public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: "Ilpo Järvinen" <ilpo.jarvinen@linux.intel.com>
To: "Michał Winiarski" <michal.winiarski@intel.com>
Cc: linux-pci@vger.kernel.org, intel-xe@lists.freedesktop.org,
	dri-devel@lists.freedesktop.org,
	LKML <linux-kernel@vger.kernel.org>,
	"Bjorn Helgaas" <bhelgaas@google.com>,
	"Christian König" <christian.koenig@amd.com>,
	"Krzysztof Wilczyński" <kw@linux.com>,
	"Rodrigo Vivi" <rodrigo.vivi@intel.com>,
	"Michal Wajdeczko" <michal.wajdeczko@intel.com>,
	"Lucas De Marchi" <lucas.demarchi@intel.com>,
	"Thomas Hellström" <thomas.hellstrom@linux.intel.com>,
	"Maarten Lankhorst" <maarten.lankhorst@linux.intel.com>,
	"Maxime Ripard" <mripard@kernel.org>,
	"Thomas Zimmermann" <tzimmermann@suse.de>,
	"David Airlie" <airlied@gmail.com>,
	"Simona Vetter" <simona@ffwll.ch>,
	"Matt Roper" <matthew.d.roper@intel.com>
Subject: Re: [PATCH v6 5/6] PCI: Allow drivers to control VF BAR size
Date: Wed, 26 Mar 2025 17:22:50 +0200 (EET)	[thread overview]
Message-ID: <7374beef-46ed-ab53-ccb5-48565526545c@linux.intel.com> (raw)
In-Reply-To: <20250320110854.3866284-6-michal.winiarski@intel.com>

[-- Attachment #1: Type: text/plain, Size: 5134 bytes --]

On Thu, 20 Mar 2025, Michał Winiarski wrote:

> Drivers could leverage the fact that the VF BAR MMIO reservation is
> created for total number of VFs supported by the device by resizing the
> BAR to larger size when smaller number of VFs is enabled.
> 
> Add a pci_iov_vf_bar_set_size() function to control the size and a
> pci_iov_vf_bar_get_sizes() helper to get the VF BAR sizes that will
> allow up to num_vfs to be successfully enabled with the current
> underlying reservation size.
> 
> Signed-off-by: Michał Winiarski <michal.winiarski@intel.com>
> ---
>  drivers/pci/iov.c   | 78 +++++++++++++++++++++++++++++++++++++++++++++
>  include/linux/pci.h |  6 ++++
>  2 files changed, 84 insertions(+)
> 
> diff --git a/drivers/pci/iov.c b/drivers/pci/iov.c
> index 861273ad9a580..751eef232685c 100644
> --- a/drivers/pci/iov.c
> +++ b/drivers/pci/iov.c
> @@ -1291,3 +1291,81 @@ int pci_sriov_configure_simple(struct pci_dev *dev, int nr_virtfn)
>  	return nr_virtfn;
>  }
>  EXPORT_SYMBOL_GPL(pci_sriov_configure_simple);
> +
> +/**
> + * pci_iov_vf_bar_set_size - set a new size for a VF BAR
> + * @dev: the PCI device
> + * @resno: the resource number
> + * @size: new size as defined in the spec (0=1MB, 31=128TB)
> + *
> + * Set the new size of a VF BAR that supports VF resizable BAR capability.
> + * Unlike pci_resize_resource(), this does not cause the resource that
> + * reserves the MMIO space (originally up to total_VFs) to be resized, which
> + * means that following calls to pci_enable_sriov() can fail if the resources
> + * no longer fit.
> + *
> + * Returns 0 on success, or negative on failure.

Return: is the correct kernel doc style.

> + */
> +int pci_iov_vf_bar_set_size(struct pci_dev *dev, int resno, int size)
> +{
> +	int ret;
> +	u32 sizes;
> +
> +	if (!pci_resource_is_iov(resno))
> +		return -EINVAL;
> +
> +	if (pci_iov_is_memory_decoding_enabled(dev))
> +		return -EBUSY;
> +
> +	sizes = pci_rebar_get_possible_sizes(dev, resno);
> +	if (!sizes)
> +		return -ENOTSUPP;
> +
> +	if (!(sizes & BIT(size)))
> +		return -EINVAL;
> +
> +	ret = pci_rebar_set_size(dev, resno, size);
> +	if (ret)
> +		return ret;
> +
> +	pci_iov_resource_set_size(dev, resno, pci_rebar_size_to_bytes(size));
> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_GPL(pci_iov_vf_bar_set_size);
> +
> +/**
> + * pci_iov_vf_bar_get_sizes - get VF BAR sizes allowing to create up to num_vfs
> + * @dev: the PCI device
> + * @resno: the resource number
> + * @num_vfs: number of VFs
> + *
> + * Get the sizes of a VF resizable BAR that can be accommodated within the
> + * resource that reserves the MMIO space if num_vfs are enabled.

I'd rephrase to:

Get the sizes of a VF resizable BAR that can be accommodate @num_vfs 
within the currently assigned size of the resource @resno.

