* [Qemu-devel] [PATCH v2 01/23] Add memory API documentation
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 02/23] Hierarchical memory region API Avi Kivity
` (22 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Signed-off-by: Avi Kivity <avi@redhat.com>
---
docs/memory.txt | 172 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 172 insertions(+), 0 deletions(-)
create mode 100644 docs/memory.txt
diff --git a/docs/memory.txt b/docs/memory.txt
new file mode 100644
index 0000000..4460c06
--- /dev/null
+++ b/docs/memory.txt
@@ -0,0 +1,172 @@
+The memory API
+==============
+
+The memory API models the memory and I/O buses and controllers of a QEMU
+machine. It attempts to allow modelling of:
+
+ - ordinary RAM
+ - memory-mapped I/O (MMIO)
+ - memory controllers that can dynamically reroute physical memory regions
+ to different destinations
+
+The memory model provides support for
+
+ - tracking RAM changes by the guest
+ - setting up coalesced memory for kvm
+ - setting up ioeventfd regions for kvm
+
+Memory is modelled as an tree (really acyclic graph) of MemoryRegion objects.
+The root of the tree is memory as seen from the CPU's viewpoint (the system
+bus). Nodes in the tree represent other buses, memory controllers, and
+memory regions that have been rerouted. Leaves are RAM and MMIO regions.
+
+Types of regions
+----------------
+
+There are four types of memory regions (all represented by a single C type
+MemoryRegion):
+
+- RAM: a RAM region is simply a range of host memory that can be made available
+ to the guest.
+
+- MMIO: a range of guest memory that is implemented by host callbacks;
+ each read or write causes a callback to be called on the host.
+
+- container: a container simply includes other memory regions, each at
+ a different offset. Containers are useful for grouping several regions
+ into one unit. For example, a PCI BAR may be composed of a RAM region
+ and an MMIO region.
+
+ A container's subregions are usually non-overlapping. In some cases it is
+ useful to have overlapping regions; for example a memory controller that
+ can overlay a subregion of RAM with MMIO or ROM, or a PCI controller
+ that does not prevent card from claiming overlapping BARs.
+
+- alias: a subsection of another region. Aliases allow a region to be
+ split apart into discontiguous regions. Examples of uses are memory banks
+ used when the guest address space is smaller than the amount of RAM
+ addressed, or a memory controller that splits main memory to expose a "PCI
+ hole". Aliases may point to any type of region, including other aliases,
+ but an alias may not point back to itself, directly or indirectly.
+
+
+Region names
+------------
+
+Regions are assigned names by the constructor. For most regions these are
+only used for debugging purposes, but RAM regions also use the name to identify
+live migration sections. This means that RAM region names need to have ABI
+stability.
+
+Region lifecycle
+----------------
+
+A region is created by one of the constructor functions (memory_region_init*())
+and destroyed by the destructor (memory_region_destroy()). In between,
+a region can be added to an address space by using memory_region_add_subregion()
+and removed using memory_region_del_subregion(). Region attributes may be
+changed at any point; they take effect once the region becomes exposed to the
+guest.
+
+Overlapping regions and priority
+--------------------------------
+Usually, regions may not overlap each other; a memory address decodes into
+exactly one target. In some cases it is useful to allow regions to overlap,
+and sometimes to control which of an overlapping regions is visible to the
+guest. This is done with memory_region_add_subregion_overlap(), which
+allows the region to overlap any other region in the same container, and
+specifies a priority that allows the core to decide which of two regions at
+the same address are visible (highest wins).
+
+Visibility
+----------
+The memory core uses the following rules to select a memory region when the
+guest accesses an address:
+
+- all direct subregions of the root region are matched against the address, in
+ descending priority order
+ - if the address lies outside the region offset/size, the subregion is
+ discarded
+ - if the subregion is a leaf (RAM or MMIO), the seach terminates
+ - if the subregion is a container, the same algorithm is used within the
+ subregion (after the address is adjusted by the subregion offset)
+ - if the subregion is an alias, the search is continues at the alias target
+ (after the address is adjusted by the subregion offset and alias offset)
+
+Example memory map
+------------------
+
+system_memory: container@0-2^48-1
+ |
+ +---- lomem: alias@0-0xdfffffff ---> #ram (0-0xdfffffff)
+ |
+ +---- himem: alias@0x100000000-0x11fffffff ---> #ram (0xe0000000-0xffffffff)
+ |
+ +---- vga-window: alias@0xa0000-0xbfffff ---> #pci (0xa0000-0xbffff)
+ | (prio 1)
+ |
+ +---- pci-hole: alias@0xe0000000-0xffffffff ---> #pci (0xe0000000-0xffffffff)
+
+pci (0-2^32-1)
+ |
+ +--- vga-area: container@0xa0000-0xbffff
+ | |
+ | +--- alias@0x00000-0x7fff ---> #vram (0x010000-0x017fff)
+ | |
+ | +--- alias@0x08000-0xffff ---> #vram (0x020000-0x027fff)
+ |
+ +---- vram: ram@0xe1000000-0xe1ffffff
+ |
+ +---- vga-mmio: mmio@0xe2000000-0xe200ffff
+
+ram: ram@0x00000000-0xffffffff
+
+The is a (simplified) PC memory map. The 4GB RAM block is mapped into the
+system address space via two aliases: "lomem" is a 1:1 mapping of the first
+3.5GB; "himem" maps the last 0.5GB at address 4GB. This leaves 0.5GB for the
+so-called PCI hole, that allows a 32-bit PCI bus to exist in a system with
+4GB of memory.
+
+The memory controller diverts addresses in the range 640K-768K to the PCI
+address space. This is modeled using the "vga-window" alias, mapped at a
+higher priority so it obscures the RAM at the same addresses. The vga window
+can be removed by programming the memory controller; this is modelled by
+removing the alias and exposing the RAM underneath.
+
+The pci address space is not a direct child of the system address space, since
+we only want parts of it to be visible (we accomplish this using aliases).
+It has two subregions: vga-area models the legacy vga window and is occupied
+by two 32K memory banks pointing at two sections of the framebuffer.
+In addition the vram is mapped as a BAR at address e1000000, and an additional
+BAR containing MMIO registers is mapped after it.
+
+Note that if the guest maps a BAR outside the PCI hole, it would not be
+visible as the pci-hole alias clips it to a 0.5GB range.
+
+Attributes
+----------
+
+Various region attributes (read-only, dirty logging, coalesced mmio, ioeventfd)
+can be changed during the region lifecycle. They take effect once the region
+is made visible (which can be immediately, later, or never).
+
+MMIO Operations
+---------------
+
+MMIO regions are provided with ->read() and ->write() callbacks; in addition
+various constraints can be supplied to control how these callbacks are called:
+
+ - .valid.min_access_size, .valid.max_access_size define the access sizes
+ (in bytes) which the device accepts; accesses outside this range will
+ have device and bus specific behaviour (ignored, or machine check)
+ - .valid.aligned specifies that the device only accepts naturally aligned
+ accesses. Unaligned accesses invoke device and bus specific behaviour.
+ - .impl.min_access_size, .impl.max_access_size define the access sizes
+ (in bytes) supported by the *implementation*; other access sizes will be
+ emulated using the ones available. For example a 4-byte write will be
+ emulated using four 1-byte write, is .impl.max_access_size = 1.
+ - .impl.valid specifies that the *implementation* only supports unaligned
+ accesses; unaligned accesses will be emulated by two aligned accesses.
+ - .old_portio and .old_mmio can be used to ease porting from code using
+ cpu_register_io_memory() and register_ioport(). They should not be used
+ in new code.
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 02/23] Hierarchical memory region API
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 01/23] Add memory API documentation Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 03/23] memory: implement dirty tracking Avi Kivity
` (21 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
The memory API separates the attributes of a memory region (its size, how
reads or writes are handled, dirty logging, and coalescing) from where it
is mapped and whether it is enabled. This allows a device to configure
a memory region once, then hand it off to its parent bus to map it according
to the bus configuration.
Hierarchical registration also allows a device to compose a region out of
a number of sub-regions with different properties; for example some may be
RAM while others may be MMIO.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
Makefile.target | 1 +
memory.c | 653 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
memory.h | 385 ++++++++++++++++++++++++++++++++
3 files changed, 1039 insertions(+), 0 deletions(-)
create mode 100644 memory.c
create mode 100644 memory.h
diff --git a/Makefile.target b/Makefile.target
index cde509b..8884a56 100644
--- a/Makefile.target
+++ b/Makefile.target
@@ -198,6 +198,7 @@ obj-$(CONFIG_REALLY_VIRTFS) += 9pfs/virtio-9p-device.o
obj-y += rwhandler.o
obj-$(CONFIG_KVM) += kvm.o kvm-all.o
obj-$(CONFIG_NO_KVM) += kvm-stub.o
+obj-y += memory.o
LIBS+=-lz
QEMU_CFLAGS += $(VNC_TLS_CFLAGS)
diff --git a/memory.c b/memory.c
new file mode 100644
index 0000000..a9cf317
--- /dev/null
+++ b/memory.c
@@ -0,0 +1,653 @@
+/*
+ * Physical memory management
+ *
+ * Copyright 2011 Red Hat, Inc. and/or its affiliates
+ *
+ * Authors:
+ * Avi Kivity <avi@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2. See
+ * the COPYING file in the top-level directory.
+ *
+ */
+
+#include "memory.h"
+#include <assert.h>
+
+typedef struct AddrRange AddrRange;
+
+struct AddrRange {
+ uint64_t start;
+ uint64_t size;
+};
+
+static AddrRange addrrange_make(uint64_t start, uint64_t size)
+{
+ return (AddrRange) { start, size };
+}
+
+static bool addrrange_equal(AddrRange r1, AddrRange r2)
+{
+ return r1.start == r2.start && r1.size == r2.size;
+}
+
+static uint64_t addrrange_end(AddrRange r)
+{
+ return r.start + r.size;
+}
+
+static AddrRange addrrange_shift(AddrRange range, int64_t delta)
+{
+ range.start += delta;
+ return range;
+}
+
+static bool addrrange_intersects(AddrRange r1, AddrRange r2)
+{
+ return (r1.start >= r2.start && r1.start < r2.start + r2.size)
+ || (r2.start >= r1.start && r2.start < r1.start + r1.size);
+}
+
+static AddrRange addrrange_intersection(AddrRange r1, AddrRange r2)
+{
+ uint64_t start = MAX(r1.start, r2.start);
+ /* off-by-one arithmetic to prevent overflow */
+ uint64_t end = MIN(addrrange_end(r1) - 1, addrrange_end(r2) - 1);
+ return addrrange_make(start, end - start + 1);
+}
+
+struct CoalescedMemoryRange {
+ AddrRange addr;
+ QTAILQ_ENTRY(CoalescedMemoryRange) link;
+};
+
+typedef struct FlatRange FlatRange;
+typedef struct FlatView FlatView;
+
+/* Range of memory in the global map. Addresses are absolute. */
+struct FlatRange {
+ MemoryRegion *mr;
+ target_phys_addr_t offset_in_region;
+ AddrRange addr;
+};
+
+/* Flattened global view of current active memory hierarchy. Kept in sorted
+ * order.
+ */
+struct FlatView {
+ FlatRange *ranges;
+ unsigned nr;
+ unsigned nr_allocated;
+};
+
+#define FOR_EACH_FLAT_RANGE(var, view) \
+ for (var = (view)->ranges; var < (view)->ranges + (view)->nr; ++var)
+
+static FlatView current_memory_map;
+static MemoryRegion *root_memory_region;
+
+static bool flatrange_equal(FlatRange *a, FlatRange *b)
+{
+ return a->mr == b->mr
+ && addrrange_equal(a->addr, b->addr)
+ && a->offset_in_region == b->offset_in_region;
+}
+
+static void flatview_init(FlatView *view)
+{
+ view->ranges = NULL;
+ view->nr = 0;
+ view->nr_allocated = 0;
+}
+
+/* Insert a range into a given position. Caller is responsible for maintaining
+ * sorting order.
+ */
+static void flatview_insert(FlatView *view, unsigned pos, FlatRange *range)
+{
+ if (view->nr == view->nr_allocated) {
+ view->nr_allocated = MAX(2 * view->nr, 10);
+ view->ranges = qemu_realloc(view->ranges,
+ view->nr_allocated * sizeof(*view->ranges));
+ }
+ memmove(view->ranges + pos + 1, view->ranges + pos,
+ (view->nr - pos) * sizeof(FlatRange));
+ view->ranges[pos] = *range;
+ ++view->nr;
+}
+
+static void flatview_destroy(FlatView *view)
+{
+ qemu_free(view->ranges);
+}
+
+/* Render a memory region into the global view. Ranges in @view obscure
+ * ranges in @mr.
+ */
+static void render_memory_region(FlatView *view,
+ MemoryRegion *mr,
+ target_phys_addr_t base,
+ AddrRange clip)
+{
+ MemoryRegion *subregion;
+ unsigned i;
+ target_phys_addr_t offset_in_region;
+ uint64_t remain;
+ uint64_t now;
+ FlatRange fr;
+ AddrRange tmp;
+
+ base += mr->addr;
+
+ tmp = addrrange_make(base, mr->size);
+
+ if (!addrrange_intersects(tmp, clip)) {
+ return;
+ }
+
+ clip = addrrange_intersection(tmp, clip);
+
+ if (mr->alias) {
+ base -= mr->alias->addr;
+ base -= mr->alias_offset;
+ render_memory_region(view, mr->alias, base, clip);
+ return;
+ }
+
+ /* Render subregions in priority order. */
+ QTAILQ_FOREACH(subregion, &mr->subregions, subregions_link) {
+ render_memory_region(view, subregion, base, clip);
+ }
+
+ if (!mr->has_ram_addr) {
+ return;
+ }
+
+ offset_in_region = clip.start - base;
+ base = clip.start;
+ remain = clip.size;
+
+ /* Render the region itself into any gaps left by the current view. */
+ for (i = 0; i < view->nr && remain; ++i) {
+ if (base >= addrrange_end(view->ranges[i].addr)) {
+ continue;
+ }
+ if (base < view->ranges[i].addr.start) {
+ now = MIN(remain, view->ranges[i].addr.start - base);
+ fr.mr = mr;
+ fr.offset_in_region = offset_in_region;
+ fr.addr = addrrange_make(base, now);
+ flatview_insert(view, i, &fr);
+ ++i;
+ base += now;
+ offset_in_region += now;
+ remain -= now;
+ }
+ if (base == view->ranges[i].addr.start) {
+ now = MIN(remain, view->ranges[i].addr.size);
+ base += now;
+ offset_in_region += now;
+ remain -= now;
+ }
+ }
+ if (remain) {
+ fr.mr = mr;
+ fr.offset_in_region = offset_in_region;
+ fr.addr = addrrange_make(base, remain);
+ flatview_insert(view, i, &fr);
+ }
+}
+
+/* Render a memory topology into a list of disjoint absolute ranges. */
+static FlatView generate_memory_topology(MemoryRegion *mr)
+{
+ FlatView view;
+
+ flatview_init(&view);
+
+ render_memory_region(&view, mr, 0, addrrange_make(0, UINT64_MAX));
+
+ return view;
+}
+
+static void memory_region_update_topology(void)
+{
+ FlatView old_view = current_memory_map;
+ FlatView new_view = generate_memory_topology(root_memory_region);
+ unsigned iold, inew;
+ FlatRange *frold, *frnew;
+ ram_addr_t phys_offset, region_offset;
+
+ /* Generate a symmetric difference of the old and new memory maps.
+ * Kill ranges in the old map, and instantiate ranges in the new map.
+ */
+ iold = inew = 0;
+ while (iold < old_view.nr || inew < new_view.nr) {
+ if (iold < old_view.nr) {
+ frold = &old_view.ranges[iold];
+ } else {
+ frold = NULL;
+ }
+ if (inew < new_view.nr) {
+ frnew = &new_view.ranges[inew];
+ } else {
+ frnew = NULL;
+ }
+
+ if (frold
+ && (!frnew
+ || frold->addr.start < frnew->addr.start
+ || (frold->addr.start == frnew->addr.start
+ && !flatrange_equal(frold, frnew)))) {
+ /* In old, but (not in new, or in new but attributes changed). */
+
+ cpu_register_physical_memory(frold->addr.start, frold->addr.size,
+ IO_MEM_UNASSIGNED);
+ ++iold;
+ } else if (frold && frnew && flatrange_equal(frold, frnew)) {
+ /* In both (logging may have changed) */
+
+ ++iold;
+ ++inew;
+ /* FIXME: dirty logging */
+ } else {
+ /* In new */
+
+ phys_offset = frnew->mr->ram_addr;
+ region_offset = frnew->offset_in_region;
+ /* cpu_register_physical_memory_log() wants region_offset for
+ * mmio, but prefers offseting phys_offset for RAM. Humour it.
