Linux CXL
 help / color / mirror / Atom feed
From: Jonathan Cameron <Jonathan.Cameron@huawei.com>
To: Robert Richter <rrichter@amd.com>
Cc: Alison Schofield <alison.schofield@intel.com>,
	Vishal Verma <vishal.l.verma@intel.com>,
	Ira Weiny <ira.weiny@intel.com>,
	Dan Williams <dan.j.williams@intel.com>,
	Dave Jiang <dave.jiang@intel.com>,
	"Davidlohr Bueso" <dave@stgolabs.net>,
	<linux-cxl@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
	Gregory Price <gourry@gourry.net>,
	"Terry Bowman" <terry.bowman@amd.com>
Subject: Re: [PATCH 2/2] cxl/pci: Check link status and only enable active dports
Date: Fri, 14 Mar 2025 12:14:20 +0000	[thread overview]
Message-ID: <20250314121420.0000069b@huawei.com> (raw)
In-Reply-To: <20250305100123.3077031-3-rrichter@amd.com>

On Wed, 5 Mar 2025 11:01:23 +0100
Robert Richter <rrichter@amd.com> wrote:

> When downstream ports are enumerated, some of them may not be
> connected to a corresponding endpoint or upstream switch port. The
> dport is inactive and its link status is down then. For permanently
> disabled ports a HwInit configuration mechanism (set by hardware or
> firmware) may assign a (further unused) default port number. The port
> number may be set to the same value accross other inactive links.

Spec reference needed.  This sounds like it came from the spec but
searching for it isn't finding anything.

> Those duplicate port numbers cause the downstream port enumeration to
> fail including the root or switch port initialization
> (cxl_switch_port_probe()) and all its active downstream ports.
> 
> Prevent a port initialization failure by checking the link status and
> only enabling active dports. If a dport is inactive, there is no
> matching component (endpoint or switch) connected to and thus, it must
> not be enumerated and added to the kernel's CXL device hierarchy.
> There is no device that will connect to an inactive dport.

> 
> Signed-off-by: Robert Richter <rrichter@amd.com>
> ---
>  drivers/cxl/core/pci.c | 30 ++++++++++++++++++++++++++----
>  1 file changed, 26 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/cxl/core/pci.c b/drivers/cxl/core/pci.c
> index 524b8749cc0b..72683e1b41e3 100644
> --- a/drivers/cxl/core/pci.c
> +++ b/drivers/cxl/core/pci.c
> @@ -32,6 +32,26 @@ struct cxl_walk_context {
>  	int count;
>  };
>  
> +static int get_port_num(struct pci_dev *pdev)
> +{
> +	u32 lnkcap, port_num;
> +	u16 lnksta;
> +
> +	if (pcie_capability_read_dword(pdev, PCI_EXP_LNKCAP, &lnkcap))
> +		return -ENXIO;
> +
> +	/* Skip inactive links */
> +	if (lnkcap & PCI_EXP_LNKCAP_DLLLARC) {
> +		pcie_capability_read_word(pdev, PCI_EXP_LNKSTA, &lnksta);
> +		if (!(lnksta & PCI_EXP_LNKSTA_DLLLA))
> +			return -ENOENT;
> +	}
> +
> +	port_num = FIELD_GET(PCI_EXP_LNKCAP_PN, lnkcap);
> +
> +	return port_num;
> +}
> +
>  static int match_add_dports(struct pci_dev *pdev, void *data)
>  {
>  	struct cxl_walk_context *ctx = data;
> @@ -39,7 +59,7 @@ static int match_add_dports(struct pci_dev *pdev, void *data)
>  	int type = pci_pcie_type(pdev);
>  	struct cxl_register_map map;
>  	struct cxl_dport *dport;
> -	u32 lnkcap, port_num;
> +	int port_num;
>  	int rc;
>  
>  	if (pdev->bus != ctx->bus)
> @@ -48,15 +68,17 @@ static int match_add_dports(struct pci_dev *pdev, void *data)
>  		return 0;
>  	if (type != ctx->type)
>  		return 0;
> -	if (pci_read_config_dword(pdev, pci_pcie_cap(pdev) + PCI_EXP_LNKCAP,
> -				  &lnkcap))
> +
> +	port_num = get_port_num(pdev);
> +	if (port_num == -ENOENT)
> +		pci_dbg(pdev, "Skipping dport, link is down\n");
> +	if (port_num < 0)
>  		return 0;
>  
>  	rc = cxl_find_regblock(pdev, CXL_REGLOC_RBI_COMPONENT, &map);
>  	if (rc)
>  		dev_dbg(&port->dev, "failed to find component registers\n");
>  
> -	port_num = FIELD_GET(PCI_EXP_LNKCAP_PN, lnkcap);
>  	dport = devm_cxl_add_dport(port, &pdev->dev, port_num, map.resource);
>  	if (IS_ERR(dport)) {
>  		rc = PTR_ERR(dport);


  parent reply	other threads:[~2025-03-14 12:14 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-05 10:01 [PATCH 0/2] cxl/pci: Inactive downstream port handling Robert Richter
2025-03-05 10:01 ` [PATCH 1/2] cxl/pci: Ignore downstream ports with duplicate port IDs Robert Richter
2025-03-05 15:09   ` Ira Weiny
2025-03-07 15:28     ` Robert Richter
2025-03-14 12:11   ` Jonathan Cameron
2025-03-05 10:01 ` [PATCH 2/2] cxl/pci: Check link status and only enable active dports Robert Richter
2025-03-05 15:19   ` Ira Weiny
2025-03-07 15:43     ` Robert Richter
2025-03-07 20:51       ` Ira Weiny
2025-03-14 12:14   ` Jonathan Cameron [this message]
2025-03-05 19:06 ` [PATCH 0/2] cxl/pci: Inactive downstream port handling Dan Williams
2025-03-07 14:06   ` Robert Richter
2025-03-07 20:58     ` Dan Williams

Reply instructions:

You may reply publicly 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=20250314121420.0000069b@huawei.com \
    --to=jonathan.cameron@huawei.com \
    --cc=alison.schofield@intel.com \
    --cc=dan.j.williams@intel.com \
    --cc=dave.jiang@intel.com \
    --cc=dave@stgolabs.net \
    --cc=gourry@gourry.net \
    --cc=ira.weiny@intel.com \
    --cc=linux-cxl@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rrichter@amd.com \
    --cc=terry.bowman@amd.com \
    --cc=vishal.l.verma@intel.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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox