| Message ID | 20210109095353.13417-1-ardb@kernel.org (mailing list archive) |
|---|---|
| State | Accepted |
| Delegated to: | Bjorn Helgaas |
| Headers | show |
| Series | PCI: decline to resize resources if boot config must be preserved | expand |
On Sat, Jan 09, 2021 at 10:53:53AM +0100, Ard Biesheuvel wrote: > The _DSM #5 method in the ACPI host bridge object tells us whether the > OS is permitted to deviate from the resource assignment configured by > the firmware. If this is not the case, we should not permit drivers to > resize BARs on the fly. So make pci_resize_resource() take this into > account. > > Cc: <stable@vger.kernel.org> # v5.4+ > Signed-off-by: Ard Biesheuvel <ardb@kernel.org> Capitalized subject to match convention and applied to pci/resource for v5.11, thanks! Is there an email, bug report, etc that prompted this change? > --- > drivers/pci/setup-res.c | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c > index 43eda101fcf4..3b38be081e93 100644 > --- a/drivers/pci/setup-res.c > +++ b/drivers/pci/setup-res.c > @@ -410,10 +410,16 @@ EXPORT_SYMBOL(pci_release_resource); > int pci_resize_resource(struct pci_dev *dev, int resno, int size) > { > struct resource *res = dev->resource + resno; > + struct pci_host_bridge *host; > int old, ret; > u32 sizes; > u16 cmd; > > + /* Check if we must preserve the firmware's resource assignment */ > + host = pci_find_host_bridge(dev->bus); > + if (host->preserve_config) > + return -ENOTSUPP; > + > /* Make sure the resource isn't assigned before resizing it. */ > if (!(res->flags & IORESOURCE_UNSET)) > return -EBUSY; > -- > 2.17.1 >
On Tue, 12 Jan 2021 at 23:49, Bjorn Helgaas <helgaas@kernel.org> wrote: > > On Sat, Jan 09, 2021 at 10:53:53AM +0100, Ard Biesheuvel wrote: > > The _DSM #5 method in the ACPI host bridge object tells us whether the > > OS is permitted to deviate from the resource assignment configured by > > the firmware. If this is not the case, we should not permit drivers to > > resize BARs on the fly. So make pci_resize_resource() take this into > > account. > > > > Cc: <stable@vger.kernel.org> # v5.4+ > > Signed-off-by: Ard Biesheuvel <ardb@kernel.org> > > Capitalized subject to match convention and applied to pci/resource > for v5.11, thanks! > > Is there an email, bug report, etc that prompted this change? > No, I was just reviewing the recent Tianocore changes to perform BAR resizing before resource assignment even takes places, which is obviously a more appropriate time to do it, as it does not require the OS to modify the firmware configuration at all. This reminded me of _DSM #5 and the fact that the OS may not even be permitted to make any changes. > > --- > > drivers/pci/setup-res.c | 6 ++++++ > > 1 file changed, 6 insertions(+) > > > > diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c > > index 43eda101fcf4..3b38be081e93 100644 > > --- a/drivers/pci/setup-res.c > > +++ b/drivers/pci/setup-res.c > > @@ -410,10 +410,16 @@ EXPORT_SYMBOL(pci_release_resource); > > int pci_resize_resource(struct pci_dev *dev, int resno, int size) > > { > > struct resource *res = dev->resource + resno; > > + struct pci_host_bridge *host; > > int old, ret; > > u32 sizes; > > u16 cmd; > > > > + /* Check if we must preserve the firmware's resource assignment */ > > + host = pci_find_host_bridge(dev->bus); > > + if (host->preserve_config) > > + return -ENOTSUPP; > > + > > /* Make sure the resource isn't assigned before resizing it. */ > > if (!(res->flags & IORESOURCE_UNSET)) > > return -EBUSY; > > -- > > 2.17.1 > >
diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c index 43eda101fcf4..3b38be081e93 100644 --- a/drivers/pci/setup-res.c +++ b/drivers/pci/setup-res.c @@ -410,10 +410,16 @@ EXPORT_SYMBOL(pci_release_resource); int pci_resize_resource(struct pci_dev *dev, int resno, int size) { struct resource *res = dev->resource + resno; + struct pci_host_bridge *host; int old, ret; u32 sizes; u16 cmd; + /* Check if we must preserve the firmware's resource assignment */ + host = pci_find_host_bridge(dev->bus); + if (host->preserve_config) + return -ENOTSUPP; + /* Make sure the resource isn't assigned before resizing it. */ if (!(res->flags & IORESOURCE_UNSET)) return -EBUSY;
The _DSM #5 method in the ACPI host bridge object tells us whether the OS is permitted to deviate from the resource assignment configured by the firmware. If this is not the case, we should not permit drivers to resize BARs on the fly. So make pci_resize_resource() take this into account. Cc: <stable@vger.kernel.org> # v5.4+ Signed-off-by: Ard Biesheuvel <ardb@kernel.org> --- drivers/pci/setup-res.c | 6 ++++++ 1 file changed, 6 insertions(+)