+ */
+ if ((phys_offset & ~TARGET_PAGE_MASK) <= IO_MEM_ROM) {
+ phys_offset += region_offset;
+ region_offset = 0;
+ }
+
+ cpu_register_physical_memory_log(frnew->addr.start,
+ frnew->addr.size,
+ phys_offset,
+ region_offset,
+ 0);
+ ++inew;
+ }
+ }
+ current_memory_map = new_view;
+ flatview_destroy(&old_view);
+}
+
+void memory_region_init(MemoryRegion *mr,
+ const char *name,
+ uint64_t size)
+{
+ mr->ops = NULL;
+ mr->parent = NULL;
+ mr->size = size;
+ mr->addr = 0;
+ mr->offset = 0;
+ mr->has_ram_addr = false;
+ mr->priority = 0;
+ mr->may_overlap = false;
+ mr->alias = NULL;
+ QTAILQ_INIT(&mr->subregions);
+ memset(&mr->subregions_link, 0, sizeof mr->subregions_link);
+ QTAILQ_INIT(&mr->coalesced);
+ mr->name = qemu_strdup(name);
+}
+
+static bool memory_region_access_valid(MemoryRegion *mr,
+ target_phys_addr_t addr,
+ unsigned size)
+{
+ if (!mr->ops->valid.unaligned && (addr & (size - 1))) {
+ return false;
+ }
+
+ /* Treat zero as compatibility all valid */
+ if (!mr->ops->valid.max_access_size) {
+ return true;
+ }
+
+ if (size > mr->ops->valid.max_access_size
+ || size < mr->ops->valid.min_access_size) {
+ return false;
+ }
+ return true;
+}
+
+static uint32_t memory_region_read_thunk_n(void *_mr,
+ target_phys_addr_t addr,
+ unsigned size)
+{
+ MemoryRegion *mr = _mr;
+ unsigned access_size, access_size_min, access_size_max;
+ uint64_t access_mask;
+ uint32_t data = 0, tmp;
+ unsigned i;
+
+ if (!memory_region_access_valid(mr, addr, size)) {
+ return -1U; /* FIXME: better signalling */
+ }
+
+ /* FIXME: support unaligned access */
+
+ access_size_min = mr->ops->impl.min_access_size;
+ if (!access_size_min) {
+ access_size_min = 1;
+ }
+ access_size_max = mr->ops->impl.max_access_size;
+ if (!access_size_max) {
+ access_size_max = 4;
+ }
+ access_size = MAX(MIN(size, access_size_max), access_size_min);
+ access_mask = -1ULL >> (64 - access_size * 8);
+ addr += mr->offset;
+ for (i = 0; i < size; i += access_size) {
+ /* FIXME: big-endian support */
+ tmp = mr->ops->read(mr->opaque, addr + i, access_size);
+ data |= (tmp & access_mask) << (i * 8);
+ }
+
+ return data;
+}
+
+static void memory_region_write_thunk_n(void *_mr,
+ target_phys_addr_t addr,
+ unsigned size,
+ uint64_t data)
+{
+ MemoryRegion *mr = _mr;
+ unsigned access_size, access_size_min, access_size_max;
+ uint64_t access_mask;
+ unsigned i;
+
+ if (!memory_region_access_valid(mr, addr, size)) {
+ return; /* FIXME: better signalling */
+ }
+
+ /* FIXME: support unaligned access */
+
+ access_size_min = mr->ops->impl.min_access_size;
+ if (!access_size_min) {
+ access_size_min = 1;
+ }
+ access_size_max = mr->ops->impl.max_access_size;
+ if (!access_size_max) {
+ access_size_max = 4;
+ }
+ access_size = MAX(MIN(size, access_size_max), access_size_min);
+ access_mask = -1ULL >> (64 - access_size * 8);
+ addr += mr->offset;
+ for (i = 0; i < size; i += access_size) {
+ /* FIXME: big-endian support */
+ mr->ops->write(mr->opaque, addr + i, (data >> (i * 8)) & access_mask,
+ access_size);
+ }
+}
+
+static uint32_t memory_region_read_thunk_b(void *mr, target_phys_addr_t addr)
+{
+ return memory_region_read_thunk_n(mr, addr, 1);
+}
+
+static uint32_t memory_region_read_thunk_w(void *mr, target_phys_addr_t addr)
+{
+ return memory_region_read_thunk_n(mr, addr, 2);
+}
+
+static uint32_t memory_region_read_thunk_l(void *mr, target_phys_addr_t addr)
+{
+ return memory_region_read_thunk_n(mr, addr, 4);
+}
+
+static void memory_region_write_thunk_b(void *mr, target_phys_addr_t addr,
+ uint32_t data)
+{
+ memory_region_write_thunk_n(mr, addr, 1, data);
+}
+
+static void memory_region_write_thunk_w(void *mr, target_phys_addr_t addr,
+ uint32_t data)
+{
+ memory_region_write_thunk_n(mr, addr, 2, data);
+}
+
+static void memory_region_write_thunk_l(void *mr, target_phys_addr_t addr,
+ uint32_t data)
+{
+ memory_region_write_thunk_n(mr, addr, 4, data);
+}
+
+static CPUReadMemoryFunc * const memory_region_read_thunk[] = {
+ memory_region_read_thunk_b,
+ memory_region_read_thunk_w,
+ memory_region_read_thunk_l,
+};
+
+static CPUWriteMemoryFunc * const memory_region_write_thunk[] = {
+ memory_region_write_thunk_b,
+ memory_region_write_thunk_w,
+ memory_region_write_thunk_l,
+};
+
+void memory_region_init_io(MemoryRegion *mr,
+ const MemoryRegionOps *ops,
+ void *opaque,
+ const char *name,
+ uint64_t size)
+{
+ memory_region_init(mr, name, size);
+ mr->ops = ops;
+ mr->opaque = opaque;
+ mr->has_ram_addr = true;
+ mr->ram_addr = cpu_register_io_memory(memory_region_read_thunk,
+ memory_region_write_thunk,
+ mr,
+ mr->ops->endianness);
+}
+
+void memory_region_init_ram(MemoryRegion *mr,
+ DeviceState *dev,
+ const char *name,
+ uint64_t size)
+{
+ memory_region_init(mr, name, size);
+ mr->has_ram_addr = true;
+ mr->ram_addr = qemu_ram_alloc(dev, name, size);
+}
+
+void memory_region_init_ram_ptr(MemoryRegion *mr,
+ DeviceState *dev,
+ const char *name,
+ uint64_t size,
+ void *ptr)
+{
+ memory_region_init(mr, name, size);
+ mr->has_ram_addr = true;
+ mr->ram_addr = qemu_ram_alloc_from_ptr(dev, name, size, ptr);
+}
+
+void memory_region_init_alias(MemoryRegion *mr,
+ const char *name,
+ MemoryRegion *orig,
+ target_phys_addr_t offset,
+ uint64_t size)
+{
+ memory_region_init(mr, name, size);
+ mr->alias = orig;
+ mr->alias_offset = offset;
+}
+
+void memory_region_destroy(MemoryRegion *mr)
+{
+ assert(QTAILQ_EMPTY(&mr->subregions));
+ memory_region_clear_coalescing(mr);
+ qemu_free((char *)mr->name);
+}
+
+uint64_t memory_region_size(MemoryRegion *mr)
+{
+ return mr->size;
+}
+
+void memory_region_set_offset(MemoryRegion *mr, target_phys_addr_t offset)
+{
+ mr->offset = offset;
+}
+
+void memory_region_set_log(MemoryRegion *mr, bool log, unsigned client)
+{
+ /* FIXME */
+}
+
+bool memory_region_get_dirty(MemoryRegion *mr, target_phys_addr_t addr,
+ unsigned client)
+{
+ /* FIXME */
+ return true;
+}
+
+void memory_region_set_dirty(MemoryRegion *mr, target_phys_addr_t addr)
+{
+ /* FIXME */
+}
+
+void memory_region_sync_dirty_bitmap(MemoryRegion *mr)
+{
+ /* FIXME */
+}
+
+void memory_region_set_readonly(MemoryRegion *mr, bool readonly)
+{
+ /* FIXME */
+}
+
+void memory_region_reset_dirty(MemoryRegion *mr, target_phys_addr_t addr,
+ target_phys_addr_t size, unsigned client)
+{
+ /* FIXME */
+}
+
+void *memory_region_get_ram_ptr(MemoryRegion *mr)
+{
+ if (mr->alias) {
+ return memory_region_get_ram_ptr(mr->alias) + mr->alias_offset;
+ }
+
+ assert(mr->has_ram_addr);
+
+ return qemu_get_ram_ptr(mr->ram_addr);
+}
+
+static void memory_region_update_coalesced_range(MemoryRegion *mr)
+{
+ FlatRange *fr;
+ CoalescedMemoryRange *cmr;
+ AddrRange tmp;
+
+ FOR_EACH_FLAT_RANGE(fr, ¤t_memory_map) {
+ if (fr->mr == mr) {
+ qemu_unregister_coalesced_mmio(fr->addr.start, fr->addr.size);
+ QTAILQ_FOREACH(cmr, &mr->coalesced, link) {
+ tmp = addrrange_shift(cmr->addr,
+ fr->addr.start - fr->offset_in_region);
+ if (!addrrange_intersects(tmp, fr->addr)) {
+ continue;
+ }
+ tmp = addrrange_intersection(tmp, fr->addr);
+ qemu_register_coalesced_mmio(tmp.start, tmp.size);
+ }
+ }
+ }
+}
+
+void memory_region_set_coalescing(MemoryRegion *mr)
+{
+ memory_region_clear_coalescing(mr);
+ memory_region_add_coalescing(mr, 0, mr->size);
+}
+
+void memory_region_add_coalescing(MemoryRegion *mr,
+ target_phys_addr_t offset,
+ uint64_t size)
+{
+ CoalescedMemoryRange *cmr = qemu_malloc(sizeof(*cmr));
+
+ cmr->addr = addrrange_make(offset, size);
+ QTAILQ_INSERT_TAIL(&mr->coalesced, cmr, link);
+ memory_region_update_coalesced_range(mr);
+}
+
+void memory_region_clear_coalescing(MemoryRegion *mr)
+{
+ CoalescedMemoryRange *cmr;
+
+ while (!QTAILQ_EMPTY(&mr->coalesced)) {
+ cmr = QTAILQ_FIRST(&mr->coalesced);
+ QTAILQ_REMOVE(&mr->coalesced, cmr, link);
+ qemu_free(cmr);
+ }
+ memory_region_update_coalesced_range(mr);
+}
+
+static void memory_region_add_subregion_common(MemoryRegion *mr,
+ target_phys_addr_t offset,
+ MemoryRegion *subregion)
+{
+ MemoryRegion *other;
+
+ assert(!subregion->parent);
+ subregion->parent = mr;
+ subregion->addr = offset;
+ QTAILQ_FOREACH(other, &mr->subregions, subregions_link) {
+ if (subregion->may_overlap || other->may_overlap) {
+ continue;
+ }
+ if (offset >= other->offset + other->size
+ || offset + subregion->size <= other->offset) {
+ continue;
+ }
+ printf("warning: subregion collision %llx/%llx vs %llx/%llx\n",
+ (unsigned long long)offset,
+ (unsigned long long)subregion->size,
+ (unsigned long long)other->offset,
+ (unsigned long long)other->size);
+ }
+ QTAILQ_FOREACH(other, &mr->subregions, subregions_link) {
+ if (subregion->priority >= other->priority) {
+ QTAILQ_INSERT_BEFORE(other, subregion, subregions_link);
+ goto done;
+ }
+ }
+ QTAILQ_INSERT_TAIL(&mr->subregions, subregion, subregions_link);
+done:
+ memory_region_update_topology();
+}
+
+
+void memory_region_add_subregion(MemoryRegion *mr,
+ target_phys_addr_t offset,
+ MemoryRegion *subregion)
+{
+ subregion->may_overlap = false;
+ subregion->priority = 0;
+ memory_region_add_subregion_common(mr, offset, subregion);
+}
+
+void memory_region_add_subregion_overlap(MemoryRegion *mr,
+ target_phys_addr_t offset,
+ MemoryRegion *subregion,
+ unsigned priority)
+{
+ subregion->may_overlap = true;
+ subregion->priority = priority;
+ memory_region_add_subregion_common(mr, offset, subregion);
+}
+
+void memory_region_del_subregion(MemoryRegion *mr,
+ MemoryRegion *subregion)
+{
+ assert(subregion->parent == mr);
+ subregion->parent = NULL;
+ QTAILQ_REMOVE(&mr->subregions, subregion, subregions_link);
+ memory_region_update_topology();
+}
diff --git a/memory.h b/memory.h
new file mode 100644
index 0000000..a4c94bd
--- /dev/null
+++ b/memory.h
@@ -0,0 +1,385 @@
+/*
+ * Physical memory management API
+ *
+ * Copyright 2011 Red Hat, Inc. and/or its affiliates
+ *
+ * Authors:
+ * Avi Kivity <avi@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2. See
+ * the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef MEMORY_H
+#define MEMORY_H
+
+#ifndef CONFIG_USER_ONLY
+
+#include <stdint.h>
+#include <stdbool.h>
+#include "qemu-common.h"
+#include "cpu-common.h"
+#include "targphys.h"
+#include "qemu-queue.h"
+
+typedef struct MemoryRegionOps MemoryRegionOps;
+typedef struct MemoryRegion MemoryRegion;
+
+/* Must match *_DIRTY_FLAGS in cpu-all.h. To be replaced with dynamic
+ * registration.
+ */
+#define DIRTY_MEMORY_VGA 0
+#define DIRTY_MEMORY_CODE 1
+#define DIRTY_MEMORY_MIGRATION 3
+
+/*
+ * Memory region callbacks
+ */
+struct MemoryRegionOps {
+ /* Read from the memory region. @addr is relative to @mr; @size is
+ * in bytes. */
+ uint64_t (*read)(void *opaque,
+ target_phys_addr_t addr,
+ unsigned size);
+ /* Write to the memory region. @addr is relative to @mr; @size is
+ * in bytes. */
+ void (*write)(void *opaque,
+ target_phys_addr_t addr,
+ uint64_t data,
+ unsigned size);
+
+ enum device_endian endianness;
+ /* Guest-visible constraints: */
+ struct {
+ /* If nonzero, specify bounds on access sizes beyond which a machine
+ * check is thrown.
+ */
+ unsigned min_access_size;
+ unsigned max_access_size;
+ /* If true, unaligned accesses are supported. Otherwise unaligned
+ * accesses throw machine checks.
+ */
+ bool unaligned;
+ } valid;
+ /* Internal implementation constraints: */
+ struct {
+ /* If nonzero, specifies the minimum size implemented. Smaller sizes
+ * will be rounded upwards and a partial result will be returned.
+ */
+ unsigned min_access_size;
+ /* If nonzero, specifies the maximum size implemented. Larger sizes
+ * will be done as a series of accesses with smaller sizes.
+ */
+ unsigned max_access_size;
+ /* If true, unaligned accesses are supported. Otherwise all accesses
+ * are converted to (possibly multiple) naturally aligned accesses.
+ */
+ bool unaligned;
+ } impl;
+};
+
+typedef struct CoalescedMemoryRange CoalescedMemoryRange;
+
+struct MemoryRegion {
+ /* All fields are private - violators will be prosecuted */
+ const MemoryRegionOps *ops;
+ void *opaque;
+ MemoryRegion *parent;
+ uint64_t size;
+ target_phys_addr_t addr;
+ target_phys_addr_t offset;
+ ram_addr_t ram_addr;
+ bool has_ram_addr;
+ MemoryRegion *alias;
+ target_phys_addr_t alias_offset;
+ unsigned priority;
+ bool may_overlap;
+ QTAILQ_HEAD(subregions, MemoryRegion) subregions;
+ QTAILQ_ENTRY(MemoryRegion) subregions_link;
+ QTAILQ_HEAD(coalesced_ranges, CoalescedMemoryRange) coalesced;
+ const char *name;
+};
+
+/**
+ * memory_region_init: Initialize a memory region
+ *
+ * The region typically acts as a container for other memory regions. Us
+ * memory_region_add_subregion() to add subregions.
+ *
+ * @mr: the #MemoryRegion to be initialized
+ * @name: used for debugging; not visible to the user or ABI
+ * @size: size of the region; any subregions beyond this size will be clipped
+ */
+void memory_region_init(MemoryRegion *mr,
+ const char *name,
+ uint64_t size);
+/**
+ * memory_region_init_io: Initialize an I/O memory region.
+ *
+ * Accesses into the region will be cause the callbacks in @ops to be called.
+ * if @size is nonzero, subregions will be clipped to @size.
+ *
+ * @mr: the #MemoryRegion to be initialized.
+ * @ops: a structure containing read and write callbacks to be used when
+ * I/O is performed on the region.
+ * @opaque: passed to to the read and write callbacks of the @ops structure.
+ * @name: used for debugging; not visible to the user or ABI
+ * @size: size of the region.
+ */
+void memory_region_init_io(MemoryRegion *mr,
+ const MemoryRegionOps *ops,
+ void *opaque,
+ const char *name,
+ uint64_t size);
+
+/**
+ * memory_region_init_ram: Initialize RAM memory region. Accesses into the
+ * region will be modify memory directly.
