Linux-PCI Archive on lore.kernel.org
 help / color / Atom feed
From: Heiner Kallweit <hkallweit1@gmail.com>
To: Greg KH <gregkh@linuxfoundation.org>
Cc: Frederick Lawler <fred@fredlawl.com>,
	Bjorn Helgaas <bhelgaas@google.com>,
	Rajat Jain <rajatja@google.com>,
	"linux-pci@vger.kernel.org" <linux-pci@vger.kernel.org>
Subject: Re: [PATCH v2 3/3] PCI/ASPM: add sysfs attributes for controlling ASPM
Date: Thu, 22 Aug 2019 21:15:31 +0200
Message-ID: <e4a278f1-ac78-a2fd-ddfe-a4dd0d1685e3@gmail.com> (raw)
In-Reply-To: <20190822130503.GD12941@kroah.com>

On 22.08.2019 15:05, Greg KH wrote:
> On Wed, Aug 21, 2019 at 08:18:41PM +0200, Heiner Kallweit wrote:
>> Background of this extension is a problem with the r8169 network driver.
>> Several combinations of board chipsets and network chip versions have
>> problems if ASPM is enabled, therefore we have to disable ASPM per default.
>> However especially on notebooks ASPM can provide significant power-saving,
>> therefore we want to give users the option to enable ASPM. With the new sysfs
>> attributes users can control which ASPM link-states are enabled/disabled.
>>
>> Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
>> ---
>> v2:
>> - use a dedicated sysfs attribute per link state
>> - allow separate control of ASPM and PCI PM L1 sub-states
>> ---
>>  Documentation/ABI/testing/sysfs-bus-pci |  13 ++
>>  drivers/pci/pci.h                       |   8 +-
>>  drivers/pci/pcie/aspm.c                 | 263 +++++++++++++++++++++++-
>>  3 files changed, 276 insertions(+), 8 deletions(-)
>>
>> diff --git a/Documentation/ABI/testing/sysfs-bus-pci b/Documentation/ABI/testing/sysfs-bus-pci
>> index 8bfee557e..38b565c30 100644
>> --- a/Documentation/ABI/testing/sysfs-bus-pci
>> +++ b/Documentation/ABI/testing/sysfs-bus-pci
>> @@ -347,3 +347,16 @@ Description:
>>  		If the device has any Peer-to-Peer memory registered, this
>>  	        file contains a '1' if the memory has been published for
>>  		use outside the driver that owns the device.
>> +
>> +What		/sys/bus/pci/devices/.../power/aspm_l0s
>> +What		/sys/bus/pci/devices/.../power/aspm_l1
>> +What		/sys/bus/pci/devices/.../power/aspm_l1_1
>> +What		/sys/bus/pci/devices/.../power/aspm_l1_2
>> +What		/sys/bus/pci/devices/.../power/aspm_l1_1_pcipm
>> +What		/sys/bus/pci/devices/.../power/aspm_l1_2_pcipm
>> +What		/sys/bus/pci/devices/.../power/aspm_clkpm
> 
> Wait, there already is a "power" subdirectory for all devices in the
> system, are you adding another one, or files to that already-created
> directory?
> 
This is the standard "power" directory dynamically created by
dpm_sysfs_add().

