linux-kernel - Re: [PATCH v2 2/3] PCI: endpoint: Add API pci_epf_set_inbound

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <bmz36usaey7skutdukgryq22sgvyidfnmtx6z4zrpcvrgpvcdh@4l47wob6grkn>
Date: Wed, 24 Sep 2025 23:39:56 +0530
From: Manivannan Sadhasivam <mani@...nel.org>
To: Frank Li <Frank.Li@....com>
Cc: Krzysztof Wilczyński <kwilczynski@...nel.org>, 
	Kishon Vijay Abraham I <kishon@...nel.org>, Bjorn Helgaas <bhelgaas@...gle.com>, 
	Jon Mason <jdmason@...zu.us>, Dave Jiang <dave.jiang@...el.com>, 
	Allen Hubbe <allenbh@...il.com>, linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org, 
	ntb@...ts.linux.dev, imx@...ts.linux.dev
Subject: Re: [PATCH v2 2/3] PCI: endpoint: Add API pci_epf_set_inbound_space()

On Mon, Sep 15, 2025 at 06:22:45PM -0400, Frank Li wrote:
> Add pci_epf_set_inbound_space() to allow setting any physical address as
> inbound memory space, such as an MSI message base address.
> 
> Since PCI BAR size must be a power of two, the input MMIO range
> [inbound_addr, inbound_addr + size) is mapped by finding n such that
> [base, base + 2^n) covers the range.

> The base address is also required
> to be aligned to 2^n.
>

Where does this alignment restriction gets checked?
 
> Signed-off-by: Frank Li <Frank.Li@....com>
> ---
> change in v2
> - add new API pci_epf_set_inbound_space()
> - fix bits 8 * size_of(dma_addr_t);
> ---
>  drivers/pci/endpoint/pci-epf-core.c | 86 +++++++++++++++++++++++++++++++++++++
>  include/linux/pci-epf.h             |  6 +++
>  2 files changed, 92 insertions(+)
> 
> diff --git a/drivers/pci/endpoint/pci-epf-core.c b/drivers/pci/endpoint/pci-epf-core.c
> index 4281067d4a62da6fbf6c2fb807b0f1b5afd1f45b..cd10e8619d03c7e2ffe48e437b0aecf0e8a499f4 100644
> --- a/drivers/pci/endpoint/pci-epf-core.c
> +++ b/drivers/pci/endpoint/pci-epf-core.c
> @@ -344,6 +344,92 @@ void *pci_epf_alloc_space(struct pci_epf *epf, size_t size, enum pci_barno bar,
>  }
>  EXPORT_SYMBOL_GPL(pci_epf_alloc_space);
>  
> +/**
> + * pci_epf_set_inbound_space() - set MMIO for the PCI EPF register space

This is just assigning the memory for @bar in epf_bar. So, how about,
pci_epf_assign_bar_space()?

'Assign PCI EPF BAR space'

> + * @epf: the EPF device to whom allocate the memory
> + * @size: the size of the memory that has to be allocated

s/allocated/assigned since this API is not allocating the space.

> + * @bar: the BAR number corresponding to the allocated register space
> + * @epc_features: the features provided by the EPC specific to this EPF
> + * @type: Identifies if the allocation is for primary EPC or secondary EPC
> + * @inbound_addr: Any physical address space such as MSI message address that

s/inbound_addr/bar_addr

"Address to be assigned for the @bar"

> + *		work as inbound address space. from_memory need be false.
> + *
> + * Invoke to allocate memory for the PCI EPF register space.

s/allocate/assign

> + * Flag PCI_BASE_ADDRESS_MEM_TYPE_64 will automatically get set if the BAR
> + * can only be a 64-bit BAR, or if the requested size is larger than 2 GB.
> + */
> +int pci_epf_set_inbound_space(struct pci_epf *epf, size_t size,
> +			      enum pci_barno bar,
> +			      const struct pci_epc_features *epc_features,
> +			      enum pci_epc_interface_type type,
> +			      dma_addr_t inbound_addr)
> +{
> +	size_t aligned_size, align = epc_features->align;
> +	struct pci_epf_bar *epf_bar;
> +	struct pci_epc *epc;
> +	dma_addr_t up;
> +	int pos;
> +
> +	if (!size)
> +		return -EINVAL;
> +
> +	up = inbound_addr + size - 1;

s/up/limit?

> +
> +	/*
> +	 *  Bits:		15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
> +	 *  Inbound_addr:	U  U  U  U  U  U  0 X X X X X X X X X
> +	 *  Up:			U  U  U  U  U  U  1 X X X X X X X X X
> +	 *
> +	 *  U means address bits have not change in Range [Inbound_Addr, Up]
> +	 *  X means bit 0 or 1.
> +	 *
> +	 *  Inbound_addr^Up  0  0  0  0  0  0  1 X X X X X X X X X
> +	 *  Find first bit 1 pos from MSB, 2 ^ pos windows will cover
> +	 *  [Inbound_Addr, Up] range.
> +	 */
> +	for (pos = 8 * sizeof(dma_addr_t) - 1; pos > 0; pos--)
> +		if ((up ^ inbound_addr) & BIT_ULL(pos))
> +			break;
> +
> +	if (pos == 8 * sizeof(dma_addr_t) - 1)
> +		return -EINVAL;
> +
> +	size = pci_epf_get_bar_required_size(epf, BIT_ULL(pos + 1), bar,
> +					     epc_features, type);
> +
> +	if (size == 0)
> +		return -EINVAL;
> +
> +	/*
> +	 * Allocate enough memory to accommodate the iATU alignment

s/iATU/EPC

> +	 * requirement.  In most cases, this will be the same as .size but
> +	 * it might be different if, for example, the fixed size of a BAR
> +	 * is smaller than align.
> +	 */
> +	aligned_size = align ? ALIGN(size, align) : size;

This should be handled inside the above API.

- Mani

-- 
மணிவண்ணன் சதாசிவம்