+ *
+ * @mr: the #MemoryRegion to be initialized.
+ * @dev: a device associated with the region; may be %NULL.
+ * @name: the name of the region; the pair (@dev, @name) must be globally
+ * unique. The name is part of the save/restore ABI and so cannot be
+ * changed.
+ * @size: size of the region.
+ */
+void memory_region_init_ram(MemoryRegion *mr,
+ DeviceState *dev, /* FIXME: layering violation */
+ const char *name,
+ uint64_t size);
+
+/**
+ * memory_region_init_ram: Initialize RAM memory region from a user-provided.
+ * pointer. Accesses into the region will be modify
+ * memory directly.
+ *
+ * @mr: the #MemoryRegion to be initialized.
+ * @dev: a device associated with the region; may be %NULL.
+ * @name: the name of the region; the pair (@dev, @name) must be globally
+ * unique. The name is part of the save/restore ABI and so cannot be
+ * changed.
+ * @size: size of the region.
+ * @ptr: memory to be mapped; must contain at least @size bytes.
+ */
+void memory_region_init_ram_ptr(MemoryRegion *mr,
+ DeviceState *dev, /* FIXME: layering violation */
+ const char *name,
+ uint64_t size,
+ void *ptr);
+
+/**
+ * memory_region_init_alias: Initialize a memory region that aliases all or a
+ * part of another memory region.
+ *
+ * @mr: the #MemoryRegion to be initialized.
+ * @name: used for debugging; not visible to the user or ABI
+ * @orig: the region to be referenced; @mr will be equivalent to
+ * @orig between @offset and @offset + @size - 1.
+ * @offset: start of the section in @orig to be referenced.
+ * @size: size of the region.
+ */
+void memory_region_init_alias(MemoryRegion *mr,
+ const char *name,
+ MemoryRegion *orig,
+ target_phys_addr_t offset,
+ uint64_t size);
+/**
+ * memory_region_destroy: Destroy a memory region and relaim all resources.
+ *
+ * @mr: the region to be destroyed. May not currently be a subregion
+ * (see memory_region_add_subregion()) or referenced in an alias
+ * (see memory_region_init_alias()).
+ */
+void memory_region_destroy(MemoryRegion *mr);
+
+/**
+ * memory_region_size: get a memory region's size.
+ *
+ * @mr: the memory region being queried.
+ */
+uint64_t memory_region_size(MemoryRegion *mr);
+
+/**
+ * memory_region_get_ram_ptr: Get a pointer into a RAM memory region.
+ *
+ * Returns a host pointer to a RAM memory region (created with
+ * memory_region_init_ram() or memory_region_init_ram_ptr()). Use with
+ * care.
+ *
+ * @mr: the memory region being queried.
+ */
+void *memory_region_get_ram_ptr(MemoryRegion *mr);
+
+/**
+ * memory_region_set_offset: Sets an offset to be added to MemoryRegionOps
+ * callbacks.
+ *
+ * This function is deprecated and should not be used in new code.
+ */
+void memory_region_set_offset(MemoryRegion *mr, target_phys_addr_t offset);
+
+/**
+ * memory_region_set_log: Turn dirty logging on or off for a region.
+ *
+ * Turns dirty logging on or off for a specified client (display, migration).
+ * Only meaningful for RAM regions.
+ *
+ * @mr: the memory region being updated.
+ * @log: whether dirty logging is to be enabled or disabled.
+ * @client: the user of the logging information; %DIRTY_MEMORY_MIGRATION or
+ * %DIRTY_MEMORY_VGA.
+ */
+void memory_region_set_log(MemoryRegion *mr, bool log, unsigned client);
+
+/**
+ * memory_region_get_dirty: Check whether a page is dirty for a specified
+ * client.
+ *
+ * Checks whether a page has been written to since the last
+ * call to memory_region_reset_dirty() with the same @client. Dirty logging
+ * must be enabled.
+ *
+ * @mr: the memory region being queried.
+ * @addr: the address (relative to the start of the region) being queried.
+ * @client: the user of the logging information; %DIRTY_MEMORY_MIGRATION or
+ * %DIRTY_MEMORY_VGA.
+ */
+bool memory_region_get_dirty(MemoryRegion *mr, target_phys_addr_t addr,
+ unsigned client);
+
+/**
+ * memory_region_set_dirty: Mark a page as dirty in a memory region.
+ *
+ * Marks a page as dirty, after it has been dirtied outside guest code.
+ *
+ * @mr: the memory region being queried.
+ * @addr: the address (relative to the start of the region) being dirtied.
+ */
+void memory_region_set_dirty(MemoryRegion *mr, target_phys_addr_t addr);
+
+/**
+ * memory_region_sync_dirty_bitmap: Synchronize a region's dirty bitmap with
+ * any external TLBs (e.g. kvm)
+ *
+ * Flushes dirty information from accelerators such as kvm and vhost-net
+ * and makes it available to users of the memory API.
+ *
+ * @mr: the region being flushed.
+ */
+void memory_region_sync_dirty_bitmap(MemoryRegion *mr);
+
+/**
+ * memory_region_reset_dirty: Mark a range of pages as clean, for a specified
+ * client.
+ *
+ * Marks a range of pages as no longer dirty.
+ *
+ * @mr: the region being updated.
+ * @addr: the start of the subrange being cleaned.
+ * @size: the size of the subrange being cleaned.
+ * @client: the user of the logging information; %DIRTY_MEMORY_MIGRATION or
+ * %DIRTY_MEMORY_VGA.
+ */
+void memory_region_reset_dirty(MemoryRegion *mr, target_phys_addr_t addr,
+ target_phys_addr_t size, unsigned client);
+
+/**
+ * memory_region_set_readonly: Turn a memory region read-only (or read-write)
+ *
+ * Allows a memory region to be marked as read-only (turning it into a ROM).
+ * only useful on RAM regions.
+ *
+ * @mr: the region being updated.
+ * @readonly: whether rhe region is to be ROM or RAM.
+ */
+void memory_region_set_readonly(MemoryRegion *mr, bool readonly);
+
+/**
+ * memory_region_set_coalescing: Enable memory coalescing for the region.
+ *
+ * Enabled writes to a region to be queued for later processing. MMIO ->write
+ * callbacks may be delayed until a non-coalesced MMIO is issued.
+ * Only useful for IO regions. Roughly similar to write-combining hardware.
+ *
+ * @mr: the memory region to be write coalesced
+ */
+void memory_region_set_coalescing(MemoryRegion *mr);
+
+/**
+ * memory_region_add_coalescing: Enable memory coalescing for a sub-range of
+ * a region.
+ *
+ * Like memory_region_set_coalescing(), but works on a sub-range of a region.
+ * Multiple calls can be issued coalesced disjoint ranges.
+ *
+ * @mr: the memory region to be updated.
+ * @offset: the start of the range within the region to be coalesced.
+ * @size: the size of the subrange to be coalesced.
+ */
+void memory_region_add_coalescing(MemoryRegion *mr,
+ target_phys_addr_t offset,
+ uint64_t size);
+
+/**
+ * memory_region_clear_coalescing: Disable MMIO coalescing for the region.
+ *
+ * Disables any coalescing caused by memory_region_set_coalescing() or
+ * memory_region_add_coalescing(). Roughly equivalent to uncacheble memory
+ * hardware.
+ *
+ * @mr: the memory region to be updated.
+ */
+void memory_region_clear_coalescing(MemoryRegion *mr);
+
+/**
+ * memory_region_add_subregion: Add a sub-region to a container.
+ *
+ * Adds a sub-region at @offset. The sub-region may not overlap with other
+ * subregions (except for those explicitly marked as overlapping). A region
+ * may only be added once as a subregion (unless removed with
+ * memory_region_del_subregion()); use memory_region_init_alias() if you
+ * want a region to be a subregion in multiple locations.
+ *
+ * @mr: the region to contain the new subregion; must be a container
+ * initialized with memory_region_init().
+ * @offset: the offset relative to @mr where @subregion is added.
+ * @subregion: the subregion to be added.
+ */
+void memory_region_add_subregion(MemoryRegion *mr,
+ target_phys_addr_t offset,
+ MemoryRegion *subregion);
+/**
+ * memory_region_add_subregion: Add a sub-region to a container, with overlap.
+ *
+ * Adds a sub-region at @offset. The sub-region may overlap with other
+ * subregions. Conflicts are resolved by having a higher @priority hide a
+ * lower @priority. Subregions without priority are taken as @priority 0.
+ * A region may only be added once as a subregion (unless removed with
+ * memory_region_del_subregion()); use memory_region_init_alias() if you
+ * want a region to be a subregion in multiple locations.
+ *
+ * @mr: the region to contain the new subregion; must be a container
+ * initialized with memory_region_init().
+ * @offset: the offset relative to @mr where @subregion is added.
+ * @subregion: the subregion to be added.
+ * @priority: used for resolving overlaps; highest priority wins.
+ */
+void memory_region_add_subregion_overlap(MemoryRegion *mr,
+ target_phys_addr_t offset,
+ MemoryRegion *subregion,
+ unsigned priority);
+/**
+ * memory_region_del_subregion: Remove a subregion.
+ *
+ * Removes a subregion from its container.
+ *
+ * @mr: the container to be updated.
+ * @subregion: the region being removed; must be a current subregion of @mr.
+ */
+void memory_region_del_subregion(MemoryRegion *mr,
+ MemoryRegion *subregion);
+
+#endif
+
+#endif
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 03/23] memory: implement dirty tracking
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 01/23] Add memory API documentation Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 02/23] Hierarchical memory region API Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region Avi Kivity
` (20 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Currently dirty tracking is implemented by passing through
all calls to the underlying cpu_physical_memory_*() calls.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 39 +++++++++++++++++++++++++++++++--------
memory.h | 1 +
2 files changed, 32 insertions(+), 8 deletions(-)
diff --git a/memory.c b/memory.c
index a9cf317..d858b47 100644
--- a/memory.c
+++ b/memory.c
@@ -69,6 +69,7 @@ struct FlatRange {
MemoryRegion *mr;
target_phys_addr_t offset_in_region;
AddrRange addr;
+ uint8_t dirty_log_mask;
};
/* Flattened global view of current active memory hierarchy. Kept in sorted
@@ -177,6 +178,7 @@ static void render_memory_region(FlatView *view,
fr.mr = mr;
fr.offset_in_region = offset_in_region;
fr.addr = addrrange_make(base, now);
+ fr.dirty_log_mask = mr->dirty_log_mask;
flatview_insert(view, i, &fr);
++i;
base += now;
@@ -194,6 +196,7 @@ static void render_memory_region(FlatView *view,
fr.mr = mr;
fr.offset_in_region = offset_in_region;
fr.addr = addrrange_make(base, remain);
+ fr.dirty_log_mask = mr->dirty_log_mask;
flatview_insert(view, i, &fr);
}
}
@@ -247,9 +250,14 @@ static void memory_region_update_topology(void)
} else if (frold && frnew && flatrange_equal(frold, frnew)) {
/* In both (logging may have changed) */
+ if (frold->dirty_log_mask && !frnew->dirty_log_mask) {
+ cpu_physical_log_stop(frnew->addr.start, frnew->addr.size);
+ } else if (frnew->dirty_log_mask && !frold->dirty_log_mask) {
+ cpu_physical_log_start(frnew->addr.start, frnew->addr.size);
+ }
+
++iold;
++inew;
- /* FIXME: dirty logging */
} else {
/* In new */
@@ -267,7 +275,7 @@ static void memory_region_update_topology(void)
frnew->addr.size,
phys_offset,
region_offset,
- 0);
+ frnew->dirty_log_mask);
++inew;
}
}
@@ -292,6 +300,7 @@ void memory_region_init(MemoryRegion *mr,
memset(&mr->subregions_link, 0, sizeof mr->subregions_link);
QTAILQ_INIT(&mr->coalesced);
mr->name = qemu_strdup(name);
+ mr->dirty_log_mask = 0;
}
static bool memory_region_access_valid(MemoryRegion *mr,
@@ -496,24 +505,35 @@ void memory_region_set_offset(MemoryRegion *mr, target_phys_addr_t offset)
void memory_region_set_log(MemoryRegion *mr, bool log, unsigned client)
{
- /* FIXME */
+ uint8_t mask = 1 << client;
+
+ mr->dirty_log_mask = (mr->dirty_log_mask & ~mask) | (log * mask);
+ memory_region_update_topology();
}
bool memory_region_get_dirty(MemoryRegion *mr, target_phys_addr_t addr,
unsigned client)
{
- /* FIXME */
- return true;
+ assert(mr->has_ram_addr);
+ return cpu_physical_memory_get_dirty(mr->ram_addr + addr, 1 << client);
}
void memory_region_set_dirty(MemoryRegion *mr, target_phys_addr_t addr)
{
- /* FIXME */
+ assert(mr->has_ram_addr);
+ return cpu_physical_memory_set_dirty(mr->ram_addr + addr);
}
void memory_region_sync_dirty_bitmap(MemoryRegion *mr)
{
- /* FIXME */
+ FlatRange *fr;
+
+ FOR_EACH_FLAT_RANGE(fr, ¤t_memory_map) {
+ if (fr->mr == mr) {
+ cpu_physical_sync_dirty_bitmap(fr->addr.start,
+ fr->addr.start + fr->addr.size);
+ }
+ }
}
void memory_region_set_readonly(MemoryRegion *mr, bool readonly)
@@ -524,7 +544,10 @@ void memory_region_set_readonly(MemoryRegion *mr, bool readonly)
void memory_region_reset_dirty(MemoryRegion *mr, target_phys_addr_t addr,
target_phys_addr_t size, unsigned client)
{
- /* FIXME */
+ assert(mr->has_ram_addr);
+ cpu_physical_memory_reset_dirty(mr->ram_addr + addr,
+ mr->ram_addr + addr + size,
+ 1 << client);
}
void *memory_region_get_ram_ptr(MemoryRegion *mr)
diff --git a/memory.h b/memory.h
index a4c94bd..d441bd8 100644
--- a/memory.h
+++ b/memory.h
@@ -99,6 +99,7 @@ struct MemoryRegion {
QTAILQ_ENTRY(MemoryRegion) subregions_link;
QTAILQ_HEAD(coalesced_ranges, CoalescedMemoryRange) coalesced;
const char *name;
+ uint8_t dirty_log_mask;
};
/**
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (2 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 03/23] memory: implement dirty tracking Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:36 ` Paolo Bonzini
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 05/23] Internal interfaces for memory API Avi Kivity
` (19 subsequent siblings)
23 siblings, 1 reply; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Simple implementations of memory routers, for example the Cirrus VGA memory banks
or the 440FX PAM registers can generate adjacent memory regions which are contiguous.
Detect these and merge them; this saves kvm memory slots and shortens lookup times.
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 29 +++++++++++++++++++++++++++++
1 files changed, 29 insertions(+), 0 deletions(-)
diff --git a/memory.c b/memory.c
index d858b47..121f9e1 100644
--- a/memory.c
+++ b/memory.c
@@ -122,6 +122,34 @@ static void flatview_destroy(FlatView *view)
qemu_free(view->ranges);
}
+static bool can_merge(FlatRange *r1, FlatRange *r2)
+{
+ return addrrange_end(r1->addr) == r2->addr.start
+ && r1->mr == r2->mr
+ && r1->offset_in_region + r1->addr.size == r2->offset_in_region
+ && r1->dirty_log_mask == r2->dirty_log_mask;
+}
+
+/* Attempt to simplify a view by merging ajacent ranges */
+static void flatview_simplify(FlatView *view)
+{
+ unsigned i, j;
+
+ i = 0;
+ while (i < view->nr) {
+ j = i + 1;
+ while (j < view->nr
+ && can_merge(&view->ranges[j-1], &view->ranges[j])) {
+ view->ranges[i].addr.size += view->ranges[j].addr.size;
+ ++j;
+ }
+ ++i;
+ memmove(&view->ranges[i], &view->ranges[j],
+ (view->nr - j) * sizeof(view->ranges[j]));
+ view->nr -= j - i;
+ }
+}
+
/* Render a memory region into the global view. Ranges in @view obscure
* ranges in @mr.