>> +date:		August 2019
>> +Contact:	Heiner Kallweit <hkallweit1@gmail.com>
>> +Description:	If ASPM is supported for an endpoint, then these files
>> +		can be used to disable or enable the individual
>> +		power management states.
>> diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
>> index 3f126f808..e51c91f38 100644
>> --- a/drivers/pci/pci.h
>> +++ b/drivers/pci/pci.h
>> @@ -525,17 +525,13 @@ void pcie_aspm_init_link_state(struct pci_dev *pdev);
>>  void pcie_aspm_exit_link_state(struct pci_dev *pdev);
>>  void pcie_aspm_pm_state_change(struct pci_dev *pdev);
>>  void pcie_aspm_powersave_config_link(struct pci_dev *pdev);
>> +void pcie_aspm_create_sysfs_dev_files(struct pci_dev *pdev);
>> +void pcie_aspm_remove_sysfs_dev_files(struct pci_dev *pdev);
>>  #else
>>  static inline void pcie_aspm_init_link_state(struct pci_dev *pdev) { }
>>  static inline void pcie_aspm_exit_link_state(struct pci_dev *pdev) { }
>>  static inline void pcie_aspm_pm_state_change(struct pci_dev *pdev) { }
>>  static inline void pcie_aspm_powersave_config_link(struct pci_dev *pdev) { }
>> -#endif
>> -
>> -#ifdef CONFIG_PCIEASPM_DEBUG
>> -void pcie_aspm_create_sysfs_dev_files(struct pci_dev *pdev);
>> -void pcie_aspm_remove_sysfs_dev_files(struct pci_dev *pdev);
>> -#else
>>  static inline void pcie_aspm_create_sysfs_dev_files(struct pci_dev *pdev) { }
>>  static inline void pcie_aspm_remove_sysfs_dev_files(struct pci_dev *pdev) { }
>>  #endif
>> diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c
>> index 149b876c9..e7dfcd1bd 100644
>> --- a/drivers/pci/pcie/aspm.c
>> +++ b/drivers/pci/pcie/aspm.c
>> @@ -1275,38 +1275,297 @@ static ssize_t clk_ctl_store(struct device *dev,
>>  
>>  static DEVICE_ATTR_RW(link_state);
>>  static DEVICE_ATTR_RW(clk_ctl);
>> +#endif
>> +
>> +static const char power_group[] = "power";
>> +
>> +static struct pcie_link_state *aspm_get_parent_link(struct pci_dev *pdev)
>> +{
>> +	struct pci_dev *parent = pdev->bus->self;
>> +
>> +	if (pdev->has_secondary_link)
>> +		parent = pdev;
>> +
>> +	return parent ? parent->link_state : NULL;
>> +}
>> +
>> +static bool pcie_check_valid_aspm_endpoint(struct pci_dev *pdev)
>> +{
>> +	struct pcie_link_state *link;
>> +
>> +	if (pci_pcie_type(pdev) != PCI_EXP_TYPE_ENDPOINT)
>> +		return false;
>> +
>> +	link = aspm_get_parent_link(pdev);
>> +
>> +	return link && link->aspm_capable;
>> +}
>> +
>> +static ssize_t aspm_attr_show_common(struct device *dev,
>> +				     struct device_attribute *attr,
>> +				     char *buf, int state)
>> +{
>> +	struct pci_dev *pdev = to_pci_dev(dev);
>> +	struct pcie_link_state *link;