> + *
> + * Returns 0 if BAR isn't resizable, otherwise returns a bitmask in format

Return:

a bitmask of sizes

> + * defined in the spec (bit 0=1MB, bit 31=128TB).
> + */
> +u32 pci_iov_vf_bar_get_sizes(struct pci_dev *dev, int resno, int num_vfs)
> +{
> +	resource_size_t size;
> +	u32 sizes;
> +	int i;
> +
> +	sizes = pci_rebar_get_possible_sizes(dev, resno);
> +	if (!sizes)
> +		return 0;
> +
> +	while (sizes > 0) {
> +		i = __fls(sizes);
> +		size = pci_rebar_size_to_bytes(i);
> +
> +		if (size * num_vfs <= pci_resource_len(dev, resno))
> +			break;
> +
> +		sizes &= ~BIT(i);
> +	}

Couldn't this be handled without a loop:

	bar_sizes = (round_up(pci_resource_len(dev, resno) / num_vfs) - 1) >>
		    ilog2(SZ_1M);

	sizes &= bar_sizes;

(Just to given an idea, I wrote this into the email so it might contain 
some off-by-one errors or like).

> +
> +	return sizes;
> +}
> +EXPORT_SYMBOL_GPL(pci_iov_vf_bar_get_sizes);
> diff --git a/include/linux/pci.h b/include/linux/pci.h
> index 0e8e3fd77e967..c8708f3749757 100644
> --- a/include/linux/pci.h
> +++ b/include/linux/pci.h
> @@ -2389,6 +2389,8 @@ int pci_sriov_set_totalvfs(struct pci_dev *dev, u16 numvfs);
>  int pci_sriov_get_totalvfs(struct pci_dev *dev);
>  int pci_sriov_configure_simple(struct pci_dev *dev, int nr_virtfn);
>  resource_size_t pci_iov_resource_size(struct pci_dev *dev, int resno);
> +int pci_iov_vf_bar_set_size(struct pci_dev *dev, int resno, int size);
> +u32 pci_iov_vf_bar_get_sizes(struct pci_dev *dev, int resno, int num_vfs);
>  void pci_vf_drivers_autoprobe(struct pci_dev *dev, bool probe);
>  
>  /* Arch may override these (weak) */
> @@ -2441,6 +2443,10 @@ static inline int pci_sriov_get_totalvfs(struct pci_dev *dev)
>  #define pci_sriov_configure_simple	NULL
>  static inline resource_size_t pci_iov_resource_size(struct pci_dev *dev, int resno)
>  { return 0; }
> +static inline int pci_iov_vf_bar_set_size(struct pci_dev *dev, int resno, int size)
> +{ return -ENODEV; }
> +static inline u32 pci_iov_vf_bar_get_sizes(struct pci_dev *dev, int resno, int num_vfs)
> +{ return 0; }
>  static inline void pci_vf_drivers_autoprobe(struct pci_dev *dev, bool probe) { }
>  #endif
>  
> 

-- 
 i.

  reply	other threads:[~2025-03-26 15:23 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-20 11:08 [PATCH v6 0/6] PCI: VF resizable BAR Michał Winiarski
2025-03-20 11:08 ` [PATCH v6 1/6] PCI/IOV: Restore VF resizable BAR state after reset Michał Winiarski
2025-03-26 14:42   ` Ilpo Järvinen
2025-03-26 14:52     ` Ilpo Järvinen
2025-04-02 10:20       ` Michał Winiarski
2025-03-20 11:08 ` [PATCH v6 2/6] PCI: Add a helper to convert between VF BAR number and IOV resource Michał Winiarski
2025-03-26 14:46   ` Ilpo Järvinen
2025-04-02 10:23     ` Michał Winiarski
2025-03-20 11:08 ` [PATCH v6 3/6] PCI: Allow IOV resources to be resized in pci_resize_resource() Michał Winiarski
2025-03-26 14:58   ` Ilpo Järvinen
2025-04-02 10:25     ` Michał Winiarski
2025-03-20 11:08 ` [PATCH v6 4/6] PCI/IOV: Check that VF BAR fits within the reservation Michał Winiarski
2025-03-26 15:11   ` Ilpo Järvinen
2025-03-28 16:39     ` Ilpo Järvinen
2025-04-02 10:33       ` Michał Winiarski
2025-04-02 10:31     ` Michał Winiarski
2025-03-20 11:08 ` [PATCH v6 5/6] PCI: Allow drivers to control VF BAR size Michał Winiarski
2025-03-26 15:22   ` Ilpo Järvinen [this message]
2025-04-02 10:43     ` Michał Winiarski
2025-04-02 12:04       ` Ilpo Järvinen
2025-03-20 11:08 ` [PATCH v6 6/6] drm/xe/pf: Set VF LMEM " Michał Winiarski
2025-03-26 15:29   ` Ilpo Järvinen
2025-04-02 10:44     ` Michał Winiarski

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=7374beef-46ed-ab53-ccb5-48565526545c@linux.intel.com \
    --to=ilpo.jarvinen@linux.intel.com \
    --cc=airlied@gmail.com \
    --cc=bhelgaas@google.com \
    --cc=christian.koenig@amd.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=kw@linux.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=lucas.demarchi@intel.com \
    --cc=maarten.lankhorst@linux.intel.com \
    --cc=matthew.d.roper@intel.com \
    --cc=michal.wajdeczko@intel.com \
    --cc=michal.winiarski@intel.com \
    --cc=mripard@kernel.org \
    --cc=rodrigo.vivi@intel.com \
    --cc=simona@ffwll.ch \
    --cc=thomas.hellstrom@linux.intel.com \
    --cc=tzimmermann@suse.de \
    /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