*/
@@ -209,6 +237,7 @@ static FlatView generate_memory_topology(MemoryRegion *mr)
flatview_init(&view);
render_memory_region(&view, mr, 0, addrrange_make(0, UINT64_MAX));
+ flatview_simplify(&view);
return view;
}
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region Avi Kivity
@ 2011-07-26 11:36 ` Paolo Bonzini
2011-07-26 11:38 ` Avi Kivity
2011-07-26 15:41 ` Richard Henderson
0 siblings, 2 replies; 31+ messages in thread
From: Paolo Bonzini @ 2011-07-26 11:36 UTC (permalink / raw)
To: Avi Kivity; +Cc: qemu-devel, kvm
On 07/26/2011 01:26 PM, Avi Kivity wrote:
> + while (i < view->nr) {
> + j = i + 1;
> + while (j < view->nr
> + && can_merge(&view->ranges[j-1], &view->ranges[j])) {
> + view->ranges[i].addr.size += view->ranges[j].addr.size;
> + ++j;
> + }
> + ++i;
if (j != i) {
> + memmove(&view->ranges[i], &view->ranges[j],
> + (view->nr - j) * sizeof(view->ranges[j]));
> + view->nr -= j - i;
> + }
}
Paolo
^ permalink raw reply [flat|nested] 31+ messages in thread* Re: [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region
2011-07-26 11:36 ` Paolo Bonzini
@ 2011-07-26 11:38 ` Avi Kivity
2011-07-26 11:51 ` Paolo Bonzini
2011-07-26 15:41 ` Richard Henderson
1 sibling, 1 reply; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:38 UTC (permalink / raw)
To: Paolo Bonzini; +Cc: qemu-devel, kvm
On 07/26/2011 02:36 PM, Paolo Bonzini wrote:
> On 07/26/2011 01:26 PM, Avi Kivity wrote:
>> + while (i < view->nr) {
>> + j = i + 1;
>> + while (j < view->nr
>> + && can_merge(&view->ranges[j-1], &view->ranges[j])) {
>> + view->ranges[i].addr.size += view->ranges[j].addr.size;
>> + ++j;
>> + }
>> + ++i;
>
> if (j != i) {
>
>> + memmove(&view->ranges[i], &view->ranges[j],
>> + (view->nr - j) * sizeof(view->ranges[j]));
>> + view->nr -= j - i;
>> + }
>
> }
Seems to work both ways?
--
error compiling committee.c: too many arguments to function
^ permalink raw reply [flat|nested] 31+ messages in thread* Re: [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region
2011-07-26 11:38 ` Avi Kivity
@ 2011-07-26 11:51 ` Paolo Bonzini
2011-07-26 12:04 ` Avi Kivity
0 siblings, 1 reply; 31+ messages in thread
From: Paolo Bonzini @ 2011-07-26 11:51 UTC (permalink / raw)
To: Avi Kivity; +Cc: qemu-devel, kvm
On 07/26/2011 01:38 PM, Avi Kivity wrote:
>>
>> if (j != i) {
>>
>>> + memmove(&view->ranges[i], &view->ranges[j],
>>> + (view->nr - j) * sizeof(view->ranges[j]));
>>> + view->nr -= j - i;
>>> + }
>>
>> }
>
> Seems to work both ways?
Sure, but you're pointlessly memmove-ing memory over itself. I'm not
sure how many segments a single memory region will usually have, but
it's better to be safe.
Paolo
^ permalink raw reply [flat|nested] 31+ messages in thread* Re: [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region
2011-07-26 11:51 ` Paolo Bonzini
@ 2011-07-26 12:04 ` Avi Kivity
0 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 12:04 UTC (permalink / raw)
To: Paolo Bonzini; +Cc: qemu-devel, kvm
On 07/26/2011 02:51 PM, Paolo Bonzini wrote:
> On 07/26/2011 01:38 PM, Avi Kivity wrote:
>>>
>>> if (j != i) {
>>>
>>>> + memmove(&view->ranges[i], &view->ranges[j],
>>>> + (view->nr - j) * sizeof(view->ranges[j]));
>>>> + view->nr -= j - i;
>>>> + }
>>>
>>> }
>>
>> Seems to work both ways?
>
> Sure, but you're pointlessly memmove-ing memory over itself. I'm not
> sure how many segments a single memory region will usually have, but
> it's better to be safe.
This will never be an issue in practice. We expect to have few
FlatRanges (O(100)), simplify() will be called very rarely (never during
normal operations; a few dozen times during boot; a few during hotplug);
everything is in L1 cache; and the cost will be dwarfed by any calls to
kvm (if enabled).
--
error compiling committee.c: too many arguments to function
^ permalink raw reply [flat|nested] 31+ messages in thread
* Re: [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region
2011-07-26 11:36 ` Paolo Bonzini
2011-07-26 11:38 ` Avi Kivity
@ 2011-07-26 15:41 ` Richard Henderson
2011-07-26 16:04 ` Avi Kivity
1 sibling, 1 reply; 31+ messages in thread
From: Richard Henderson @ 2011-07-26 15:41 UTC (permalink / raw)
To: Paolo Bonzini; +Cc: Avi Kivity, kvm, qemu-devel
On 07/26/2011 04:36 AM, Paolo Bonzini wrote:
> On 07/26/2011 01:26 PM, Avi Kivity wrote:
>> + while (i < view->nr) {
>> + j = i + 1;
>> + while (j < view->nr
>> + && can_merge(&view->ranges[j-1], &view->ranges[j])) {
>> + view->ranges[i].addr.size += view->ranges[j].addr.size;
>> + ++j;
>> + }
>> + ++i;
>
> if (j != i) {
>
>> + memmove(&view->ranges[i], &view->ranges[j],
>> + (view->nr - j) * sizeof(view->ranges[j]));
>> + view->nr -= j - i;
>> + }
>
> }
Err, j = i + 1, and increments. Your conditional will always be true.
r~
^ permalink raw reply [flat|nested] 31+ messages in thread* Re: [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region
2011-07-26 15:41 ` Richard Henderson
@ 2011-07-26 16:04 ` Avi Kivity
0 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 16:04 UTC (permalink / raw)
To: Richard Henderson; +Cc: Paolo Bonzini, qemu-devel, kvm
On 07/26/2011 06:41 PM, Richard Henderson wrote:
> On 07/26/2011 04:36 AM, Paolo Bonzini wrote:
> > On 07/26/2011 01:26 PM, Avi Kivity wrote:
> >> + while (i< view->nr) {
> >> + j = i + 1;
> >> + while (j< view->nr
> >> +&& can_merge(&view->ranges[j-1],&view->ranges[j])) {
> >> + view->ranges[i].addr.size += view->ranges[j].addr.size;
> >> + ++j;
> >> + }
> >> + ++i;
> >
> > if (j != i) {
> >
> >> + memmove(&view->ranges[i],&view->ranges[j],
> >> + (view->nr - j) * sizeof(view->ranges[j]));
> >> + view->nr -= j - i;
> >> + }
> >
> > }
>
> Err, j = i + 1, and increments. Your conditional will always be true.
>
If the while loop body is never executed, then i will equal j (both
incremented once).
--
error compiling committee.c: too many arguments to function
^ permalink raw reply [flat|nested] 31+ messages in thread
* [Qemu-devel] [PATCH v2 05/23] Internal interfaces for memory API
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (3 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 04/23] memory: merge adjacent segments of a single memory region Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 06/23] memory: abstract address space operations Avi Kivity
` (18 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
get_system_memory() provides the root of the memory hierarchy.
This interface is intended to be private between memory.c and exec.c.
If this file is included elsewhere, it should be regarded as a bug (or
TODO item). However, it will be temporarily needed for the conversion
to hierarchical memory routing.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
exec-memory.h | 36 ++++++++++++++++++++++++++++++++++++
memory.c | 7 +++++++
2 files changed, 43 insertions(+), 0 deletions(-)
create mode 100644 exec-memory.h
diff --git a/exec-memory.h b/exec-memory.h
new file mode 100644
index 0000000..aea1b45
--- /dev/null
+++ b/exec-memory.h
@@ -0,0 +1,36 @@
+/*
+ * Internal memory managment interfaces
+ *
+ * Copyright 2011 Red Hat, Inc. and/or its affiliates
+ *
+ * Authors:
+ * Avi Kivity <avi@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2. See
+ * the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef EXEC_MEMORY_H
+#define EXEC_MEMORY_H
+
+/*
+ * Internal interfaces between memory.c/exec.c/vl.c. Do not #include unless
+ * you're one of them.
+ */
+
+#include "memory.h"
+
+#ifndef CONFIG_USER_ONLY
+
+/* Get the root memory region. This interface should only be used temporarily
+ * until a proper bus interface is available.
+ */
+MemoryRegion *get_system_memory(void);
+
+/* Set the root memory region. This region is the system memory map. */
+void set_system_memory_map(MemoryRegion *mr);
+
+#endif
+
+#endif
diff --git a/memory.c b/memory.c
index 121f9e1..fcb612e 100644
--- a/memory.c
+++ b/memory.c
@@ -12,6 +12,7 @@
*/
#include "memory.h"
+#include "exec-memory.h"
#include <assert.h>
typedef struct AddrRange AddrRange;
@@ -703,3 +704,9 @@ void memory_region_del_subregion(MemoryRegion *mr,
QTAILQ_REMOVE(&mr->subregions, subregion, subregions_link);
memory_region_update_topology();
}
+
+void set_system_memory_map(MemoryRegion *mr)
+{
+ root_memory_region = mr;
+ memory_region_update_topology();
+}
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 06/23] memory: abstract address space operations
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (4 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 05/23] Internal interfaces for memory API Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 07/23] memory: rename MemoryRegion::has_ram_addr to ::terminates Avi Kivity
` (17 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Prepare for multiple address space support by abstracting away the details
of registering a memory range with qemu's flat representation into an
AddressSpace object.
Note operations which are memory specific are not abstracted, since they will
never be called on I/O address spaces anyway.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 111 +++++++++++++++++++++++++++++++++++++++++++++-----------------
1 files changed, 81 insertions(+), 30 deletions(-)
diff --git a/memory.c b/memory.c
index fcb612e..bae7765 100644
--- a/memory.c
+++ b/memory.c
@@ -82,12 +82,26 @@ struct FlatView {
unsigned nr_allocated;
};
+typedef struct AddressSpace AddressSpace;
+typedef struct AddressSpaceOps AddressSpaceOps;
+
+/* A system address space - I/O, memory, etc. */
+struct AddressSpace {
+ const AddressSpaceOps *ops;
+ MemoryRegion *root;
+ FlatView current_map;
+};
+
+struct AddressSpaceOps {
+ void (*range_add)(AddressSpace *as, FlatRange *fr);
+ void (*range_del)(AddressSpace *as, FlatRange *fr);
+ void (*log_start)(AddressSpace *as, FlatRange *fr);
+ void (*log_stop)(AddressSpace *as, FlatRange *fr);
+};
+
#define FOR_EACH_FLAT_RANGE(var, view) \
for (var = (view)->ranges; var < (view)->ranges + (view)->nr; ++var)
-static FlatView current_memory_map;
-static MemoryRegion *root_memory_region;
-
static bool flatrange_equal(FlatRange *a, FlatRange *b)
{
return a->mr == b->mr
@@ -151,6 +165,54 @@ static void flatview_simplify(FlatView *view)
}
}
+static void as_memory_range_add(AddressSpace *as, FlatRange *fr)
+{
+ ram_addr_t phys_offset, region_offset;
+
+ phys_offset = fr->mr->ram_addr;
+ region_offset = fr->offset_in_region;
+ /* cpu_register_physical_memory_log() wants region_offset for
+ * mmio, but prefers offseting phys_offset for RAM. Humour it.
+ */
+ if ((phys_offset & ~TARGET_PAGE_MASK) <= IO_MEM_ROM) {
+ phys_offset += region_offset;
+ region_offset = 0;
+ }
+
+ cpu_register_physical_memory_log(fr->addr.start,
+ fr->addr.size,
+ phys_offset,
+ region_offset,
+ fr->dirty_log_mask);
+}
+
+static void as_memory_range_del(AddressSpace *as, FlatRange *fr)
+{
+ cpu_register_physical_memory(fr->addr.start, fr->addr.size,
+ IO_MEM_UNASSIGNED);
+}
+
+static void as_memory_log_start(AddressSpace *as, FlatRange *fr)
+{
+ cpu_physical_log_start(fr->addr.start, fr->addr.size);
+}
+
+static void as_memory_log_stop(AddressSpace *as, FlatRange *fr)
+{
+ cpu_physical_log_stop(fr->addr.start, fr->addr.size);
+}
+
+static const AddressSpaceOps address_space_ops_memory = {
+ .range_add = as_memory_range_add,
+ .range_del = as_memory_range_del,
+ .log_start = as_memory_log_start,
+ .log_stop = as_memory_log_stop,
+};
+
+static AddressSpace address_space_memory = {
+ .ops = &address_space_ops_memory,
+};
+
/* Render a memory region into the global view. Ranges in @view obscure
* ranges in @mr.
*/
@@ -243,13 +305,12 @@ static FlatView generate_memory_topology(MemoryRegion *mr)
return view;
}
-static void memory_region_update_topology(void)
+static void address_space_update_topology(AddressSpace *as)
{
- FlatView old_view = current_memory_map;
- FlatView new_view = generate_memory_topology(root_memory_region);
+ FlatView old_view = as->current_map;
+ FlatView new_view = generate_memory_topology(as->root);
unsigned iold, inew;
FlatRange *frold, *frnew;
- ram_addr_t phys_offset, region_offset;
/* Generate a symmetric difference of the old and new memory maps.
* Kill ranges in the old map, and instantiate ranges in the new map.
@@ -274,16 +335,15 @@ static void memory_region_update_topology(void)
&& !flatrange_equal(frold, frnew)))) {
/* In old, but (not in new, or in new but attributes changed). */
- cpu_register_physical_memory(frold->addr.start, frold->addr.size,
- IO_MEM_UNASSIGNED);
+ as->ops->range_del(as, frold);
++iold;
} else if (frold && frnew && flatrange_equal(frold, frnew)) {
/* In both (logging may have changed) */
if (frold->dirty_log_mask && !frnew->dirty_log_mask) {
- cpu_physical_log_stop(frnew->addr.start, frnew->addr.size);
+ as->ops->log_stop(as, frnew);
} else if (frnew->dirty_log_mask && !frold->dirty_log_mask) {
- cpu_physical_log_start(frnew->addr.start, frnew->addr.size);
+ as->ops->log_start(as, frnew);
}
++iold;
@@ -291,28 +351,19 @@ static void memory_region_update_topology(void)
} else {
/* In new */
- phys_offset = frnew->mr->ram_addr;
- region_offset = frnew->offset_in_region;
- /* cpu_register_physical_memory_log() wants region_offset for
- * mmio, but prefers offseting phys_offset for RAM. Humour it.