>> +	int val;
>> +
>> +	link = aspm_get_parent_link(pdev);
>> +	if (!link)
>> +		return -EOPNOTSUPP;
>> +
>> +	mutex_lock(&aspm_lock);
>> +	val = !!(link->aspm_enabled & state);
>> +	mutex_unlock(&aspm_lock);
>> +
>> +	return snprintf(buf, PAGE_SIZE, "%d\n", val);
> 
> For sysfs files, you never need to use snprintf() as you "know" the size
> of the buffer is big, and you are just printing out a single number.  So
> that can be cleaned up in all of these attribute files.
> 
>> +}
>> +
>> +static ssize_t aspm_l0s_show(struct device *dev,
>> +			     struct device_attribute *attr, char *buf)
>> +{
>> +	return aspm_attr_show_common(dev, attr, buf, ASPM_STATE_L0S);
>> +}
>> +
>> +static ssize_t aspm_l1_show(struct device *dev,
>> +			    struct device_attribute *attr, char *buf)
>> +{
>> +	return aspm_attr_show_common(dev, attr, buf, ASPM_STATE_L1);
>> +}
>> +
>> +static ssize_t aspm_l1_1_show(struct device *dev,
>> +			      struct device_attribute *attr, char *buf)
>> +{
>> +	return aspm_attr_show_common(dev, attr, buf, ASPM_STATE_L1_1);
>> +}
>> +
>> +static ssize_t aspm_l1_2_show(struct device *dev,
>> +			      struct device_attribute *attr, char *buf)
>> +{
>> +	return aspm_attr_show_common(dev, attr, buf, ASPM_STATE_L1_2);
>> +}
>> +
>> +static ssize_t aspm_l1_1_pcipm_show(struct device *dev,
>> +				    struct device_attribute *attr, char *buf)
>> +{
>> +	return aspm_attr_show_common(dev, attr, buf, ASPM_STATE_L1_1_PCIPM);
>> +}
>> +
>> +static ssize_t aspm_l1_2_pcipm_show(struct device *dev,
>> +				    struct device_attribute *attr, char *buf)
>> +{
>> +	return aspm_attr_show_common(dev, attr, buf, ASPM_STATE_L1_2_PCIPM);
>> +}
>> +
>> +static ssize_t aspm_attr_store_common(struct device *dev,
>> +				      struct device_attribute *attr,
>> +				      const char *buf, size_t len, int state)
>> +{
>> +	struct pci_dev *pdev = to_pci_dev(dev);
>> +	struct pcie_link_state *link;
>> +	bool state_enable;
>> +
>> +	if (aspm_disabled)
>> +		return -EPERM;
>> +
>> +	link = aspm_get_parent_link(pdev);
>> +	if (!link)
>> +		return -EOPNOTSUPP;
>> +
>> +	if (!(link->aspm_capable & state))
>> +		return -EOPNOTSUPP;
>> +
>> +	if (strtobool(buf, &state_enable) < 0)
>> +		return -EINVAL;
>> +
>> +	down_read(&pci_bus_sem);
>> +	mutex_lock(&aspm_lock);
>> +
>> +	if (state_enable)
>> +		link->aspm_disable &= ~state;
>> +	else
>> +		link->aspm_disable |= state;
>> +
>> +	if (link->aspm_disable & ASPM_STATE_L1)
>> +		link->aspm_disable |= ASPM_STATE_L1SS;
>> +
>> +	pcie_config_aspm_link(link, policy_to_aspm_state(link));
> 
> This function can't fail?
> 
It has no return value.

