linux-kernel - Re: [PATCH v3 1/3] PCI: endpoint: Add helper function pci_epf_get_bar_required

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <aNauaSVs5ytzsVFt@ryzen>
Date: Fri, 26 Sep 2025 17:16:57 +0200
From: Niklas Cassel <cassel@...nel.org>
To: Frank Li <Frank.li@....com>
Cc: Manivannan Sadhasivam <mani@...nel.org>,
	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 v3 1/3] PCI: endpoint: Add helper function
 pci_epf_get_bar_required_size()

On Fri, Sep 26, 2025 at 10:56:46AM -0400, Frank Li wrote:
> On Fri, Sep 26, 2025 at 02:31:42PM +0200, Niklas Cassel wrote:
> > On Thu, Sep 25, 2025 at 01:01:47PM -0400, Frank Li wrote:
> > > Introduce pci_epf_get_bar_required_size() to retrieve the required BAR
> > > size and memory size. Prepare for adding support to set an MMIO address to
> > > a specific BAR.
> > >
> > > Use two variables 'aligned_bar_size' and 'aligned_mem_size' to avoid
> > > confuse.
> >
> > s/confuse/confusion/
> >
> >
> > >
> > > No functional changes.
> > >
> > > Signed-off-by: Frank Li <Frank.Li@....com>
> > > ---
> > > change in v3
> > > - change return value to int.
> > > - use two pointers return bar size aligned and memory start address aligned
> > > - update comments about why need memory align size. Actually iATU require
> > > start address match aligned requirement. Since kernel return align to
> > > size's address.
> > > - use two varible aligned_bar_size and aligned_mem_size to avoid confuse
> > > use 'size'.
> > >
> > > change in v2
> > > - new patch
> > > ---
> > >  drivers/pci/endpoint/pci-epf-core.c | 84 +++++++++++++++++++++++--------------
> > >  1 file changed, 53 insertions(+), 31 deletions(-)
> > >
> > > diff --git a/drivers/pci/endpoint/pci-epf-core.c b/drivers/pci/endpoint/pci-epf-core.c
> > > index d54e18872aefc07c655c94c104a347328ff7a432..2cd0257831f9885a4381c087ed8f3326f5960966 100644
> > > --- a/drivers/pci/endpoint/pci-epf-core.c
> > > +++ b/drivers/pci/endpoint/pci-epf-core.c
> > > @@ -208,6 +208,49 @@ void pci_epf_remove_vepf(struct pci_epf *epf_pf, struct pci_epf *epf_vf)
> > >  }
> > >  EXPORT_SYMBOL_GPL(pci_epf_remove_vepf);
> > >
> > > +static int
> > > +pci_epf_get_bar_required_size(struct pci_epf *epf, size_t size,
> > > +			      size_t *aligned_bar_size,
> > > +			      size_t *aligned_mem_size,
> > > +			      enum pci_barno bar,
> > > +			      const struct pci_epc_features *epc_features,
> > > +			      enum pci_epc_interface_type type)
> > > +{
> > > +	u64 bar_fixed_size = epc_features->bar[bar].fixed_size;
> > > +	size_t align = epc_features->align;
> > > +
> > > +	if (size < 128)
> > > +		size = 128;
> > > +
> > > +	/* According to PCIe base spec, min size for a resizable BAR is 1 MB. */
> > > +	if (epc_features->bar[bar].type == BAR_RESIZABLE && size < SZ_1M)
> > > +		size = SZ_1M;
> > > +
> > > +	if (epc_features->bar[bar].type == BAR_FIXED && bar_fixed_size) {
> > > +		if (size > bar_fixed_size) {
> > > +			dev_err(&epf->dev,
> > > +				"requested BAR size is larger than fixed size\n");
> > > +			return -ENOMEM;
> > > +		}
> > > +		size = bar_fixed_size;
> > > +	} else {
> > > +		/* BAR size must be power of two */
> > > +		size = roundup_pow_of_two(size);
> > > +	}
> > > +
> > > +	*aligned_bar_size = size;
> >
> > I think this name is wrong.
> > The BAR size has not been aligned to anything.
> > The BAR size has to be a power of two, but that is a requirement of the PCI
> > specification, so that in an inherent property of a BAR.
> >
> > Perhaps just name it size or bar_size?
> 
> there already have 'size' for input.  It should match epc required's size.

Why do you need both "size_t size" and "size_t *bar_size"?

Isn't it enough with "size_t *bar_size" ?

The user can supply a value, and the function could update that value.


Kind regards,
Niklas