- */
- if ((phys_offset & ~TARGET_PAGE_MASK) <= IO_MEM_ROM) {
- phys_offset += region_offset;
- region_offset = 0;
- }
-
- cpu_register_physical_memory_log(frnew->addr.start,
- frnew->addr.size,
- phys_offset,
- region_offset,
- frnew->dirty_log_mask);
+ as->ops->range_add(as, frnew);
++inew;
}
}
- current_memory_map = new_view;
+ as->current_map = new_view;
flatview_destroy(&old_view);
}
+static void memory_region_update_topology(void)
+{
+ address_space_update_topology(&address_space_memory);
+}
+
void memory_region_init(MemoryRegion *mr,
const char *name,
uint64_t size)
@@ -558,7 +609,7 @@ void memory_region_sync_dirty_bitmap(MemoryRegion *mr)
{
FlatRange *fr;
- FOR_EACH_FLAT_RANGE(fr, ¤t_memory_map) {
+ FOR_EACH_FLAT_RANGE(fr, &address_space_memory.current_map) {
if (fr->mr == mr) {
cpu_physical_sync_dirty_bitmap(fr->addr.start,
fr->addr.start + fr->addr.size);
@@ -597,7 +648,7 @@ static void memory_region_update_coalesced_range(MemoryRegion *mr)
CoalescedMemoryRange *cmr;
AddrRange tmp;
- FOR_EACH_FLAT_RANGE(fr, ¤t_memory_map) {
+ FOR_EACH_FLAT_RANGE(fr, &address_space_memory.current_map) {
if (fr->mr == mr) {
qemu_unregister_coalesced_mmio(fr->addr.start, fr->addr.size);
QTAILQ_FOREACH(cmr, &mr->coalesced, link) {
@@ -707,6 +758,6 @@ void memory_region_del_subregion(MemoryRegion *mr,
void set_system_memory_map(MemoryRegion *mr)
{
- root_memory_region = mr;
+ address_space_memory.root = mr;
memory_region_update_topology();
}
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 07/23] memory: rename MemoryRegion::has_ram_addr to ::terminates
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (5 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 06/23] memory: abstract address space operations Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 08/23] memory: late initialization of ram_addr Avi Kivity
` (16 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
I/O regions will not have ram_addrs, so this is a better name.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 18 +++++++++---------
memory.h | 2 +-
2 files changed, 10 insertions(+), 10 deletions(-)
diff --git a/memory.c b/memory.c
index bae7765..9e1a838 100644
--- a/memory.c
+++ b/memory.c
@@ -251,7 +251,7 @@ static void render_memory_region(FlatView *view,
render_memory_region(view, subregion, base, clip);
}
- if (!mr->has_ram_addr) {
+ if (!mr->terminates) {
return;
}
@@ -373,7 +373,7 @@ void memory_region_init(MemoryRegion *mr,
mr->size = size;
mr->addr = 0;
mr->offset = 0;
- mr->has_ram_addr = false;
+ mr->terminates = false;
mr->priority = 0;
mr->may_overlap = false;
mr->alias = NULL;
@@ -528,7 +528,7 @@ void memory_region_init_io(MemoryRegion *mr,
memory_region_init(mr, name, size);
mr->ops = ops;
mr->opaque = opaque;
- mr->has_ram_addr = true;
+ mr->terminates = true;
mr->ram_addr = cpu_register_io_memory(memory_region_read_thunk,
memory_region_write_thunk,
mr,
@@ -541,7 +541,7 @@ void memory_region_init_ram(MemoryRegion *mr,
uint64_t size)
{
memory_region_init(mr, name, size);
- mr->has_ram_addr = true;
+ mr->terminates = true;
mr->ram_addr = qemu_ram_alloc(dev, name, size);
}
@@ -552,7 +552,7 @@ void memory_region_init_ram_ptr(MemoryRegion *mr,
void *ptr)
{
memory_region_init(mr, name, size);
- mr->has_ram_addr = true;
+ mr->terminates = true;
mr->ram_addr = qemu_ram_alloc_from_ptr(dev, name, size, ptr);
}
@@ -595,13 +595,13 @@ void memory_region_set_log(MemoryRegion *mr, bool log, unsigned client)
bool memory_region_get_dirty(MemoryRegion *mr, target_phys_addr_t addr,
unsigned client)
{
- assert(mr->has_ram_addr);
+ assert(mr->terminates);
return cpu_physical_memory_get_dirty(mr->ram_addr + addr, 1 << client);
}
void memory_region_set_dirty(MemoryRegion *mr, target_phys_addr_t addr)
{
- assert(mr->has_ram_addr);
+ assert(mr->terminates);
return cpu_physical_memory_set_dirty(mr->ram_addr + addr);
}
@@ -625,7 +625,7 @@ void memory_region_set_readonly(MemoryRegion *mr, bool readonly)
void memory_region_reset_dirty(MemoryRegion *mr, target_phys_addr_t addr,
target_phys_addr_t size, unsigned client)
{
- assert(mr->has_ram_addr);
+ assert(mr->terminates);
cpu_physical_memory_reset_dirty(mr->ram_addr + addr,
mr->ram_addr + addr + size,
1 << client);
@@ -637,7 +637,7 @@ void *memory_region_get_ram_ptr(MemoryRegion *mr)
return memory_region_get_ram_ptr(mr->alias) + mr->alias_offset;
}
- assert(mr->has_ram_addr);
+ assert(mr->terminates);
return qemu_get_ram_ptr(mr->ram_addr);
}
diff --git a/memory.h b/memory.h
index d441bd8..47d6b9d 100644
--- a/memory.h
+++ b/memory.h
@@ -90,7 +90,7 @@ struct MemoryRegion {
target_phys_addr_t addr;
target_phys_addr_t offset;
ram_addr_t ram_addr;
- bool has_ram_addr;
+ bool terminates;
MemoryRegion *alias;
target_phys_addr_t alias_offset;
unsigned priority;
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 08/23] memory: late initialization of ram_addr
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (6 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 07/23] memory: rename MemoryRegion::has_ram_addr to ::terminates Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 09/23] memory: I/O address space support Avi Kivity
` (15 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
For non-RAM memory regions, we cannot tell whether this is an I/O region
or an MMIO region. Since the qemu backing registration is different for
the two, we have to defer initialization until we know which address
space we are in.
These shenanigans will be removed once the backing registration is unified
with the memory API.
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 24 ++++++++++++++++++++----
memory.h | 1 +
2 files changed, 21 insertions(+), 4 deletions(-)
diff --git a/memory.c b/memory.c
index 9e1a838..e839c9e 100644
--- a/memory.c
+++ b/memory.c
@@ -165,10 +165,14 @@ static void flatview_simplify(FlatView *view)
}
}
+static void memory_region_prepare_ram_addr(MemoryRegion *mr);
+
static void as_memory_range_add(AddressSpace *as, FlatRange *fr)
{
ram_addr_t phys_offset, region_offset;
+ memory_region_prepare_ram_addr(fr->mr);
+
phys_offset = fr->mr->ram_addr;
region_offset = fr->offset_in_region;
/* cpu_register_physical_memory_log() wants region_offset for
@@ -519,6 +523,19 @@ static CPUWriteMemoryFunc * const memory_region_write_thunk[] = {
memory_region_write_thunk_l,
};
+static void memory_region_prepare_ram_addr(MemoryRegion *mr)
+{
+ if (mr->backend_registered) {
+ return;
+ }
+
+ mr->ram_addr = cpu_register_io_memory(memory_region_read_thunk,
+ memory_region_write_thunk,
+ mr,
+ mr->ops->endianness);
+ mr->backend_registered = true;
+}
+
void memory_region_init_io(MemoryRegion *mr,
const MemoryRegionOps *ops,
void *opaque,
@@ -529,10 +546,7 @@ void memory_region_init_io(MemoryRegion *mr,
mr->ops = ops;
mr->opaque = opaque;
mr->terminates = true;
- mr->ram_addr = cpu_register_io_memory(memory_region_read_thunk,
- memory_region_write_thunk,
- mr,
- mr->ops->endianness);
+ mr->backend_registered = false;
}
void memory_region_init_ram(MemoryRegion *mr,
@@ -543,6 +557,7 @@ void memory_region_init_ram(MemoryRegion *mr,
memory_region_init(mr, name, size);
mr->terminates = true;
mr->ram_addr = qemu_ram_alloc(dev, name, size);
+ mr->backend_registered = true;
}
void memory_region_init_ram_ptr(MemoryRegion *mr,
@@ -554,6 +569,7 @@ void memory_region_init_ram_ptr(MemoryRegion *mr,
memory_region_init(mr, name, size);
mr->terminates = true;
mr->ram_addr = qemu_ram_alloc_from_ptr(dev, name, size, ptr);
+ mr->backend_registered = true;
}
void memory_region_init_alias(MemoryRegion *mr,
diff --git a/memory.h b/memory.h
index 47d6b9d..c481038 100644
--- a/memory.h
+++ b/memory.h
@@ -89,6 +89,7 @@ struct MemoryRegion {
uint64_t size;
target_phys_addr_t addr;
target_phys_addr_t offset;
+ bool backend_registered;
ram_addr_t ram_addr;
bool terminates;
MemoryRegion *alias;
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 09/23] memory: I/O address space support
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (7 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 08/23] memory: late initialization of ram_addr Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 10/23] memory: add backward compatibility for old portio registration Avi Kivity
` (14 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Allow registering I/O ports via the same mechanism as mmio ranges.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
exec-memory.h | 3 ++
memory.c | 60 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
memory.h | 2 +
3 files changed, 64 insertions(+), 1 deletions(-)
diff --git a/exec-memory.h b/exec-memory.h
index aea1b45..c439aba 100644
--- a/exec-memory.h
+++ b/exec-memory.h
@@ -31,6 +31,9 @@ MemoryRegion *get_system_memory(void);
/* Set the root memory region. This region is the system memory map. */
void set_system_memory_map(MemoryRegion *mr);
+/* Set the I/O memory region. This region is the I/O memory map. */
+void set_system_io_map(MemoryRegion *mr);
+
#endif
#endif
diff --git a/memory.c b/memory.c
index e839c9e..df0ed0e 100644
--- a/memory.c
+++ b/memory.c
@@ -13,6 +13,7 @@
#include "memory.h"
#include "exec-memory.h"
+#include "ioport.h"
#include <assert.h>
typedef struct AddrRange AddrRange;
@@ -217,6 +218,52 @@ static AddressSpace address_space_memory = {
.ops = &address_space_ops_memory,
};
+static void memory_region_iorange_read(IORange *iorange,
+ uint64_t offset,
+ unsigned width,
+ uint64_t *data)
+{
+ MemoryRegion *mr = container_of(iorange, MemoryRegion, iorange);
+
+ *data = mr->ops->read(mr->opaque, offset, width);
+}
+
+static void memory_region_iorange_write(IORange *iorange,
+ uint64_t offset,
+ unsigned width,
+ uint64_t data)
+{
+ MemoryRegion *mr = container_of(iorange, MemoryRegion, iorange);
+
+ mr->ops->write(mr->opaque, offset, data, width);
+}
+
+static const IORangeOps memory_region_iorange_ops = {
+ .read = memory_region_iorange_read,
+ .write = memory_region_iorange_write,
+};
+
+static void as_io_range_add(AddressSpace *as, FlatRange *fr)
+{
+ iorange_init(&fr->mr->iorange, &memory_region_iorange_ops,
+ fr->addr.start,fr->addr.size);
+ ioport_register(&fr->mr->iorange);
+}
+
+static void as_io_range_del(AddressSpace *as, FlatRange *fr)
+{
+ isa_unassign_ioport(fr->addr.start, fr->addr.size);
+}
+
+static const AddressSpaceOps address_space_ops_io = {
+ .range_add = as_io_range_add,
+ .range_del = as_io_range_del,
+};
+
+static AddressSpace address_space_io = {
+ .ops = &address_space_ops_io,
+};
+
/* Render a memory region into the global view. Ranges in @view obscure
* ranges in @mr.
*/
@@ -365,7 +412,12 @@ static void address_space_update_topology(AddressSpace *as)
static void memory_region_update_topology(void)
{
- address_space_update_topology(&address_space_memory);
+ if (address_space_memory.root) {
+ address_space_update_topology(&address_space_memory);
+ }
+ if (address_space_io.root) {
+ address_space_update_topology(&address_space_io);
+ }
}
void memory_region_init(MemoryRegion *mr,
@@ -777,3 +829,9 @@ void set_system_memory_map(MemoryRegion *mr)
address_space_memory.root = mr;
memory_region_update_topology();
}
+
+void set_system_io_map(MemoryRegion *mr)
+{
+ address_space_io.root = mr;
+ memory_region_update_topology();
+}
diff --git a/memory.h b/memory.h
index c481038..88ba428 100644
--- a/memory.h
+++ b/memory.h
@@ -22,6 +22,7 @@
#include "cpu-common.h"
#include "targphys.h"
#include "qemu-queue.h"
+#include "iorange.h"
typedef struct MemoryRegionOps MemoryRegionOps;
typedef struct MemoryRegion MemoryRegion;
@@ -91,6 +92,7 @@ struct MemoryRegion {
target_phys_addr_t offset;
bool backend_registered;
ram_addr_t ram_addr;
+ IORange iorange;
bool terminates;
MemoryRegion *alias;
target_phys_addr_t alias_offset;
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 10/23] memory: add backward compatibility for old portio registration
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (8 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 09/23] memory: I/O address space support Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 11/23] memory: add backward compatibility for old mmio registration Avi Kivity
` (13 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 32 ++++++++++++++++++++++++++++++++
memory.h | 17 +++++++++++++++++
2 files changed, 49 insertions(+), 0 deletions(-)
diff --git a/memory.c b/memory.c
index df0ed0e..5f2c9ef 100644
--- a/memory.c
+++ b/memory.c
@@ -218,6 +218,21 @@ static AddressSpace address_space_memory = {
.ops = &address_space_ops_memory,
};
+static const MemoryRegionPortio *find_portio(MemoryRegion *mr, uint64_t offset,
+ unsigned width, bool write)
+{
+ const MemoryRegionPortio *mrp;
+
+ for (mrp = mr->ops->old_portio; mrp->size; ++mrp) {
+ if (offset >= mrp->offset && offset < mrp->offset + mrp->len
+ && width == mrp->size
+ && (write ? (bool)mrp->write : (bool)mrp->read)) {
+ return mrp;
+ }
+ }
+ return NULL;
+}
+
static void memory_region_iorange_read(IORange *iorange,
uint64_t offset,
unsigned width,
@@ -225,6 +240,15 @@ static void memory_region_iorange_read(IORange *iorange,
{
MemoryRegion *mr = container_of(iorange, MemoryRegion, iorange);
+ if (mr->ops->old_portio) {
+ const MemoryRegionPortio *mrp = find_portio(mr, offset, width, false);
+
+ *data = ((uint64_t)1 << (width * 8)) - 1;
+ if (mrp) {
+ *data = mrp->read(mr->opaque, offset - mrp->offset);
+ }
+ return;
+ }
*data = mr->ops->read(mr->opaque, offset, width);
}
@@ -235,6 +259,14 @@ static void memory_region_iorange_write(IORange *iorange,
{
MemoryRegion *mr = container_of(iorange, MemoryRegion, iorange);
+ if (mr->ops->old_portio) {
+ const MemoryRegionPortio *mrp = find_portio(mr, offset, width, true);
+
+ if (mrp) {
+ mrp->write(mr->opaque, offset - mrp->offset, data);
+ }
+ return;
+ }
mr->ops->write(mr->opaque, offset, data, width);
}
diff --git a/memory.h b/memory.h
index 88ba428..40ab95a 100644
--- a/memory.h
+++ b/memory.h
@@ -23,9 +23,11 @@
#include "targphys.h"
#include "qemu-queue.h"
#include "iorange.h"
+#include "ioport.h"
typedef struct MemoryRegionOps MemoryRegionOps;
typedef struct MemoryRegion MemoryRegion;
+typedef struct MemoryRegionPortio MemoryRegionPortio;
/* Must match *_DIRTY_FLAGS in cpu-all.h. To be replaced with dynamic
* registration.
@@ -78,6 +80,11 @@ struct MemoryRegionOps {
*/
bool unaligned;
} impl;
+
+ /* If .read and .write are not present, old_portio may be used for
+ * backwards compatibility with old portio registration
+ */
+ const MemoryRegionPortio *old_portio;
};
typedef struct CoalescedMemoryRange CoalescedMemoryRange;
@@ -105,6 +112,16 @@ struct MemoryRegion {
uint8_t dirty_log_mask;
};
+struct MemoryRegionPortio {
+ uint32_t offset;
+ uint32_t len;
+ unsigned size;
+ IOPortReadFunc *read;
+ IOPortWriteFunc *write;
+};
+
+#define PORTIO_END { }
+
/**
* memory_region_init: Initialize a memory region
*
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 11/23] memory: add backward compatibility for old mmio registration
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (9 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 10/23] memory: add backward compatibility for old portio registration Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 12/23] memory: add ioeventfd support Avi Kivity
` (12 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
This eases the transition to the new API.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 10 ++++++++++
memory.h | 10 ++++++++++
2 files changed, 20 insertions(+), 0 deletions(-)
diff --git a/memory.c b/memory.c
index 5f2c9ef..7dd7cac 100644
--- a/memory.c
+++ b/memory.c
@@ -14,6 +14,7 @@
#include "memory.h"
#include "exec-memory.h"
#include "ioport.h"
+#include "bitops.h"
#include <assert.h>
typedef struct AddrRange AddrRange;
@@ -506,6 +507,10 @@ static uint32_t memory_region_read_thunk_n(void *_mr,
return -1U; /* FIXME: better signalling */
}
+ if (!mr->ops->read) {
+ return mr->ops->old_mmio.read[bitops_ffsl(size)](mr->opaque, addr);
+ }
+
/* FIXME: support unaligned access */
access_size_min = mr->ops->impl.min_access_size;
@@ -542,6 +547,11 @@ static void memory_region_write_thunk_n(void *_mr,
return; /* FIXME: better signalling */
}
+ if (!mr->ops->write) {
+ mr->ops->old_mmio.write[bitops_ffsl(size)](mr->opaque, addr, data);
+ return;
+ }
+
/* FIXME: support unaligned access */
access_size_min = mr->ops->impl.min_access_size;
diff --git a/memory.h b/memory.h
index 40ab95a..003c999 100644
--- a/memory.h
+++ b/memory.h
@@ -28,6 +28,7 @@
typedef struct MemoryRegionOps MemoryRegionOps;
typedef struct MemoryRegion MemoryRegion;
typedef struct MemoryRegionPortio MemoryRegionPortio;
+typedef struct MemoryRegionMmio MemoryRegionMmio;
/* Must match *_DIRTY_FLAGS in cpu-all.h. To be replaced with dynamic
* registration.
@@ -36,6 +37,11 @@ typedef struct MemoryRegionPortio MemoryRegionPortio;
#define DIRTY_MEMORY_CODE 1
#define DIRTY_MEMORY_MIGRATION 3
+struct MemoryRegionMmio {
+ CPUReadMemoryFunc *read[3];
+ CPUWriteMemoryFunc *write[3];
+};
+
/*
* Memory region callbacks
*/
@@ -85,6 +91,10 @@ struct MemoryRegionOps {
* backwards compatibility with old portio registration
*/
const MemoryRegionPortio *old_portio;
+ /* If .read and .write are not present, old_mmio may be used for
+ * backwards compatibility with old mmio registration
+ */
+ const MemoryRegionMmio old_mmio;
};
typedef struct CoalescedMemoryRange CoalescedMemoryRange;
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 12/23] memory: add ioeventfd support
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (10 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 11/23] memory: add backward compatibility for old mmio registration Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 13/23] memory: separate building the final memory map into two steps Avi Kivity
` (11 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
As with the rest of the memory API, the caller associates an eventfd
with an address, and the memory API takes care of registering or
unregistering when the address is made visible or invisible to the
guest.
