[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v5 09/10] xen/arm: Do not map PCI ECAM and MMIO space to Domain-0's p2m



Hi Oleksandr,

On 08/10/2021 06:55, Oleksandr Andrushchenko wrote:
From: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>

PCI host bridges are special devices in terms of implementing PCI
passthrough. According to [1] the current implementation depends on
Domain-0 to perform the initialization of the relevant PCI host
bridge hardware and perform PCI device enumeration. In order to
achieve that one of the required changes is to not map all the memory
ranges in map_range_to_domain as we traverse the device tree on startup
and perform some additional checks if the range needs to be mapped to
Domain-0.

The generic PCI host controller device tree binding says [2]:
- ranges: As described in IEEE Std 1275-1994, but must provide
           at least a definition of non-prefetchable memory. One
           or both of prefetchable Memory and IO Space may also
           be provided.

- reg   : The Configuration Space base address and size, as accessed
           from the parent bus.  The base address corresponds to
           the first bus in the "bus-range" property.  If no
           "bus-range" is specified, this will be bus 0 (the default).

 From the above none of the memory ranges from the "ranges" property

NIT: The space before "From" looks odd.

needs to be mapped to Domain-0 at startup as MMIO mapping is going to
be handled dynamically by vPCI as we assign PCI devices, e.g. each
device assigned to Domain-0/guest will have its MMIOs mapped/unmapped
as needed by Xen.

The "reg" property covers not only ECAM space, but may also have other
then the configuration memory ranges described, for example [3]:
- reg: Should contain rc_dbi, config registers location and length.
- reg-names: Must include the following entries:
    "rc_dbi": controller configuration registers;
    "config": PCIe configuration space registers.

This patch makes it possible to not map all the ranges from the
"ranges" property and also ECAM from the "reg". All the rest from the
"reg" property still needs to be mapped to Domain-0, so the PCI
host bridge remains functional in Domain-0.

[1] https://lists.xenproject.org/archives/html/xen-devel/2020-07/msg00777.html
[2] 
https://www.kernel.org/doc/Documentation/devicetree/bindings/pci/host-generic-pci.txt
[3] 
https://www.kernel.org/doc/Documentation/devicetree/bindings/pci/hisilicon-pcie.txt

Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>
---
Since v4:
- update skip_mapping comment
- add comment why we need to map interrupts to Dom0
Since v3:
  - pass struct map_range_data to map_dt_irq_to_domain
  - remove redundant check from map_range_to_domain
  - fix handle_device's .skip_mapping
Since v2:
  - removed check in map_range_to_domain for PCI_DEV
    and moved it to handle_device, so the code is
    simpler
  - s/map_pci_bridge/skip_mapping
  - extended comment in pci_host_bridge_mappings
  - minor code restructure in construct_dom0
  - s/.need_p2m_mapping/.need_p2m_hwdom_mapping and related
    callbacks
  - unsigned int i; in pci_host_bridge_mappings
Since v1:
  - Added better description of why and what needs to be mapped into
    Domain-0's p2m and what doesn't
  - Do not do any mappings for PCI devices while traversing the DT
  - Walk all the bridges and make required mappings in one go
---
  xen/arch/arm/domain_build.c        | 57 ++++++++++++++++++------------
  xen/arch/arm/pci/ecam.c            | 14 ++++++++
  xen/arch/arm/pci/pci-host-common.c | 49 +++++++++++++++++++++++++
  xen/arch/arm/pci/pci-host-zynqmp.c |  1 +
  xen/include/asm-arm/pci.h          | 10 ++++++
  xen/include/asm-arm/setup.h        | 13 +++++++
  6 files changed, 121 insertions(+), 23 deletions(-)

diff --git a/xen/arch/arm/domain_build.c b/xen/arch/arm/domain_build.c
index b51176b31bef..0d673b06a2f3 100644
--- a/xen/arch/arm/domain_build.c
+++ b/xen/arch/arm/domain_build.c
@@ -10,7 +10,6 @@
  #include <asm/regs.h>
  #include <xen/errno.h>
  #include <xen/err.h>
-#include <xen/device_tree.h>
  #include <xen/libfdt/libfdt.h>
  #include <xen/guest_access.h>
  #include <xen/iocap.h>
@@ -51,12 +50,6 @@ static int __init parse_dom0_mem(const char *s)
  }
  custom_param("dom0_mem", parse_dom0_mem);
-struct map_range_data
-{
-    struct domain *d;
-    p2m_type_t p2mt;
-};
-
  /* Override macros from asm/page.h to make them work with mfn_t */
  #undef virt_to_mfn
  #define virt_to_mfn(va) _mfn(__virt_to_mfn(va))
@@ -1663,10 +1656,11 @@ static int __init map_dt_irq_to_domain(const struct 
dt_device_node *dev,
                                         const struct dt_irq *dt_irq,
                                         void *data)
  {
-    struct domain *d = data;
+    struct map_range_data *mr_data = data;
+    struct domain *d = mr_data->d;
      unsigned int irq = dt_irq->irq;
      int res;
-    bool need_mapping = !dt_device_for_passthrough(dev);
+    bool need_mapping = !mr_data->skip_mapping;