>> +
>> +	mutex_unlock(&aspm_lock);
>> +	up_read(&pci_bus_sem);
>> +
>> +	return len;
>> +}
>> +
>> +static ssize_t aspm_l0s_store(struct device *dev,
>> +			      struct device_attribute *attr,
>> +			      const char *buf, size_t len)
>> +{
>> +	return aspm_attr_store_common(dev, attr, buf, len, ASPM_STATE_L0S);
>> +}
>> +
>> +static ssize_t aspm_l1_store(struct device *dev,
>> +			     struct device_attribute *attr,
>> +			     const char *buf, size_t len)
>> +{
>> +	return aspm_attr_store_common(dev, attr, buf, len, ASPM_STATE_L1);
>> +}
>> +
>> +static ssize_t aspm_l1_1_store(struct device *dev,
>> +			       struct device_attribute *attr,
>> +			       const char *buf, size_t len)
>> +{
>> +	return aspm_attr_store_common(dev, attr, buf, len, ASPM_STATE_L1_1);
>> +}
>> +
>> +static ssize_t aspm_l1_2_store(struct device *dev,
>> +			       struct device_attribute *attr,
>> +			       const char *buf, size_t len)
>> +{
>> +	return aspm_attr_store_common(dev, attr, buf, len, ASPM_STATE_L1_2);
>> +}
>> +
>> +static ssize_t aspm_l1_1_pcipm_store(struct device *dev,
>> +				     struct device_attribute *attr,
>> +				     const char *buf, size_t len)
>> +{
>> +	return aspm_attr_store_common(dev, attr, buf, len,
>> +				      ASPM_STATE_L1_1_PCIPM);
>> +}
>> +
>> +static ssize_t aspm_l1_2_pcipm_store(struct device *dev,
>> +				     struct device_attribute *attr,
>> +				     const char *buf, size_t len)
>> +{
>> +	return aspm_attr_store_common(dev, attr, buf, len,
>> +				      ASPM_STATE_L1_2_PCIPM);
>> +}
>> +
>> +static ssize_t aspm_clkpm_show(struct device *dev,
>> +			       struct device_attribute *attr, char *buf)
>> +{
>> +	struct pci_dev *pdev = to_pci_dev(dev);
>> +	struct pcie_link_state *link;
>> +	int val;
>> +
>> +	link = aspm_get_parent_link(pdev);
>> +	if (!link)
>> +		return -EOPNOTSUPP;
>> +
>> +	mutex_lock(&aspm_lock);
>> +	val = link->clkpm_enabled;
>> +	mutex_unlock(&aspm_lock);
>> +
>> +	return snprintf(buf, PAGE_SIZE, "%d\n", val);
>> +}
>> +
>> +static ssize_t aspm_clkpm_store(struct device *dev,
>> +				struct device_attribute *attr,
>> +				const char *buf, size_t len)
>> +{
>> +	struct pci_dev *pdev = to_pci_dev(dev);
>> +	struct pcie_link_state *link;
>> +	bool state_enable;
>> +
>> +	if (aspm_disabled)
>> +		return -EPERM;
>> +
>> +	link = aspm_get_parent_link(pdev);
>> +	if (!link)
>> +		return -EOPNOTSUPP;
>> +
>> +	if (!link->clkpm_capable)
>> +		return -EOPNOTSUPP;
>> +
>> +	if (strtobool(buf, &state_enable) < 0)
>> +		return -EINVAL;
>> +
>> +	down_read(&pci_bus_sem);
>> +	mutex_lock(&aspm_lock);
>> +
>> +	link->clkpm_disable = !state_enable;
>> +	pcie_set_clkpm(link, policy_to_clkpm_state(link));
>> +
>> +	mutex_unlock(&aspm_lock);
>> +	up_read(&pci_bus_sem);
>> +
>> +	return len;
>> +}
>> +
>> +static DEVICE_ATTR_RW(aspm_l0s);
>> +static DEVICE_ATTR_RW(aspm_l1);
>> +static DEVICE_ATTR_RW(aspm_l1_1);
>> +static DEVICE_ATTR_RW(aspm_l1_2);
>> +static DEVICE_ATTR_RW(aspm_l1_1_pcipm);
>> +static DEVICE_ATTR_RW(aspm_l1_2_pcipm);
>> +static DEVICE_ATTR_RW(aspm_clkpm);
>>  
>> -static char power_group[] = "power";
>>  void pcie_aspm_create_sysfs_dev_files(struct pci_dev *pdev)
>>  {
>>  	struct pcie_link_state *link_state = pdev->link_state;
>>  
>> +	if (pcie_check_valid_aspm_endpoint(pdev)) {
>> +		sysfs_add_file_to_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l0s.attr, power_group);
>> +		sysfs_add_file_to_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1.attr, power_group);
>> +		sysfs_add_file_to_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1_1.attr, power_group);
>> +		sysfs_add_file_to_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1_2.attr, power_group);
>> +		sysfs_add_file_to_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1_1_pcipm.attr, power_group);
>> +		sysfs_add_file_to_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1_2_pcipm.attr, power_group);
>> +		sysfs_add_file_to_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_clkpm.attr, power_group);
> 
> Huh?  First of, if you are ever in a driver, and you have to call a
> sysfs_* function, you know something is wrong.
> 
> Why are you dynamically adding files to a group?  These should all be
> statically allocated and then at creation time you can dynamically
> determine if you need to show them or not.  Use the is_visable callback
> for that.
> 