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 224 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
memory.h | 45 +++++++++++++
2 files changed, 269 insertions(+), 0 deletions(-)
diff --git a/memory.c b/memory.c
index 7dd7cac..686bbf2 100644
--- a/memory.c
+++ b/memory.c
@@ -15,6 +15,7 @@
#include "exec-memory.h"
#include "ioport.h"
#include "bitops.h"
+#include "kvm.h"
#include <assert.h>
typedef struct AddrRange AddrRange;
@@ -64,6 +65,50 @@ struct CoalescedMemoryRange {
QTAILQ_ENTRY(CoalescedMemoryRange) link;
};
+struct MemoryRegionIoeventfd {
+ AddrRange addr;
+ bool match_data;
+ uint64_t data;
+ int fd;
+};
+
+static bool memory_region_ioeventfd_before(MemoryRegionIoeventfd a,
+ MemoryRegionIoeventfd b)
+{
+ if (a.addr.start < b.addr.start) {
+ return true;
+ } else if (a.addr.start > b.addr.start) {
+ return false;
+ } else if (a.addr.size < b.addr.size) {
+ return true;
+ } else if (a.addr.size > b.addr.size) {
+ return false;
+ } else if (a.match_data < b.match_data) {
+ return true;
+ } else if (a.match_data > b.match_data) {
+ return false;
+ } else if (a.match_data) {
+ if (a.data < b.data) {
+ return true;
+ } else if (a.data > b.data) {
+ return false;
+ }
+ }
+ if (a.fd < b.fd) {
+ return true;
+ } else if (a.fd > b.fd) {
+ return false;
+ }
+ return false;
+}
+
+static bool memory_region_ioeventfd_equal(MemoryRegionIoeventfd a,
+ MemoryRegionIoeventfd b)
+{
+ return !memory_region_ioeventfd_before(a, b)
+ && !memory_region_ioeventfd_before(b, a);
+}
+
typedef struct FlatRange FlatRange;
typedef struct FlatView FlatView;
@@ -92,6 +137,8 @@ struct AddressSpace {
const AddressSpaceOps *ops;
MemoryRegion *root;
FlatView current_map;
+ int ioeventfd_nb;
+ MemoryRegionIoeventfd *ioeventfds;
};
struct AddressSpaceOps {
@@ -99,6 +146,8 @@ struct AddressSpaceOps {
void (*range_del)(AddressSpace *as, FlatRange *fr);
void (*log_start)(AddressSpace *as, FlatRange *fr);
void (*log_stop)(AddressSpace *as, FlatRange *fr);
+ void (*ioeventfd_add)(AddressSpace *as, MemoryRegionIoeventfd *fd);
+ void (*ioeventfd_del)(AddressSpace *as, MemoryRegionIoeventfd *fd);
};
#define FOR_EACH_FLAT_RANGE(var, view) \
@@ -208,11 +257,35 @@ static void as_memory_log_stop(AddressSpace *as, FlatRange *fr)
cpu_physical_log_stop(fr->addr.start, fr->addr.size);
}
+static void as_memory_ioeventfd_add(AddressSpace *as, MemoryRegionIoeventfd *fd)
+{
+ int r;
+
+ assert(fd->match_data && fd->addr.size == 4);
+
+ r = kvm_set_ioeventfd_mmio_long(fd->fd, fd->addr.start, fd->data, true);
+ if (r < 0) {
+ abort();
+ }
+}
+
+static void as_memory_ioeventfd_del(AddressSpace *as, MemoryRegionIoeventfd *fd)
+{
+ int r;
+
+ r = kvm_set_ioeventfd_mmio_long(fd->fd, fd->addr.start, fd->data, false);
+ if (r < 0) {
+ abort();
+ }
+}
+
static const AddressSpaceOps address_space_ops_memory = {
.range_add = as_memory_range_add,
.range_del = as_memory_range_del,
.log_start = as_memory_log_start,
.log_stop = as_memory_log_stop,
+ .ioeventfd_add = as_memory_ioeventfd_add,
+ .ioeventfd_del = as_memory_ioeventfd_del,
};
static AddressSpace address_space_memory = {
@@ -288,9 +361,33 @@ static void as_io_range_del(AddressSpace *as, FlatRange *fr)
isa_unassign_ioport(fr->addr.start, fr->addr.size);
}
+static void as_io_ioeventfd_add(AddressSpace *as, MemoryRegionIoeventfd *fd)
+{
+ int r;
+
+ assert(fd->match_data && fd->addr.size == 2);
+
+ r = kvm_set_ioeventfd_pio_word(fd->fd, fd->addr.start, fd->data, true);
+ if (r < 0) {
+ abort();
+ }
+}
+
+static void as_io_ioeventfd_del(AddressSpace *as, MemoryRegionIoeventfd *fd)
+{
+ int r;
+
+ r = kvm_set_ioeventfd_pio_word(fd->fd, fd->addr.start, fd->data, false);
+ if (r < 0) {
+ abort();
+ }
+}
+
static const AddressSpaceOps address_space_ops_io = {
.range_add = as_io_range_add,
.range_del = as_io_range_del,
+ .ioeventfd_add = as_io_ioeventfd_add,
+ .ioeventfd_del = as_io_ioeventfd_del,
};
static AddressSpace address_space_io = {
@@ -389,6 +486,69 @@ static FlatView generate_memory_topology(MemoryRegion *mr)
return view;
}
+static void address_space_add_del_ioeventfds(AddressSpace *as,
+ MemoryRegionIoeventfd *fds_new,
+ unsigned fds_new_nb,
+ MemoryRegionIoeventfd *fds_old,
+ unsigned fds_old_nb)
+{
+ unsigned iold, inew;
+
+ /* Generate a symmetric difference of the old and new fd sets, adding
+ * and deleting as necessary.
+ */
+
+ iold = inew = 0;
+ while (iold < fds_old_nb || inew < fds_new_nb) {
+ if (iold < fds_old_nb
+ && (inew == fds_new_nb
+ || memory_region_ioeventfd_before(fds_old[iold],
+ fds_new[inew]))) {
+ as->ops->ioeventfd_del(as, &fds_old[iold]);
+ ++iold;
+ } else if (inew < fds_new_nb
+ && (iold == fds_old_nb
+ || memory_region_ioeventfd_before(fds_new[inew],
+ fds_old[iold]))) {
+ as->ops->ioeventfd_add(as, &fds_new[inew]);
+ ++inew;
+ } else {
+ ++iold;
+ ++inew;
+ }
+ }
+}
+
+static void address_space_update_ioeventfds(AddressSpace *as)
+{
+ FlatRange *fr;
+ unsigned ioeventfd_nb = 0;
+ MemoryRegionIoeventfd *ioeventfds = NULL;
+ AddrRange tmp;
+ unsigned i;
+
+ FOR_EACH_FLAT_RANGE(fr, &as->current_map) {
+ for (i = 0; i < fr->mr->ioeventfd_nb; ++i) {
+ tmp = addrrange_shift(fr->mr->ioeventfds[i].addr,
+ fr->addr.start - fr->offset_in_region);
+ if (addrrange_intersects(fr->addr, tmp)) {
+ ++ioeventfd_nb;
+ ioeventfds = qemu_realloc(ioeventfds,
+ ioeventfd_nb * sizeof(*ioeventfds));
+ ioeventfds[ioeventfd_nb-1] = fr->mr->ioeventfds[i];
+ ioeventfds[ioeventfd_nb-1].addr = tmp;
+ }
+ }
+ }
+
+ address_space_add_del_ioeventfds(as, ioeventfds, ioeventfd_nb,
+ as->ioeventfds, as->ioeventfd_nb);
+
+ qemu_free(as->ioeventfds);
+ as->ioeventfds = ioeventfds;
+ as->ioeventfd_nb = ioeventfd_nb;
+}
+
static void address_space_update_topology(AddressSpace *as)
{
FlatView old_view = as->current_map;
@@ -441,6 +601,7 @@ static void address_space_update_topology(AddressSpace *as)
}
as->current_map = new_view;
flatview_destroy(&old_view);
+ address_space_update_ioeventfds(as);
}
static void memory_region_update_topology(void)
@@ -471,6 +632,8 @@ void memory_region_init(MemoryRegion *mr,
QTAILQ_INIT(&mr->coalesced);
mr->name = qemu_strdup(name);
mr->dirty_log_mask = 0;
+ mr->ioeventfd_nb = 0;
+ mr->ioeventfds = NULL;
}
static bool memory_region_access_valid(MemoryRegion *mr,
@@ -682,6 +845,7 @@ void memory_region_destroy(MemoryRegion *mr)
assert(QTAILQ_EMPTY(&mr->subregions));
memory_region_clear_coalescing(mr);
qemu_free((char *)mr->name);
+ qemu_free(mr->ioeventfds);
}
uint64_t memory_region_size(MemoryRegion *mr)
@@ -803,6 +967,66 @@ void memory_region_clear_coalescing(MemoryRegion *mr)
memory_region_update_coalesced_range(mr);
}
+void memory_region_add_eventfd(MemoryRegion *mr,
+ target_phys_addr_t addr,
+ unsigned size,
+ bool match_data,
+ uint64_t data,
+ int fd)
+{
+ MemoryRegionIoeventfd mrfd = {
+ .addr.start = addr,
+ .addr.size = size,
+ .match_data = match_data,
+ .data = data,
+ .fd = fd,
+ };
+ unsigned i;
+
+ for (i = 0; i < mr->ioeventfd_nb; ++i) {
+ if (memory_region_ioeventfd_before(mrfd, mr->ioeventfds[i])) {
+ break;
+ }
+ }
+ ++mr->ioeventfd_nb;
+ mr->ioeventfds = qemu_realloc(mr->ioeventfds,
+ sizeof(*mr->ioeventfds) * mr->ioeventfd_nb);
+ memmove(&mr->ioeventfds[i+1], &mr->ioeventfds[i],
+ sizeof(*mr->ioeventfds) * (mr->ioeventfd_nb-1 - i));
+ mr->ioeventfds[i] = mrfd;
+ memory_region_update_topology();
+}
+
+void memory_region_del_eventfd(MemoryRegion *mr,
+ target_phys_addr_t addr,
+ unsigned size,
+ bool match_data,
+ uint64_t data,
+ int fd)
+{
+ MemoryRegionIoeventfd mrfd = {
+ .addr.start = addr,
+ .addr.size = size,
+ .match_data = match_data,
+ .data = data,
+ .fd = fd,
+ };
+ unsigned i;
+
+ for (i = 0; i < mr->ioeventfd_nb; ++i) {
+ if (memory_region_ioeventfd_equal(mrfd, mr->ioeventfds[i])) {
+ break;
+ }
+ }
+ assert(i != mr->ioeventfd_nb);
+ memmove(&mr->ioeventfds[i], &mr->ioeventfds[i+1],
+ sizeof(*mr->ioeventfds) * (mr->ioeventfd_nb - (i+1)));
+ --mr->ioeventfd_nb;
+ mr->ioeventfds = qemu_realloc(mr->ioeventfds,
+ sizeof(*mr->ioeventfds)*mr->ioeventfd_nb + 1);
+ memory_region_update_topology();
+}
+
static void memory_region_add_subregion_common(MemoryRegion *mr,
target_phys_addr_t offset,
MemoryRegion *subregion)
diff --git a/memory.h b/memory.h
index 003c999..c280a39 100644
--- a/memory.h
+++ b/memory.h
@@ -98,6 +98,7 @@ struct MemoryRegionOps {
};
typedef struct CoalescedMemoryRange CoalescedMemoryRange;
+typedef struct MemoryRegionIoeventfd MemoryRegionIoeventfd;
struct MemoryRegion {
/* All fields are private - violators will be prosecuted */
@@ -120,6 +121,8 @@ struct MemoryRegion {
QTAILQ_HEAD(coalesced_ranges, CoalescedMemoryRange) coalesced;
const char *name;
uint8_t dirty_log_mask;
+ unsigned ioeventfd_nb;
+ MemoryRegionIoeventfd *ioeventfds;
};
struct MemoryRegionPortio {
@@ -364,6 +367,48 @@ void memory_region_add_coalescing(MemoryRegion *mr,
void memory_region_clear_coalescing(MemoryRegion *mr);
/**
+ * memory_region_add_eventfd: Request an eventfd to be triggered when a word
+ * is written to a location.
+ *
+ * Marks a word in an IO region (initialized with memory_region_init_io())
+ * as a trigger for an eventfd event. The I/O callback will not be called.
+ * The caller must be prepared to handle failure (hat is, take the required
+ * action if the callback _is_ called).
+ *
+ * @mr: the memory region being updated.
+ * @addr: the address within @mr that is to be monitored
+ * @size: the size of the access to trigger the eventfd
+ * @match_data: whether to match against @data, instead of just @addr
+ * @data: the data to match against the guest write
+ * @fd: the eventfd to be triggered when @addr, @size, and @data all match.
+ **/
+void memory_region_add_eventfd(MemoryRegion *mr,
+ target_phys_addr_t addr,
+ unsigned size,
+ bool match_data,
+ uint64_t data,
+ int fd);
+
+/**
+ * memory_region_del_eventfd: Cancel and eventfd.
+ *
+ * Cancels an eventfd trigger request by a previous memory_region_add_eventfd()
+ * call.
+ *
+ * @mr: the memory region being updated.
+ * @addr: the address within @mr that is to be monitored
+ * @size: the size of the access to trigger the eventfd
+ * @match_data: whether to match against @data, instead of just @addr
+ * @data: the data to match against the guest write
+ * @fd: the eventfd to be triggered when @addr, @size, and @data all match.
+ */
+void memory_region_del_eventfd(MemoryRegion *mr,
+ target_phys_addr_t addr,
+ unsigned size,
+ bool match_data,
+ uint64_t data,
+ int fd);
+/**
* memory_region_add_subregion: Add a sub-region to a container.
*
* Adds a sub-region at @offset. The sub-region may not overlap with other
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 13/23] memory: separate building the final memory map into two steps
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (11 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 12/23] memory: add ioeventfd support Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 14/23] memory: transaction API Avi Kivity
` (10 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Instead of adding and deleting regions in one pass, do a delete
pass followed by an add pass. This fixes the following case:
from:
0x0000-0x0fff ram (a1)
0x1000-0x1fff mmio (a2)
0x2000-0x2fff ram (a3)
to:
0x0000-0x2fff ram (b1)
The single pass algorithm removed a1, added b2, then removed a2 and a3,
which caused the wrong memory map to be built. The two pass algorithm
removes a1, a2, and a3, then adds b1.
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 38 +++++++++++++++++++++++++++++---------
1 files changed, 29 insertions(+), 9 deletions(-)
diff --git a/memory.c b/memory.c
index 686bbf2..7a5670e 100644
--- a/memory.c
+++ b/memory.c
@@ -549,10 +549,11 @@ static void address_space_update_ioeventfds(AddressSpace *as)
as->ioeventfd_nb = ioeventfd_nb;
}
-static void address_space_update_topology(AddressSpace *as)
+static void address_space_update_topology_pass(AddressSpace *as,
+ FlatView old_view,
+ FlatView new_view,
+ bool adding)
{
- FlatView old_view = as->current_map;
- FlatView new_view = generate_memory_topology(as->root);
unsigned iold, inew;
FlatRange *frold, *frnew;
@@ -579,15 +580,20 @@ static void address_space_update_topology(AddressSpace *as)
&& !flatrange_equal(frold, frnew)))) {
/* In old, but (not in new, or in new but attributes changed). */
- as->ops->range_del(as, frold);
+ if (!adding) {
+ as->ops->range_del(as, frold);
+ }
+
++iold;
} else if (frold && frnew && flatrange_equal(frold, frnew)) {
/* In both (logging may have changed) */
- if (frold->dirty_log_mask && !frnew->dirty_log_mask) {
- as->ops->log_stop(as, frnew);
- } else if (frnew->dirty_log_mask && !frold->dirty_log_mask) {
- as->ops->log_start(as, frnew);
+ if (adding) {
+ if (frold->dirty_log_mask && !frnew->dirty_log_mask) {
+ as->ops->log_stop(as, frnew);
+ } else if (frnew->dirty_log_mask && !frold->dirty_log_mask) {
+ as->ops->log_start(as, frnew);
+ }
}
++iold;
@@ -595,10 +601,24 @@ static void address_space_update_topology(AddressSpace *as)
} else {
/* In new */
- as->ops->range_add(as, frnew);
+ if (adding) {
+ as->ops->range_add(as, frnew);
+ }
+
++inew;
}
}
+}
+
+
+static void address_space_update_topology(AddressSpace *as)
+{
+ FlatView old_view = as->current_map;
+ FlatView new_view = generate_memory_topology(as->root);
+
+ address_space_update_topology_pass(as, old_view, new_view, false);
+ address_space_update_topology_pass(as, old_view, new_view, true);
+
as->current_map = new_view;
flatview_destroy(&old_view);
address_space_update_ioeventfds(as);
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 14/23] memory: transaction API
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (12 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 13/23] memory: separate building the final memory map into two steps Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 15/23] exec.c: initialize memory map Avi Kivity
` (9 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Allow changes to the memory hierarchy to be accumulated and
made visible all at once. This reduces computational effort,
especially when an accelerator (e.g. kvm) is involved.
