On 12.06.2018 11:57, Kai-Heng Feng wrote:
> Enable or disable ASPM should be done in PCI core instead of in the
> device driver.
> 
> Commit ba04c7c93bbc ("r8169: disable ASPM") uses
> pci_disable_link_state() to disable ASPM. This is incorrect, if the
> device really needs to disable ASPM, we should use a quirk in PCI core
> to prevent the PCI core from setting ASPM altogether.
> 
I wouldn't call using pci_disable_link_state() in a driver incorrect
(as it works), there is just a better way which is more in line with
the PCI subsystem architecture.

> Let's remove pci_disable_link_state() for now. Use PCI core quirks if
> any regression happens.
> 
The vendor driver disables ASPM unconditionally for chip version 25
(there it's METHOD_9), so I think ASPM support is broken in this chip
version. I'll cook a PCI quirk.

> Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>

Please note that netdev is closed currently. Once 4.18-RC1 is out it
will be re-opened. Then please re-submit properly annotating PATCH
with "net-next" (I've forgotten this often enough myself).

> ---
> v2:
> - Remove module parameter.
> - Remove pci_disable_link_state().
> 
>  drivers/net/ethernet/realtek/r8169.c | 5 -----
>  1 file changed, 5 deletions(-)
> 
> diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c
> index 75dfac0248f4..9b55ce513a36 100644
> --- a/drivers/net/ethernet/realtek/r8169.c
> +++ b/drivers/net/ethernet/realtek/r8169.c
> @@ -25,7 +25,6 @@
>  #include <linux/dma-mapping.h>
>  #include <linux/pm_runtime.h>
>  #include <linux/firmware.h>
> -#include <linux/pci-aspm.h>
>  #include <linux/prefetch.h>
>  #include <linux/ipv6.h>
>  #include <net/ip6_checksum.h>
> @@ -7647,10 +7646,6 @@ static int rtl_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
>  	mii->reg_num_mask = 0x1f;
>  	mii->supports_gmii = cfg->has_gmii;
>  
> -	/* disable ASPM completely as that cause random device stop working
> -	 * problems as well as full system hangs for some PCIe devices users */
> -	pci_disable_link_state(pdev, PCIE_LINK_STATE_L0S | PCIE_LINK_STATE_L1 |
> -				     PCIE_LINK_STATE_CLKPM);
>  
>  	/* enable device (incl. PCI PM wakeup and hotplug setup) */
>  	rc = pcim_enable_device(pdev);
>

Hi Heiner,

> On Jun 13, 2018, at 3:30 AM, Heiner Kallweit <hkallweit1@gmail.com> wrote:
>
> On 12.06.2018 11:57, Kai-Heng Feng wrote:
>> Enable or disable ASPM should be done in PCI core instead of in the
>> device driver.
>>
>> Commit ba04c7c93bbc ("r8169: disable ASPM") uses
>> pci_disable_link_state() to disable ASPM. This is incorrect, if the
>> device really needs to disable ASPM, we should use a quirk in PCI core
>> to prevent the PCI core from setting ASPM altogether.
> I wouldn't call using pci_disable_link_state() in a driver incorrect
> (as it works), there is just a better way which is more in line with
> the PCI subsystem architecture.

Ok, I'll amend the commit log in next version.

>
>> Let's remove pci_disable_link_state() for now. Use PCI core quirks if
>> any regression happens.
> The vendor driver disables ASPM unconditionally for chip version 25
> (there it's METHOD_9), so I think ASPM support is broken in this chip
> version. I'll cook a PCI quirk.

I actually asked Ryankao about this. He said that variant is more then a  
decades old and he can't find why it doesn't support ASPM.

Since METHOD_9 might be a platform issue instead, my intention was to  
enable ASPM for all variants. If users hit any issue, then we can introduce  
new PCI quirks.

>
>> Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
>
> Please note that netdev is closed currently. Once 4.18-RC1 is out it
> will be re-opened. Then please re-submit properly annotating PATCH
> with "net-next" (I've forgotten this often enough myself).

Will do for next version. Thanks!

Kai-Heng

>
>> ---
>> v2:
>> - Remove module parameter.
>> - Remove pci_disable_link_state().
>>
>>  drivers/net/ethernet/realtek/r8169.c | 5 -----
>>  1 file changed, 5 deletions(-)
>>
>> diff --git a/drivers/net/ethernet/realtek/r8169.c  
>> b/drivers/net/ethernet/realtek/r8169.c
>> index 75dfac0248f4..9b55ce513a36 100644
>> --- a/drivers/net/ethernet/realtek/r8169.c
>> +++ b/drivers/net/ethernet/realtek/r8169.c
>> @@ -25,7 +25,6 @@
>>  #include <linux/dma-mapping.h>
>>  #include <linux/pm_runtime.h>
>>  #include <linux/firmware.h>
>> -#include <linux/pci-aspm.h>
>>  #include <linux/prefetch.h>
>>  #include <linux/ipv6.h>
>>  #include <net/ip6_checksum.h>
>> @@ -7647,10 +7646,6 @@ static int rtl_init_one(struct pci_dev *pdev,  
>> const struct pci_device_id *ent)
>>  	mii->reg_num_mask = 0x1f;
>>  	mii->supports_gmii = cfg->has_gmii;
>>
>> -	/* disable ASPM completely as that cause random device stop working
>> -	 * problems as well as full system hangs for some PCIe devices users */
>> -	pci_disable_link_state(pdev, PCIE_LINK_STATE_L0S | PCIE_LINK_STATE_L1 |
>> -				     PCIE_LINK_STATE_CLKPM);
>>
>>  	/* enable device (incl. PCI PM wakeup and hotplug setup) */
>>  	rc = pcim_enable_device(pdev);