The "power" group is dynamically created by the base driver core
(dpm_sysfs_add). Not sure how the ASPM sysfs files could be statically
allocated. We could add an attribute group to the "pci" bus_type,
but this doesn't feel right.
What we could do to make it simpler:
Create an attribute group and then use sysfs_merge_group().
But I see no way not to use sysfs_xxx functions. Also functions
like device_add_groups() are just very simple wrappers around sysfs_
functions.

Alternatively we could not use the existing "power" group but create
our own group with device_add_group(). But I don't see that we gain much.

>> +	}
>> +
>>  	if (!link_state)
>>  		return;
>>  
>> +#ifdef CONFIG_PCIEASPM_DEBUG
>>  	if (link_state->aspm_support)
>>  		sysfs_add_file_to_group(&pdev->dev.kobj,
>>  			&dev_attr_link_state.attr, power_group);
>>  	if (link_state->clkpm_capable)
>>  		sysfs_add_file_to_group(&pdev->dev.kobj,
>>  			&dev_attr_clk_ctl.attr, power_group);
> 
> Same here.
> 
>> +#endif
>>  }
>>  
>>  void pcie_aspm_remove_sysfs_dev_files(struct pci_dev *pdev)
>>  {
>>  	struct pcie_link_state *link_state = pdev->link_state;
>>  
>> +	if (pcie_check_valid_aspm_endpoint(pdev)) {
>> +		sysfs_remove_file_from_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l0s.attr, power_group);
>> +		sysfs_remove_file_from_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1.attr, power_group);
>> +		sysfs_remove_file_from_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1_1.attr, power_group);
>> +		sysfs_remove_file_from_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1_2.attr, power_group);
>> +		sysfs_remove_file_from_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1_1_pcipm.attr, power_group);
>> +		sysfs_remove_file_from_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_l1_2_pcipm.attr, power_group);
>> +		sysfs_remove_file_from_group(&pdev->dev.kobj,
>> +			&dev_attr_aspm_clkpm.attr, power_group);
> 
> And then you never have to do this type of messy logic at all.
> 
> Ick.
> 
>> +	}
>> +
>>  	if (!link_state)
>>  		return;
>>  
>> +#ifdef CONFIG_PCIEASPM_DEBUG
>>  	if (link_state->aspm_support)
>>  		sysfs_remove_file_from_group(&pdev->dev.kobj,
>>  			&dev_attr_link_state.attr, power_group);
>>  	if (link_state->clkpm_capable)
>>  		sysfs_remove_file_from_group(&pdev->dev.kobj,
>>  			&dev_attr_clk_ctl.attr, power_group);
> 
> And you get to drop this nonsense as well.
> 
Right, with the new attributes CONFIG_PCIEASPM_DEBUG and all
related stuff could be removed most likely.

> thanks,
> 
> greg k-h
> 

Heiner

  reply index

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-21 18:14 [PATCH v2 0/3] " Heiner Kallweit
2019-08-21 18:17 ` [PATCH v2 1/3] PCI/ASPM: add L1 sub-state support to pci_disable_link_state Heiner Kallweit
2019-08-21 18:17 ` [PATCH v2 2/3] PCI/ASPM: allow to re-enable Clock PM Heiner Kallweit
2019-08-21 18:18 ` [PATCH v2 3/3] PCI/ASPM: add sysfs attributes for controlling ASPM Heiner Kallweit
2019-08-22 13:05   ` Greg KH
2019-08-22 19:15     ` Heiner Kallweit [this message]
2019-08-22 19:46       ` Heiner Kallweit

Reply instructions:

You may reply publically to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=e4a278f1-ac78-a2fd-ddfe-a4dd0d1685e3@gmail.com \
    --to=hkallweit1@gmail.com \
    --cc=bhelgaas@google.com \
    --cc=fred@fredlawl.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=rajatja@google.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Linux-PCI Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-pci/0 linux-pci/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-pci linux-pci/ https://lore.kernel.org/linux-pci \
		linux-pci@vger.kernel.org
	public-inbox-index linux-pci

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-pci


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git