Useful when a single register update causes multiple changes
to an address space.
Signed-off-by: Avi Kivity <avi@redhat.com>
---
memory.c | 18 ++++++++++++++++++
memory.h | 8 ++++++++
2 files changed, 26 insertions(+), 0 deletions(-)
diff --git a/memory.c b/memory.c
index 7a5670e..5c6e63d 100644
--- a/memory.c
+++ b/memory.c
@@ -18,6 +18,8 @@
#include "kvm.h"
#include <assert.h>
+unsigned memory_region_transaction_depth = 0;
+
typedef struct AddrRange AddrRange;
struct AddrRange {
@@ -626,6 +628,10 @@ static void address_space_update_topology(AddressSpace *as)
static void memory_region_update_topology(void)
{
+ if (memory_region_transaction_depth) {
+ return;
+ }
+
if (address_space_memory.root) {
address_space_update_topology(&address_space_memory);
}
@@ -634,6 +640,18 @@ static void memory_region_update_topology(void)
}
}
+void memory_region_transaction_begin(void)
+{
+ ++memory_region_transaction_depth;
+}
+
+void memory_region_transaction_commit(void)
+{
+ assert(memory_region_transaction_depth);
+ --memory_region_transaction_depth;
+ memory_region_update_topology();
+}
+
void memory_region_init(MemoryRegion *mr,
const char *name,
uint64_t size)
diff --git a/memory.h b/memory.h
index c280a39..4e518b2 100644
--- a/memory.h
+++ b/memory.h
@@ -456,6 +456,14 @@ void memory_region_add_subregion_overlap(MemoryRegion *mr,
void memory_region_del_subregion(MemoryRegion *mr,
MemoryRegion *subregion);
+/* Start a transaction; changes will be accumulated and made visible only
+ * when the transaction ends.
+ */
+void memory_region_transaction_begin(void);
+/* Commit a transaction and make changes visible to the guest.
+ */
+void memory_region_transaction_commit(void);
+
#endif
#endif
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 15/23] exec.c: initialize memory map
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (13 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 14/23] memory: transaction API Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 16/23] ioport: register ranges by byte aligned addresses always Avi Kivity
` (8 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Allocate the root memory region and initialize it.
Signed-off-by: Avi Kivity <avi@redhat.com>
---
exec.c | 19 +++++++++++++++++++
1 files changed, 19 insertions(+), 0 deletions(-)
diff --git a/exec.c b/exec.c
index 2160ded..d51502f 100644
--- a/exec.c
+++ b/exec.c
@@ -33,6 +33,8 @@
#include "kvm.h"
#include "hw/xen.h"
#include "qemu-timer.h"
+#include "memory.h"
+#include "exec-memory.h"
#if defined(CONFIG_USER_ONLY)
#include <qemu.h>
#if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
@@ -109,6 +111,9 @@ int phys_ram_fd;
static int in_migration;
RAMList ram_list = { .blocks = QLIST_HEAD_INITIALIZER(ram_list) };
+
+static MemoryRegion *system_memory;
+
#endif
CPUState *first_cpu;
@@ -197,6 +202,7 @@ typedef struct PhysPageDesc {
static void *l1_phys_map[P_L1_SIZE];
static void io_mem_init(void);
+static void memory_map_init(void);
/* io memory support */
CPUWriteMemoryFunc *io_mem_write[IO_MEM_NB_ENTRIES][4];
@@ -571,6 +577,7 @@ void cpu_exec_init_all(unsigned long tb_size)
code_gen_ptr = code_gen_buffer;
page_init();
#if !defined(CONFIG_USER_ONLY)
+ memory_map_init();
io_mem_init();
#endif
#if !defined(CONFIG_USER_ONLY) || !defined(CONFIG_USE_GUEST_BASE)
@@ -3807,6 +3814,18 @@ static void io_mem_init(void)
DEVICE_NATIVE_ENDIAN);
}
+static void memory_map_init(void)
+{
+ system_memory = qemu_malloc(sizeof(*system_memory));
+ memory_region_init(system_memory, "system", UINT64_MAX);
+ set_system_memory_map(system_memory);
+}
+
+MemoryRegion *get_system_memory(void)
+{
+ return system_memory;
+}
+
#endif /* !defined(CONFIG_USER_ONLY) */
/* physical memory access (slow version, mainly for debug) */
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 16/23] ioport: register ranges by byte aligned addresses always
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (14 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 15/23] exec.c: initialize memory map Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 17/23] pc: grab system_memory Avi Kivity
` (7 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
The I/O port space is byte addressable, even for word and long accesses.
An example is the VMware svga card, which has long ports on offsets 0,
1, and 2.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
ioport.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/ioport.c b/ioport.c
index 0d2611d..a32483b 100644
--- a/ioport.c
+++ b/ioport.c
@@ -146,7 +146,7 @@ int register_ioport_read(pio_addr_t start, int length, int size,
hw_error("register_ioport_read: invalid size");
return -1;
}
- for(i = start; i < start + length; i += size) {
+ for(i = start; i < start + length; ++i) {
ioport_read_table[bsize][i] = func;
if (ioport_opaque[i] != NULL && ioport_opaque[i] != opaque)
hw_error("register_ioport_read: invalid opaque for address 0x%x",
@@ -166,7 +166,7 @@ int register_ioport_write(pio_addr_t start, int length, int size,
hw_error("register_ioport_write: invalid size");
return -1;
}
- for(i = start; i < start + length; i += size) {
+ for(i = start; i < start + length; ++i) {
ioport_write_table[bsize][i] = func;
if (ioport_opaque[i] != NULL && ioport_opaque[i] != opaque)
hw_error("register_ioport_write: invalid opaque for address 0x%x",
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 17/23] pc: grab system_memory
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (15 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 16/23] ioport: register ranges by byte aligned addresses always Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 18/23] pc: convert pc_memory_init() to memory API Avi Kivity
` (6 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
While eventually this should come from the machine initialization function,
take a short cut to avoid converting all machines now.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
hw/pc.c | 3 ++-
hw/pc.h | 4 +++-
hw/pc_piix.c | 8 +++++++-
3 files changed, 12 insertions(+), 3 deletions(-)
diff --git a/hw/pc.c b/hw/pc.c
index a3e8539..369566a 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -957,7 +957,8 @@ void pc_cpus_init(const char *cpu_model)
}
}
-void pc_memory_init(const char *kernel_filename,
+void pc_memory_init(MemoryRegion *system_memory,
+ const char *kernel_filename,
const char *kernel_cmdline,
const char *initrd_filename,
ram_addr_t below_4g_mem_size,
diff --git a/hw/pc.h b/hw/pc.h
index 6d5730b..fa57583 100644
--- a/hw/pc.h
+++ b/hw/pc.h
@@ -6,6 +6,7 @@
#include "isa.h"
#include "fdc.h"
#include "net.h"
+#include "memory.h"
/* PC-style peripherals (also used by other machines). */
@@ -129,7 +130,8 @@ void pc_cmos_set_s3_resume(void *opaque, int irq, int level);
void pc_acpi_smi_interrupt(void *opaque, int irq, int level);
void pc_cpus_init(const char *cpu_model);
-void pc_memory_init(const char *kernel_filename,
+void pc_memory_init(MemoryRegion *system_memory,
+ const char *kernel_filename,
const char *kernel_cmdline,
const char *initrd_filename,
ram_addr_t below_4g_mem_size,
diff --git a/hw/pc_piix.c b/hw/pc_piix.c
index c5c16b4..d83854c 100644
--- a/hw/pc_piix.c
+++ b/hw/pc_piix.c
@@ -39,6 +39,8 @@
#include "blockdev.h"
#include "smbus.h"
#include "xen.h"
+#include "memory.h"
+#include "exec-memory.h"
#ifdef CONFIG_XEN
# include <xen/hvm/hvm_info_table.h>
#endif
@@ -89,6 +91,9 @@ static void pc_init1(ram_addr_t ram_size,
DriveInfo *hd[MAX_IDE_BUS * MAX_IDE_DEVS];
BusState *idebus[MAX_IDE_BUS];
ISADevice *rtc_state;
+ MemoryRegion *system_memory;
+
+ system_memory = get_system_memory();
pc_cpus_init(cpu_model);
@@ -106,7 +111,8 @@ static void pc_init1(ram_addr_t ram_size,
/* allocate ram and load rom/bios */
if (!xen_enabled()) {
- pc_memory_init(kernel_filename, kernel_cmdline, initrd_filename,
+ pc_memory_init(system_memory,
+ kernel_filename, kernel_cmdline, initrd_filename,
below_4g_mem_size, above_4g_mem_size);
}
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 18/23] pc: convert pc_memory_init() to memory API
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (16 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 17/23] pc: grab system_memory Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 19/23] pc: move global memory map out of pc_init1() and into its callers Avi Kivity
` (5 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
hw/pc.c | 59 ++++++++++++++++++++++++++++++++++++++++-------------------
hw/pc.h | 1 +
2 files changed, 41 insertions(+), 19 deletions(-)
diff --git a/hw/pc.c b/hw/pc.c
index 369566a..1c9d89a 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -41,6 +41,7 @@
#include "sysemu.h"
#include "blockdev.h"
#include "ui/qemu-spice.h"
+#include "memory.h"
/* output Bochs bios info messages */
//#define DEBUG_BIOS
@@ -966,22 +967,30 @@ void pc_memory_init(MemoryRegion *system_memory,
{
char *filename;
int ret, linux_boot, i;
- ram_addr_t ram_addr, bios_offset, option_rom_offset;
+ MemoryRegion *ram, *bios, *isa_bios, *option_rom_mr;
+ MemoryRegion *ram_below_4g, *ram_above_4g;
int bios_size, isa_bios_size;
void *fw_cfg;
linux_boot = (kernel_filename != NULL);
- /* allocate RAM */
- ram_addr = qemu_ram_alloc(NULL, "pc.ram",
- below_4g_mem_size + above_4g_mem_size);
- cpu_register_physical_memory(0, 0xa0000, ram_addr);
- cpu_register_physical_memory(0x100000,
- below_4g_mem_size - 0x100000,
- ram_addr + 0x100000);
+ /* Allocate RAM. We allocate it as a single memory region and use
+ * aliases to address portions of it, mostly for backwards compatiblity
+ * with older qemus that used qemu_ram_alloc().
+ */
+ ram = qemu_malloc(sizeof(*ram));
+ memory_region_init_ram(ram, NULL, "pc.ram",
+ below_4g_mem_size + above_4g_mem_size);
+ ram_below_4g = qemu_malloc(sizeof(*ram_below_4g));
+ memory_region_init_alias(ram_below_4g, "ram-below-4g", ram,
+ 0, below_4g_mem_size);
+ memory_region_add_subregion(system_memory, 0, ram_below_4g);
if (above_4g_mem_size > 0) {
- cpu_register_physical_memory(0x100000000ULL, above_4g_mem_size,
- ram_addr + below_4g_mem_size);
+ ram_above_4g = qemu_malloc(sizeof(*ram_above_4g));
+ memory_region_init_alias(ram_above_4g, "ram-above-4g", ram,
+ below_4g_mem_size, above_4g_mem_size);
+ memory_region_add_subregion(system_memory, 0x100000000ULL,
+ ram_above_4g);
}
/* BIOS load */
@@ -997,7 +1006,9 @@ void pc_memory_init(MemoryRegion *system_memory,
(bios_size % 65536) != 0) {
goto bios_error;
}
- bios_offset = qemu_ram_alloc(NULL, "pc.bios", bios_size);
+ bios = qemu_malloc(sizeof(*bios));
+ memory_region_init_ram(bios, NULL, "pc.bios", bios_size);
+ memory_region_set_readonly(bios, true);
ret = rom_add_file_fixed(bios_name, (uint32_t)(-bios_size), -1);
if (ret != 0) {
bios_error:
@@ -1011,16 +1022,26 @@ void pc_memory_init(MemoryRegion *system_memory,
isa_bios_size = bios_size;
if (isa_bios_size > (128 * 1024))
isa_bios_size = 128 * 1024;
- cpu_register_physical_memory(0x100000 - isa_bios_size,
- isa_bios_size,
- (bios_offset + bios_size - isa_bios_size) | IO_MEM_ROM);
-
- option_rom_offset = qemu_ram_alloc(NULL, "pc.rom", PC_ROM_SIZE);
- cpu_register_physical_memory(PC_ROM_MIN_VGA, PC_ROM_SIZE, option_rom_offset);
+ isa_bios = qemu_malloc(sizeof(*isa_bios));
+ memory_region_init_alias(isa_bios, "isa-bios", bios,
+ bios_size - isa_bios_size, isa_bios_size);
+ memory_region_add_subregion_overlap(system_memory,
+ 0x100000 - isa_bios_size,
+ isa_bios,
+ 1);
+ memory_region_set_readonly(isa_bios, true);
+
+ option_rom_mr = qemu_malloc(sizeof(*option_rom_mr));
+ memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE);
+ memory_region_add_subregion_overlap(system_memory,
+ PC_ROM_MIN_VGA,
+ option_rom_mr,
+ 1);
/* map all the bios at the top of memory */
- cpu_register_physical_memory((uint32_t)(-bios_size),
- bios_size, bios_offset | IO_MEM_ROM);
+ memory_region_add_subregion(system_memory,
+ (uint32_t)(-bios_size),
+ bios);
fw_cfg = bochs_bios_init();
rom_set_fw(fw_cfg);
diff --git a/hw/pc.h b/hw/pc.h
index fa57583..40684f4 100644
--- a/hw/pc.h
+++ b/hw/pc.h
@@ -2,6 +2,7 @@
#define HW_PC_H
#include "qemu-common.h"
+#include "memory.h"
#include "ioport.h"
#include "isa.h"
#include "fdc.h"
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 19/23] pc: move global memory map out of pc_init1() and into its callers
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (17 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 18/23] pc: convert pc_memory_init() to memory API Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 20/23] pci: pass address space to pci bus when created Avi Kivity
` (4 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Signed-off-by: Avi Kivity <avi@redhat.com>
---
hw/pc_piix.c | 15 ++++++++-------
1 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/hw/pc_piix.c b/hw/pc_piix.c
index d83854c..f2d0476 100644
--- a/hw/pc_piix.c
+++ b/hw/pc_piix.c
@@ -68,7 +68,8 @@ static void ioapic_init(IsaIrqState *isa_irq_state)
}
/* PC hardware initialisation */
-static void pc_init1(ram_addr_t ram_size,
+static void pc_init1(MemoryRegion *system_memory,
+ ram_addr_t ram_size,
const char *boot_device,
const char *kernel_filename,
const char *kernel_cmdline,
@@ -91,9 +92,6 @@ static void pc_init1(ram_addr_t ram_size,
DriveInfo *hd[MAX_IDE_BUS * MAX_IDE_DEVS];
BusState *idebus[MAX_IDE_BUS];
ISADevice *rtc_state;