Before, the variable 'need_mapping' was helpful to understand what !dt_device_for_passthrough(). Now it feels odd to read. There is only one user, so can the local variable be dropped?

if ( irq < NR_LOCAL_IRQS )
      {
@@ -1690,13 +1684,12 @@ static int __init map_dt_irq_to_domain(const struct 
dt_device_node *dev,
      return 0;
  }
-static int __init map_range_to_domain(const struct dt_device_node *dev,
-                                      u64 addr, u64 len,
-                                      void *data)
+int __init map_range_to_domain(const struct dt_device_node *dev,
+                               u64 addr, u64 len, void *data)
  {
      struct map_range_data *mr_data = data;
      struct domain *d = mr_data->d;
-    bool need_mapping = !dt_device_for_passthrough(dev);
+    bool need_mapping = !mr_data->skip_mapping;

Same here.

      int res;
/*
@@ -1748,23 +1741,21 @@ static int __init map_range_to_domain(const struct 
dt_device_node *dev,
   * then we may need to perform additional mappings in order to make
   * the child resources available to domain 0.
   */
-static int __init map_device_children(struct domain *d,
-                                      const struct dt_device_node *dev,
-                                      p2m_type_t p2mt)
+static int __init map_device_children(const struct dt_device_node *dev,
+                                      struct map_range_data *mr_data)
  {
-    struct map_range_data mr_data = { .d = d, .p2mt = p2mt };
-    int ret;
-
      if ( dt_device_type_is_equal(dev, "pci") )
      {
+        int ret;
+
          dt_dprintk("Mapping children of %s to guest\n",
                     dt_node_full_name(dev));
- ret = dt_for_each_irq_map(dev, &map_dt_irq_to_domain, d);
+        ret = dt_for_each_irq_map(dev, &map_dt_irq_to_domain, mr_data);
          if ( ret < 0 )
              return ret;
- ret = dt_for_each_range(dev, &map_range_to_domain, &mr_data);
+        ret = dt_for_each_range(dev, &map_range_to_domain, mr_data);
          if ( ret < 0 )
              return ret;
      }
@@ -1845,6 +1836,20 @@ static int __init handle_device(struct domain *d, struct 
dt_device_node *dev,
      int res;
      u64 addr, size;
      bool need_mapping = !dt_device_for_passthrough(dev);

I find the difference between .skip_mapping and the local variable need_mapping quite puzzling. We are likely going to misuse them in the future.

I think it would be clearer if the local variable is renamed to 'own_device' or similar.

+    /*
+     * For PCI passthrough we only need to remap to Dom0 the interrupts
+     * and memory ranges from "reg" property which cover controller's
+     * configuration registers and such. PCIe configuration space registers
+     * of the PCIe Root Complex and PCIe aperture should not be mapped
+     * automatically to Dom0.
+     */
+    struct map_range_data mr_data = {
+        .d = d,
+        .p2mt = p2mt,
+        .skip_mapping = !need_mapping ||
+                        (is_pci_passthrough_enabled() &&
+                         (device_get_class(dev) == DEVICE_PCI))

The device class is confusing. When I see DEVICE_PCI, I think this is a PCI device whereas here you are referring to the hostbridge.

Unfortunate, I wasn't able to comment on the original patch before it was committed. But I would like this to be renamed to DEVICE_PCI_HOSTBRIDGE for Xen 4.16. Can you send a patch?

