linux-kernel - Re: [PATCH v3 3/4] PCI/ACPI: Add Broadcom bcm2711 MCFG quirk

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: <20211005223148.GA1125087@bhelgaas>
Date:   Tue, 5 Oct 2021 17:31:48 -0500
From:   Bjorn Helgaas <helgaas@...nel.org>
To:     Jeremy Linton <jeremy.linton@....com>
Cc:     linux-pci@...r.kernel.org, lorenzo.pieralisi@....com,
        nsaenz@...nel.org, bhelgaas@...gle.com, rjw@...ysocki.net,
        lenb@...nel.org, robh@...nel.org, kw@...ux.com,
        f.fainelli@...il.com, bcm-kernel-feedback-list@...adcom.com,
        linux-acpi@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        linux-rpi-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3 3/4] PCI/ACPI: Add Broadcom bcm2711 MCFG quirk

On Tue, Oct 05, 2021 at 10:43:32AM -0500, Jeremy Linton wrote:
> On 10/5/21 10:10 AM, Bjorn Helgaas wrote:
> > On Thu, Aug 26, 2021 at 02:15:56AM -0500, Jeremy Linton wrote:
> > > Now that there is a bcm2711 quirk, it needs to be enabled when the
> > > MCFG is missing. Use an ACPI namespace _DSD property
> > > "linux-ecam-quirk-id" as an alternative to the MCFG OEM.
> > > 
> > > Signed-off-by: Jeremy Linton <jeremy.linton@....com>
> > > Acked-by: Florian Fainelli <f.fainelli@...il.com>
> > > Acked-by: Bjorn Helgaas <bhelgaas@...gle.com>
> > > ---
> > >   drivers/acpi/pci_mcfg.c | 17 +++++++++++++++++
> > >   1 file changed, 17 insertions(+)
> > > 
> > > diff --git a/drivers/acpi/pci_mcfg.c b/drivers/acpi/pci_mcfg.c
> > > index 53cab975f612..04c517418365 100644
> > > --- a/drivers/acpi/pci_mcfg.c
> > > +++ b/drivers/acpi/pci_mcfg.c
> > > @@ -169,6 +169,9 @@ static struct mcfg_fixup mcfg_quirks[] = {
> > >   	ALTRA_ECAM_QUIRK(1, 13),
> > >   	ALTRA_ECAM_QUIRK(1, 14),
> > >   	ALTRA_ECAM_QUIRK(1, 15),
> > > +
> > > +	{ "bc2711", "", 0, 0, MCFG_BUS_ANY, &bcm2711_pcie_ops,
> > > +	  DEFINE_RES_MEM(0xFD500000, 0xA000) },
> > >   };
> > >   static char mcfg_oem_id[ACPI_OEM_ID_SIZE];
> > > @@ -198,8 +201,22 @@ static void pci_mcfg_apply_quirks(struct acpi_pci_root *root,
> > >   	u16 segment = root->segment;
> > >   	struct resource *bus_range = &root->secondary;
> > >   	struct mcfg_fixup *f;
> > > +	const char *soc;
> > >   	int i;
> > > +	/*
> > > +	 * This may be a machine with a PCI/SMC conduit, which means it doesn't
> > > +	 * have an MCFG. Use an ACPI namespace definition instead.
> > > +	 */
> > > +	if (!fwnode_property_read_string(acpi_fwnode_handle(root->device),
> > > +					 "linux-ecam-quirk-id", &soc)) {
> > > +		if (strlen(soc) != ACPI_OEM_ID_SIZE)
> > 
> > From a reviewing perspective, it's not obvious why soc must be exactly
> > ACPI_OEM_ID_SIZE.  Does that imply space-padding in the DT string or
> > something?
> 
> This is at the moment an ACPI only DSD, and it must follow the MADT OEM_ID
> format for now because we are effectively just overriding that field. The
> rest of the code in this module is just treating it as a fixed 6 bytes.
> 
> > Is there any documentation for this DT property?
> 
> Its not a DT property, and its unclear since its linux only if it
> belongs in previously linked ACPI registry.

Oh, right, it comes from a _DSD.

> > Also not obvious why strlen() is safe here.  I mean, I looked a couple
> > levels deep in fwnode_property_read_string(), but whatever guarantees
> > null termination is buried pretty deep.
> 
> I've not tracked down who, if anyone other than the AML compiler is
> guaranteeing a null. The spec says something to the effect "Most other
> string, however,  are of variable-length and are automatically null
> terminated by the compiler". Not sure if that helps any.

Doesn't help for me.  The PCI core shouldn't go in the weeds no matter
what junk we might get from an AML compiler.  Maybe
fwnode_property_read_string() guarantees null termination, but it's
not documented and not easy to verify.

I think a strncpy() here might be better.  Not sure it's worthwhile to
emit a specific error message for the wrong length.

> > It seems a little weird to use an MCFG quirk mechanism when there's no
> > MCFG at all on this platform.
> 
> Well its just a point to hook in a CFG space quirk, and since that
> is what most of the MCFG quirks are, it seemed reasonable to reuse
> it rather than recreate it.

Yeah, it's ugly no matter how we slice it.  The pci_no_msi()
especially has nothing to do with ECAM at all.  But I don't know how
to identify this thing for a quirk.  PNP0A08 devices really rely on
ECAM or a system firmware config accessor.

> > > +			dev_err(&root->device->dev, "ECAM quirk should be %d characters\n",
> > > +				ACPI_OEM_ID_SIZE);
> > > +		else
> > > +			memcpy(mcfg_oem_id, soc, ACPI_OEM_ID_SIZE);
> > > +	}
> > > +
> > >   	for (i = 0, f = mcfg_quirks; i < ARRAY_SIZE(mcfg_quirks); i++, f++) {
> > >   		if (pci_mcfg_quirk_matches(f, segment, bus_range)) {
> > >   			if (f->cfgres.start)
> > > -- 
> > > 2.31.1
> > > 
>