Linux-HyperV Archive on lore.kernel.org
 help / color / Atom feed
From: Long Li <longli@microsoft.com>
To: Dexuan Cui <decui@microsoft.com>,
	"longli@linuxonhyperv.com" <longli@linuxonhyperv.com>,
	KY Srinivasan <kys@microsoft.com>,
	Haiyang Zhang <haiyangz@microsoft.com>,
	Stephen Hemminger <sthemmin@microsoft.com>,
	Sasha Levin <sashal@kernel.org>,
	Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>,
	Andrew Murray <andrew.murray@arm.com>,
	Bjorn Helgaas <bhelgaas@google.com>,
	"linux-hyperv@vger.kernel.org" <linux-hyperv@vger.kernel.org>,
	"linux-pci@vger.kernel.org" <linux-pci@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: RE: [EXTERNAL] [PATCH 2/2] PCI: hv: Add support for protocol 1.3 and support PCI_BUS_RELATIONS2
Date: Tue, 3 Dec 2019 00:49:57 +0000
Message-ID: <MW2PR2101MB1132621C7DA3BF2EE1AD1E84CE420@MW2PR2101MB1132.namprd21.prod.outlook.com> (raw)
In-Reply-To: <PU1P153MB0169CFE46920BAEEE11CA9B7BF430@PU1P153MB0169.APCP153.PROD.OUTLOOK.COM>

>Subject: RE: [EXTERNAL] [PATCH 2/2] PCI: hv: Add support for protocol 1.3 and
>support PCI_BUS_RELATIONS2
>
>> From: linux-hyperv-owner@vger.kernel.org
>> Sent: Friday, November 22, 2019 5:57 PM  ...
>> @@ -63,6 +63,7 @@
>>  enum pci_protocol_version_t {
>>  	PCI_PROTOCOL_VERSION_1_1 = PCI_MAKE_VERSION(1, 1),	/*
>Win10
>> */
>>  	PCI_PROTOCOL_VERSION_1_2 = PCI_MAKE_VERSION(1, 2),	/* RS1
>*/
>> +	PCI_PROTOCOL_VERSION_1_3 = PCI_MAKE_VERSION(1, 3),	/* VB
>*/
>>  };
>
>What is "VB" ? Can we use a more meaningful name here? :-)

Vibranium.

>
>> +struct pci_function_description2 {
>> +	u16	v_id;	/* vendor ID */
>> +	u16	d_id;	/* device ID */
>> +	u8	rev;
>> +	u8	prog_intf;
>> +	u8	subclass;
>> +	u8	base_class;
>> +	u32	subsystem_id;
>> +	union win_slot_encoding win_slot;
>
>space -> TAB?
>
>> +/*
>> + * Set NUMA node for the devices on the bus  */ static void
>> +pci_assign_numa_node(struct hv_pcibus_device *hbus)
>
>IMO we'd better add a "hv_" prefix to this function's name, otherwise it looks
>more like a generic PCI subsystem API.

I will send v2 to address comments above.

>
>> +{
>> +	struct pci_dev *dev;
>> +	struct pci_bus *bus = hbus->pci_bus;
>> +	struct hv_pci_dev *hv_dev;
>> +
>> +	list_for_each_entry(dev, &bus->devices, bus_list) {
>> +		hv_dev = get_pcichild_wslot(hbus, devfn_to_wslot(dev-
>>devfn));
>> +		if (!hv_dev)
>> +			continue;
>> +
>> +		if (hv_dev->desc.flags &
>HV_PCI_DEVICE_FLAG_NUMA_AFFINITY)
>> +			set_dev_node(&dev->dev, hv_dev-
>>desc.virtual_numa_node);
>> +	}
>> +}
>
>Can you please give a brief background introduction to dev->numa_node, e.g.
>how is it used here? -- is it used when a PCI device driver (e.g. the mlx
>driver) asks the kernel to allocate memory for DMA, or allocate MSI interrupts?
>How big is the performance gain in the tests? I'm curious as it looks unclear to
>me how dev->numa_node is used by the PCI subsystem.

numa_node can be used by a device driver (if it's numa aware) to setup its internal data structures and allocate its MSI.

As an example, you can look at "drivers/net/ethernet/mellanox/mlx4/main.c: mlx4_load_one()":
It stores the value at : dev->numa_node = dev_to_node(&pdev->dev);
after that, dev->numa_node  is used through driver.

numa_node is also exported though /sys to user-mode, so a NUMA aware application (e.g. MPI) can figure out how to pin itself to selected CPUs for the best latency.

>
>Thanks,
>-- Dexuan

  reply index

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-23  1:57 [PATCH 1/2] PCI: hv: decouple the func definition in hv_dr_state from VSP message longli
2019-11-23  1:57 ` [PATCH 2/2] PCI: hv: Add support for protocol 1.3 and support PCI_BUS_RELATIONS2 longli
2019-11-30  4:45   ` [EXTERNAL] " Michael Kelley
2019-12-02 21:23     ` Long Li
2019-12-02 23:59   ` Dexuan Cui
2019-12-03  0:49     ` Long Li [this message]
2019-11-30  4:30 ` [EXTERNAL] [PATCH 1/2] PCI: hv: decouple the func definition in hv_dr_state from VSP message Michael Kelley
2019-12-02 23:29 ` Dexuan Cui
2019-12-03  0:39   ` Long Li

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=MW2PR2101MB1132621C7DA3BF2EE1AD1E84CE420@MW2PR2101MB1132.namprd21.prod.outlook.com \
    --to=longli@microsoft.com \
    --cc=andrew.murray@arm.com \
    --cc=bhelgaas@google.com \
    --cc=decui@microsoft.com \
    --cc=haiyangz@microsoft.com \
    --cc=kys@microsoft.com \
    --cc=linux-hyperv@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=longli@linuxonhyperv.com \
    --cc=lorenzo.pieralisi@arm.com \
    --cc=sashal@kernel.org \
    --cc=sthemmin@microsoft.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-HyperV Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-hyperv/0 linux-hyperv/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-hyperv linux-hyperv/ https://lore.kernel.org/linux-hyperv \
		linux-hyperv@vger.kernel.org
	public-inbox-index linux-hyperv

Example config snippet for mirrors

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


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