+    };
naddr = dt_number_of_address(dev); @@ -1884,7 +1889,6 @@ static int __init handle_device(struct domain *d, struct dt_device_node *dev,
      /* Give permission and map MMIOs */
      for ( i = 0; i < naddr; i++ )
      {
-        struct map_range_data mr_data = { .d = d, .p2mt = p2mt };
          res = dt_device_get_address(dev, i, &addr, &size);
          if ( res )
          {
@@ -1898,7 +1902,7 @@ static int __init handle_device(struct domain *d, struct 
dt_device_node *dev,
              return res;
      }
- res = map_device_children(d, dev, p2mt);
+    res = map_device_children(dev, &mr_data);
      if ( res )
          return res;
@@ -3056,7 +3060,14 @@ static int __init construct_dom0(struct domain *d)
          return rc;
if ( acpi_disabled )
+    {
          rc = prepare_dtb_hwdom(d, &kinfo);
+        if ( rc < 0 )
+            return rc;
+#ifdef CONFIG_HAS_PCI
+        rc = pci_host_bridge_mappings(d, p2m_mmio_direct_c);

It is not clear to me why you are passing p2m_mmio_direct_c and not p2mt here? If you really want to force a type, then I think it should be p2m_mmio_direct.

But then why is it a parameter of pci_host_bridge_mappings? Do you expect someone else to modify it?

+#endif
+    }
      else
          rc = prepare_acpi(d, &kinfo);
diff --git a/xen/arch/arm/pci/ecam.c b/xen/arch/arm/pci/ecam.c
index 602d00799c8d..b81620074a91 100644
--- a/xen/arch/arm/pci/ecam.c
+++ b/xen/arch/arm/pci/ecam.c
@@ -40,6 +40,19 @@ void __iomem *pci_ecam_map_bus(struct pci_host_bridge 
*bridge,
      return base + (PCI_DEVFN2(sbdf.bdf) << devfn_shift) + where;
  }
+bool pci_ecam_need_p2m_hwdom_mapping(struct domain *d,

AFAICT, this is only used by boot code. So you want to do add __init.

This is also exported but not used. I would prefer if we only exposed when the first external user will be introduced.

+                                     struct pci_host_bridge *bridge,
+                                     uint64_t addr)
+{
+    struct pci_config_window *cfg = bridge->cfg;
+
+    /*
+     * We do not want ECAM address space to be mapped in Domain-0's p2m,
+     * so we can trap access to it.
+     */
+    return cfg->phys_addr != addr;
+}
+
  /* ECAM ops */
  const struct pci_ecam_ops pci_generic_ecam_ops = {
      .bus_shift  = 20,
@@ -47,6 +60,7 @@ const struct pci_ecam_ops pci_generic_ecam_ops = {
          .map_bus                = pci_ecam_map_bus,
          .read                   = pci_generic_config_read,
          .write                  = pci_generic_config_write,
+        .need_p2m_hwdom_mapping = pci_ecam_need_p2m_hwdom_mapping,
      }
  };
diff --git a/xen/arch/arm/pci/pci-host-common.c b/xen/arch/arm/pci/pci-host-common.c
index 1eb4daa87365..085f08e23e0c 100644
--- a/xen/arch/arm/pci/pci-host-common.c
+++ b/xen/arch/arm/pci/pci-host-common.c
@@ -18,6 +18,7 @@
#include <xen/init.h>
  #include <xen/pci.h>
+#include <asm/setup.h>

For new code, we usually include xen/*.h first and then asm/*.h. They are then order alphabetically within themselves.

  #include <xen/rwlock.h>
  #include <xen/sched.h>
  #include <xen/vmap.h>
@@ -320,6 +321,54 @@ int pci_host_get_num_bridges(void)
      return count;
  }
+int __init pci_host_bridge_mappings(struct domain *d, p2m_type_t p2mt)
+{
+    struct pci_host_bridge *bridge;
+    struct map_range_data mr_data = {
+        .d = d,
+        .p2mt = p2mt,
+        .skip_mapping = false
+    };
+
+    /*
+     * For each PCI host bridge we need to only map those ranges
+     * which are used by Domain-0 to properly initialize the bridge,
+     * e.g. we do not want to map ECAM configuration space which lives in
+     * "reg" or "assigned-addresses" device tree property, but we want to

AFAIU, "assigned-addresses" is only relevant for the child of nodes with type="pci". This is not the case for the hostbridges. So I think you want to drop the mention of "assigned-addresses".


