Re: [PATCH -v2] PCI: Clear all bridge res MEM_64 if host bridge has non mem64

From: Bjorn Helgaas
Date: Fri Dec 19 2014 - 17:45:27 EST


On Wed, Dec 10, 2014 at 04:19:42PM -0800, Yinghai Lu wrote:
> So we could use bridge 64bit mem pref for children mem pref instead of
> forcing them into bridge mem.
>
> Could help Marek's system as his system is using _CRS, and all mem res is under
> 4G.
>
> -v2: fix checking logic problem found by Gravin and Wei.
>
> Link: https://bugzilla.kernel.org/show_bug.cgi?id=85491
> Reported-by: Marek Kordik <kordikmarek@xxxxxxxxx>
> Fixes: 5b28541552ef ("PCI: Restrict 64-bit prefetchable bridge windows to 64-bit resources")
> Signed-off-by: Yinghai Lu <yinghai@xxxxxxxxxx>
>
> ---
> drivers/pci/host-bridge.c | 7 +++++++
> drivers/pci/pci.h | 1 +
> drivers/pci/probe.c | 9 +++++++++
> drivers/pci/setup-bus.c | 3 +++
> include/linux/pci.h | 1 +
> 5 files changed, 21 insertions(+)
>
> Index: linux-2.6/drivers/pci/host-bridge.c
> ===================================================================
> --- linux-2.6.orig/drivers/pci/host-bridge.c
> +++ linux-2.6/drivers/pci/host-bridge.c
> @@ -31,6 +31,13 @@ void pci_set_host_bridge_release(struct
> bridge->release_data = release_data;
> }
>
> +bool pcibios_host_bridge_has_mem64_res(struct pci_bus *bus)
> +{
> + struct pci_host_bridge *bridge = find_pci_host_bridge(bus);
> +
> + return bridge->has_mem64_res;
> +}
> +
> void pcibios_resource_to_bus(struct pci_bus *bus, struct pci_bus_region *region,
> struct resource *res)
> {
> Index: linux-2.6/drivers/pci/pci.h
> ===================================================================
> --- linux-2.6.orig/drivers/pci/pci.h
> +++ linux-2.6/drivers/pci/pci.h
> @@ -196,6 +196,7 @@ enum pci_bar_type {
> pci_bar_mem64, /* A 64-bit memory BAR */
> };
>
> +bool pcibios_host_bridge_has_mem64_res(struct pci_bus *bus);
> bool pci_bus_read_dev_vendor_id(struct pci_bus *bus, int devfn, u32 *pl,
> int crs_timeout);
> int pci_setup_device(struct pci_dev *dev);
> Index: linux-2.6/drivers/pci/probe.c
> ===================================================================
> --- linux-2.6.orig/drivers/pci/probe.c
> +++ linux-2.6/drivers/pci/probe.c
> @@ -1980,6 +1980,15 @@ struct pci_bus *pci_create_root_bus(stru
> dev_info(&b->dev, "root bus resource %pR%s\n", res, bus_addr);
> }
>
> + list_for_each_entry(window, &bridge->windows, list) {
> + res = window->res;
> + if (resource_type(res) == IORESOURCE_MEM &&
> + (res->end - window->offset) > 0xffffffff) {
> + bridge->has_mem64_res = true;
> + break;
> + }
> + }
> +
> down_write(&pci_bus_sem);
> list_add_tail(&b->node, &pci_root_buses);
> up_write(&pci_bus_sem);
> Index: linux-2.6/include/linux/pci.h
> ===================================================================
> --- linux-2.6.orig/include/linux/pci.h
> +++ linux-2.6/include/linux/pci.h
> @@ -407,6 +407,7 @@ struct pci_host_bridge {
> struct list_head windows; /* pci_host_bridge_windows */
> void (*release_fn)(struct pci_host_bridge *);
> void *release_data;
> + bool has_mem64_res;
> };
>
> #define to_pci_host_bridge(n) container_of(n, struct pci_host_bridge, dev)
> Index: linux-2.6/drivers/pci/setup-bus.c
> ===================================================================
> --- linux-2.6.orig/drivers/pci/setup-bus.c
> +++ linux-2.6/drivers/pci/setup-bus.c
> @@ -693,6 +693,9 @@ static void pci_bridge_check_ranges(stru
> }
> }
>
> + if (!pcibios_host_bridge_has_mem64_res(bus))
> + b_res[2].flags &= ~IORESOURCE_MEM_64;

I don't want to clear the MEM_64 bit in the resource flags. I want those
flags to reflect the capability of the hardware. The bridge has a 64-bit
window, so I think the resource should have the MEM_64 bit set. I think we
should set that bit when we first enumerate the bridge and read its window
information, and it should never be changed after that.

It's too confusing if we use the resource to store information about
upstream things like host bridge windows or downstream things like the BARs
of devices below the bridge.

> +
> /* double check if bridge does support 64 bit pref */
> if (b_res[2].flags & IORESOURCE_MEM_64) {
> u32 mem_base_hi, tmp;
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/