+     * map other regions of the host bridge. The PCI aperture defined by
+     * the "ranges" device tree property should also be skipped.
+     */
+    list_for_each_entry( bridge, &pci_host_bridges, node )
+    {
+        const struct dt_device_node *dev = bridge->dt_node;
+        unsigned int i;
+
+        for ( i = 0; i < dt_number_of_address(dev); i++ )
+        {
+            uint64_t addr, size;
+            int err;
+
+            err = dt_device_get_address(dev, i, &addr, &size);
+            if ( err )
+            {
+                printk(XENLOG_ERR
+                       "Unable to retrieve address range index=%u for %s\n",
+                       i, dt_node_full_name(dev));
+                return err;
+            }
+
+            if ( bridge->ops->need_p2m_hwdom_mapping(d, bridge, addr) )
+            {
+                err = map_range_to_domain(dev, addr, size, &mr_data);
+                if ( err )
+                    return err;
+            }
+        }
+    }
+
+    return 0;
+}
+
  /*
   * Local variables:
   * mode: C
diff --git a/xen/arch/arm/pci/pci-host-zynqmp.c 
b/xen/arch/arm/pci/pci-host-zynqmp.c
index 61a9807d3d58..6ad2b31e810d 100644
--- a/xen/arch/arm/pci/pci-host-zynqmp.c
+++ b/xen/arch/arm/pci/pci-host-zynqmp.c
@@ -34,6 +34,7 @@ const struct pci_ecam_ops nwl_pcie_ops = {
          .map_bus                = pci_ecam_map_bus,
          .read                   = pci_generic_config_read,
          .write                  = pci_generic_config_write,
+        .need_p2m_hwdom_mapping = pci_ecam_need_p2m_hwdom_mapping,
      }
  };
diff --git a/xen/include/asm-arm/pci.h b/xen/include/asm-arm/pci.h
index a71b0eea8cb9..b5b85ccd0937 100644
--- a/xen/include/asm-arm/pci.h
+++ b/xen/include/asm-arm/pci.h
@@ -17,6 +17,8 @@
#ifdef CONFIG_HAS_PCI +#include <asm/p2m.h>
+
  #define pci_to_dev(pcidev) (&(pcidev)->arch.dev)
extern bool pci_passthrough_enabled;
@@ -73,6 +75,9 @@ struct pci_ops {
                  uint32_t reg, uint32_t len, uint32_t *value);
      int (*write)(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
                   uint32_t reg, uint32_t len, uint32_t value);
+    bool (*need_p2m_hwdom_mapping)(struct domain *d,
+                                   struct pci_host_bridge *bridge,
+                                   uint64_t addr);
  };
/*
@@ -96,6 +101,9 @@ int pci_generic_config_write(struct pci_host_bridge *bridge, 
pci_sbdf_t sbdf,
                               uint32_t reg, uint32_t len, uint32_t value);
  void __iomem *pci_ecam_map_bus(struct pci_host_bridge *bridge,
                                 pci_sbdf_t sbdf, uint32_t where);
+bool pci_ecam_need_p2m_hwdom_mapping(struct domain *d,
+                                     struct pci_host_bridge *bridge,
+                                     uint64_t addr);
  struct pci_host_bridge *pci_find_host_bridge(uint16_t segment, uint8_t bus);
  int pci_get_host_bridge_segment(const struct dt_device_node *node,
                                  uint16_t *segment);
@@ -113,6 +121,8 @@ int pci_host_iterate_bridges(struct domain *d,
                                          struct pci_host_bridge *bridge));
  int pci_host_get_num_bridges(void);
+int pci_host_bridge_mappings(struct domain *d, p2m_type_t p2mt);
+
  #else   /*!CONFIG_HAS_PCI*/
struct arch_pci_dev { };
diff --git a/xen/include/asm-arm/setup.h b/xen/include/asm-arm/setup.h
index 95da0b7ab9cd..88d9673db817 100644
--- a/xen/include/asm-arm/setup.h
+++ b/xen/include/asm-arm/setup.h
@@ -2,6 +2,8 @@
  #define __ARM_SETUP_H_
#include <public/version.h>
+#include <asm/p2m.h>
+#include <xen/device_tree.h>
#define MIN_FDT_ALIGN 8
  #define MAX_FDT_SIZE SZ_2M
@@ -77,6 +79,14 @@ struct bootinfo {
  #endif
  };
+struct map_range_data
+{
+    struct domain *d;
+    p2m_type_t p2mt;
+    /* Set if mapping of the memory ranges must be skipped. */
+    bool skip_mapping;
+};
+
  extern struct bootinfo bootinfo;
extern domid_t max_init_domid;
@@ -124,6 +134,9 @@ void device_tree_get_reg(const __be32 **cell, u32 
address_cells,
  u32 device_tree_get_u32(const void *fdt, int node,
                          const char *prop_name, u32 dflt);
+int map_range_to_domain(const struct dt_device_node *dev,
+                        u64 addr, u64 len, void *data);
+
  #endif
  /*
   * Local variables:


--
Julien Grall



 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.