- MemoryRegion *system_memory;
-
- system_memory = get_system_memory();
pc_cpus_init(cpu_model);
@@ -214,7 +212,8 @@ static void pc_init_pci(ram_addr_t ram_size,
const char *initrd_filename,
const char *cpu_model)
{
- pc_init1(ram_size, boot_device,
+ pc_init1(get_system_memory(),
+ ram_size, boot_device,
kernel_filename, kernel_cmdline,
initrd_filename, cpu_model, 1, 1);
}
@@ -226,7 +225,8 @@ static void pc_init_pci_no_kvmclock(ram_addr_t ram_size,
const char *initrd_filename,
const char *cpu_model)
{
- pc_init1(ram_size, boot_device,
+ pc_init1(get_system_memory(),
+ ram_size, boot_device,
kernel_filename, kernel_cmdline,
initrd_filename, cpu_model, 1, 0);
}
@@ -240,7 +240,8 @@ static void pc_init_isa(ram_addr_t ram_size,
{
if (cpu_model == NULL)
cpu_model = "486";
- pc_init1(ram_size, boot_device,
+ pc_init1(get_system_memory(),
+ ram_size, boot_device,
kernel_filename, kernel_cmdline,
initrd_filename, cpu_model, 0, 1);
}
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 20/23] pci: pass address space to pci bus when created
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (18 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 19/23] pc: move global memory map out of pc_init1() and into its callers Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 21/23] pci: add MemoryRegion based BAR management API Avi Kivity
` (3 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
This is now done sloppily, via get_system_memory(). Eventually callers
will be converted to stop using that.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
hw/apb_pci.c | 2 ++
hw/bonito.c | 4 +++-
hw/grackle_pci.c | 5 +++--
hw/gt64xxx.c | 4 +++-
hw/pc.h | 4 +++-
hw/pc_piix.c | 3 ++-
hw/pci.c | 16 +++++++++++-----
hw/pci.h | 12 +++++++++---
hw/pci_host.h | 1 +
hw/pci_internals.h | 1 +
hw/piix_pci.c | 13 +++++++++----
hw/ppc4xx_pci.c | 5 ++++-
hw/ppc_mac.h | 9 ++++++---
hw/ppc_newworld.c | 5 +++--
hw/ppc_oldworld.c | 3 ++-
hw/ppc_prep.c | 3 ++-
hw/ppce500_pci.c | 6 +++++-
hw/prep_pci.c | 5 +++--
hw/prep_pci.h | 3 ++-
hw/sh_pci.c | 4 +++-
hw/unin_pci.c | 10 ++++++----
hw/versatile_pci.c | 2 ++
22 files changed, 85 insertions(+), 35 deletions(-)
diff --git a/hw/apb_pci.c b/hw/apb_pci.c
index 974c87a..8b9939c 100644
--- a/hw/apb_pci.c
+++ b/hw/apb_pci.c
@@ -34,6 +34,7 @@
#include "rwhandler.h"
#include "apb_pci.h"
#include "sysemu.h"
+#include "exec-memory.h"
/* debug APB */
//#define DEBUG_APB
@@ -346,6 +347,7 @@ PCIBus *pci_apb_init(target_phys_addr_t special_base,
d->bus = pci_register_bus(&d->busdev.qdev, "pci",
pci_apb_set_irq, pci_pbm_map_irq, d,
+ get_system_memory(),
0, 32);
pci_bus_set_mem_base(d->bus, mem_base);
diff --git a/hw/bonito.c b/hw/bonito.c
index e8c57a3..5f62dda 100644
--- a/hw/bonito.c
+++ b/hw/bonito.c
@@ -42,6 +42,7 @@
#include "mips.h"
#include "pci_host.h"
#include "sysemu.h"
+#include "exec-memory.h"
//#define DEBUG_BONITO
@@ -773,7 +774,8 @@ PCIBus *bonito_init(qemu_irq *pic)
dev = qdev_create(NULL, "Bonito-pcihost");
pcihost = FROM_SYSBUS(BonitoState, sysbus_from_qdev(dev));
b = pci_register_bus(&pcihost->busdev.qdev, "pci", pci_bonito_set_irq,
- pci_bonito_map_irq, pic, 0x28, 32);
+ pci_bonito_map_irq, pic, get_system_memory(),
+ 0x28, 32);
pcihost->bus = b;
qdev_init_nofail(dev);
diff --git a/hw/grackle_pci.c b/hw/grackle_pci.c
index cee07e0..da67cf9 100644
--- a/hw/grackle_pci.c
+++ b/hw/grackle_pci.c
@@ -61,7 +61,8 @@ static void pci_grackle_reset(void *opaque)
{
}
-PCIBus *pci_grackle_init(uint32_t base, qemu_irq *pic)
+PCIBus *pci_grackle_init(uint32_t base, qemu_irq *pic,
+ MemoryRegion *address_space)
{
DeviceState *dev;
SysBusDevice *s;
@@ -74,7 +75,7 @@ PCIBus *pci_grackle_init(uint32_t base, qemu_irq *pic)
d->host_state.bus = pci_register_bus(&d->busdev.qdev, "pci",
pci_grackle_set_irq,
pci_grackle_map_irq,
- pic, 0, 4);
+ pic, address_space, 0, 4);
pci_create_simple(d->host_state.bus, 0, "grackle");
diff --git a/hw/gt64xxx.c b/hw/gt64xxx.c
index 8e1f6a0..65e63dd 100644
--- a/hw/gt64xxx.c
+++ b/hw/gt64xxx.c
@@ -27,6 +27,7 @@
#include "pci.h"
#include "pci_host.h"
#include "pc.h"
+#include "exec-memory.h"
//#define DEBUG
@@ -1092,7 +1093,8 @@ PCIBus *gt64120_register(qemu_irq *pic)
d = FROM_SYSBUS(GT64120State, s);
d->pci.bus = pci_register_bus(&d->busdev.qdev, "pci",
gt64120_pci_set_irq, gt64120_pci_map_irq,
- pic, PCI_DEVFN(18, 0), 4);
+ pic, get_system_memory(),
+ PCI_DEVFN(18, 0), 4);
d->ISD_handle = cpu_register_io_memory(gt64120_read, gt64120_write, d,
DEVICE_NATIVE_ENDIAN);
diff --git a/hw/pc.h b/hw/pc.h
index 40684f4..a2de0fe 100644
--- a/hw/pc.h
+++ b/hw/pc.h
@@ -178,7 +178,9 @@ int pcspk_audio_init(qemu_irq *pic);
struct PCII440FXState;
typedef struct PCII440FXState PCII440FXState;
-PCIBus *i440fx_init(PCII440FXState **pi440fx_state, int *piix_devfn, qemu_irq *pic, ram_addr_t ram_size);
+PCIBus *i440fx_init(PCII440FXState **pi440fx_state, int *piix_devfn,
+ qemu_irq *pic, MemoryRegion *address_space,
+ ram_addr_t ram_size);
void i440fx_init_memory_mappings(PCII440FXState *d);
/* piix4.c */
diff --git a/hw/pc_piix.c b/hw/pc_piix.c
index f2d0476..2b9c2b1 100644
--- a/hw/pc_piix.c
+++ b/hw/pc_piix.c
@@ -128,7 +128,8 @@ static void pc_init1(MemoryRegion *system_memory,
isa_irq = qemu_allocate_irqs(isa_irq_handler, isa_irq_state, 24);
if (pci_enabled) {
- pci_bus = i440fx_init(&i440fx_state, &piix3_devfn, isa_irq, ram_size);
+ pci_bus = i440fx_init(&i440fx_state, &piix3_devfn, isa_irq,
+ system_memory, ram_size);
} else {
pci_bus = NULL;
i440fx_state = NULL;
diff --git a/hw/pci.c b/hw/pci.c
index b904a4e..cf16f3b 100644
--- a/hw/pci.c
+++ b/hw/pci.c
@@ -263,11 +263,14 @@ int pci_find_domain(const PCIBus *bus)
}
void pci_bus_new_inplace(PCIBus *bus, DeviceState *parent,
- const char *name, uint8_t devfn_min)
+ const char *name,
+ MemoryRegion *address_space,
+ uint8_t devfn_min)
{
qbus_create_inplace(&bus->qbus, &pci_bus_info, parent, name);
assert(PCI_FUNC(devfn_min) == 0);
bus->devfn_min = devfn_min;
+ bus->address_space = address_space;
/* host bridge */
QLIST_INIT(&bus->child);
@@ -276,13 +279,14 @@ void pci_bus_new_inplace(PCIBus *bus, DeviceState *parent,
vmstate_register(NULL, -1, &vmstate_pcibus, bus);
}
-PCIBus *pci_bus_new(DeviceState *parent, const char *name, uint8_t devfn_min)
+PCIBus *pci_bus_new(DeviceState *parent, const char *name,
+ MemoryRegion *address_space, uint8_t devfn_min)
{
PCIBus *bus;
bus = qemu_mallocz(sizeof(*bus));
bus->qbus.qdev_allocated = 1;
- pci_bus_new_inplace(bus, parent, name, devfn_min);
+ pci_bus_new_inplace(bus, parent, name, address_space, devfn_min);
return bus;
}
@@ -310,11 +314,13 @@ void pci_bus_set_mem_base(PCIBus *bus, target_phys_addr_t base)
PCIBus *pci_register_bus(DeviceState *parent, const char *name,
pci_set_irq_fn set_irq, pci_map_irq_fn map_irq,
- void *irq_opaque, uint8_t devfn_min, int nirq)
+ void *irq_opaque,
+ MemoryRegion *address_space,
+ uint8_t devfn_min, int nirq)
{
PCIBus *bus;
- bus = pci_bus_new(parent, name, devfn_min);
+ bus = pci_bus_new(parent, name, address_space, devfn_min);
pci_bus_irqs(bus, set_irq, map_irq, irq_opaque, nirq);
return bus;
}
diff --git a/hw/pci.h b/hw/pci.h
index c220745..cfeb042 100644
--- a/hw/pci.h
+++ b/hw/pci.h
@@ -5,6 +5,7 @@
#include "qobject.h"
#include "qdev.h"
+#include "memory.h"
/* PCI includes legacy ISA access. */
#include "isa.h"
@@ -233,15 +234,20 @@ typedef enum {
typedef int (*pci_hotplug_fn)(DeviceState *qdev, PCIDevice *pci_dev,
PCIHotplugState state);
void pci_bus_new_inplace(PCIBus *bus, DeviceState *parent,
- const char *name, uint8_t devfn_min);
-PCIBus *pci_bus_new(DeviceState *parent, const char *name, uint8_t devfn_min);
+ const char *name,
+ MemoryRegion *address_space,
+ uint8_t devfn_min);
+PCIBus *pci_bus_new(DeviceState *parent, const char *name,
+ MemoryRegion *address_space, uint8_t devfn_min);
void pci_bus_irqs(PCIBus *bus, pci_set_irq_fn set_irq, pci_map_irq_fn map_irq,
void *irq_opaque, int nirq);
int pci_bus_get_irq_level(PCIBus *bus, int irq_num);
void pci_bus_hotplug(PCIBus *bus, pci_hotplug_fn hotplug, DeviceState *dev);
PCIBus *pci_register_bus(DeviceState *parent, const char *name,
pci_set_irq_fn set_irq, pci_map_irq_fn map_irq,
- void *irq_opaque, uint8_t devfn_min, int nirq);
+ void *irq_opaque,
+ MemoryRegion *address_space,
+ uint8_t devfn_min, int nirq);
void pci_device_reset(PCIDevice *dev);
void pci_bus_reset(PCIBus *bus);
diff --git a/hw/pci_host.h b/hw/pci_host.h
index 0a58595..05dcb66 100644
--- a/hw/pci_host.h
+++ b/hw/pci_host.h
@@ -35,6 +35,7 @@ struct PCIHostState {
SysBusDevice busdev;
ReadWriteHandler conf_handler;
ReadWriteHandler data_handler;
+ MemoryRegion *address_space;
uint32_t config_reg;
PCIBus *bus;
};
diff --git a/hw/pci_internals.h b/hw/pci_internals.h
index fbe1866..c3a463a 100644
--- a/hw/pci_internals.h
+++ b/hw/pci_internals.h
@@ -25,6 +25,7 @@ struct PCIBus {
PCIDevice *devices[PCI_SLOT_MAX * PCI_FUNC_MAX];
PCIDevice *parent_dev;
target_phys_addr_t mem_base;
+ MemoryRegion *address_space;
QLIST_HEAD(, PCIBus) child; /* this will be replaced by qdev later */
QLIST_ENTRY(PCIBus) sibling;/* this will be replaced by qdev later */
diff --git a/hw/piix_pci.c b/hw/piix_pci.c
index d08b31a..80d6665 100644
--- a/hw/piix_pci.c
+++ b/hw/piix_pci.c
@@ -241,7 +241,9 @@ static int i440fx_initfn(PCIDevice *dev)
static PCIBus *i440fx_common_init(const char *device_name,
PCII440FXState **pi440fx_state,
int *piix3_devfn,
- qemu_irq *pic, ram_addr_t ram_size)
+ qemu_irq *pic,
+ MemoryRegion *address_space,
+ ram_addr_t ram_size)
{
DeviceState *dev;
PCIBus *b;
@@ -251,7 +253,8 @@ static PCIBus *i440fx_common_init(const char *device_name,
dev = qdev_create(NULL, "i440FX-pcihost");
s = FROM_SYSBUS(I440FXState, sysbus_from_qdev(dev));
- b = pci_bus_new(&s->busdev.qdev, NULL, 0);
+ s->address_space = address_space;
+ b = pci_bus_new(&s->busdev.qdev, NULL, s->address_space, 0);
s->bus = b;
qdev_init_nofail(dev);
@@ -288,11 +291,13 @@ static PCIBus *i440fx_common_init(const char *device_name,
}
PCIBus *i440fx_init(PCII440FXState **pi440fx_state, int *piix3_devfn,
- qemu_irq *pic, ram_addr_t ram_size)
+ qemu_irq *pic, MemoryRegion *address_space,
+ ram_addr_t ram_size)
{
PCIBus *b;
- b = i440fx_common_init("i440FX", pi440fx_state, piix3_devfn, pic, ram_size);
+ b = i440fx_common_init("i440FX", pi440fx_state, piix3_devfn, pic,
+ address_space, ram_size);
return b;
}
diff --git a/hw/ppc4xx_pci.c b/hw/ppc4xx_pci.c
index 299473c..15c24f6 100644
--- a/hw/ppc4xx_pci.c
+++ b/hw/ppc4xx_pci.c
@@ -24,6 +24,7 @@
#include "ppc4xx.h"
#include "pci.h"
#include "pci_host.h"
+#include "exec-memory.h"
#undef DEBUG
#ifdef DEBUG
@@ -345,7 +346,9 @@ PCIBus *ppc4xx_pci_init(CPUState *env, qemu_irq pci_irqs[4],
controller->pci_state.bus = pci_register_bus(NULL, "pci",
ppc4xx_pci_set_irq,
ppc4xx_pci_map_irq,
- pci_irqs, 0, 4);
+ pci_irqs,
+ get_system_memory(),
+ 0, 4);
controller->pci_dev = pci_register_device(controller->pci_state.bus,
"host bridge", sizeof(PCIDevice),
diff --git a/hw/ppc_mac.h b/hw/ppc_mac.h
index 68dade7..6fad20a 100644
--- a/hw/ppc_mac.h
+++ b/hw/ppc_mac.h
@@ -25,6 +25,8 @@
#if !defined(__PPC_MAC_H__)
#define __PPC_MAC_H__
+#include "memory.h"
+
/* SMP is not enabled, for now */
#define MAX_CPUS 1
@@ -52,11 +54,12 @@ qemu_irq *heathrow_pic_init(int *pmem_index,
int nb_cpus, qemu_irq **irqs);
/* Grackle PCI */
-PCIBus *pci_grackle_init(uint32_t base, qemu_irq *pic);
+PCIBus *pci_grackle_init(uint32_t base, qemu_irq *pic,
+ MemoryRegion *address_space);
/* UniNorth PCI */
-PCIBus *pci_pmac_init(qemu_irq *pic);
-PCIBus *pci_pmac_u3_init(qemu_irq *pic);
+PCIBus *pci_pmac_init(qemu_irq *pic, MemoryRegion *address_space);
+PCIBus *pci_pmac_u3_init(qemu_irq *pic, MemoryRegion *address_space);
/* Mac NVRAM */
typedef struct MacIONVRAMState MacIONVRAMState;
diff --git a/hw/ppc_newworld.c b/hw/ppc_newworld.c
index 5bce709..2c0fae8 100644
--- a/hw/ppc_newworld.c
+++ b/hw/ppc_newworld.c
@@ -67,6 +67,7 @@
#include "kvm_ppc.h"
#include "hw/usb.h"
#include "blockdev.h"
+#include "exec-memory.h"
#define MAX_IDE_BUS 2
#define CFG_ADDR 0xf0000510
@@ -317,10 +318,10 @@ static void ppc_core99_init (ram_addr_t ram_size,
pic = openpic_init(NULL, &pic_mem_index, smp_cpus, openpic_irqs, NULL);
if (PPC_INPUT(env) == PPC_FLAGS_INPUT_970) {
/* 970 gets a U3 bus */
- pci_bus = pci_pmac_u3_init(pic);
+ pci_bus = pci_pmac_u3_init(pic, get_system_memory());
machine_arch = ARCH_MAC99_U3;
} else {
- pci_bus = pci_pmac_init(pic);
+ pci_bus = pci_pmac_init(pic, get_system_memory());
machine_arch = ARCH_MAC99;
}
/* init basic PC hardware */
diff --git a/hw/ppc_oldworld.c b/hw/ppc_oldworld.c
index 20cd8e1..585afd6 100644
--- a/hw/ppc_oldworld.c
+++ b/hw/ppc_oldworld.c
@@ -43,6 +43,7 @@
#include "kvm.h"
#include "kvm_ppc.h"
#include "blockdev.h"
+#include "exec-memory.h"
#define MAX_IDE_BUS 2
#define CFG_ADDR 0xf0000510
@@ -233,7 +234,7 @@ static void ppc_heathrow_init (ram_addr_t ram_size,
hw_error("Only 6xx bus is supported on heathrow machine\n");
}
pic = heathrow_pic_init(&pic_mem_index, 1, heathrow_irqs);
- pci_bus = pci_grackle_init(0xfec00000, pic);
+ pci_bus = pci_grackle_init(0xfec00000, pic, get_system_memory());
pci_vga_init(pci_bus);
escc_mem_index = escc_init(0x80013000, pic[0x0f], pic[0x10], serial_hds[0],
diff --git a/hw/ppc_prep.c b/hw/ppc_prep.c
index 0e9cfc2..91ebe07 100644
--- a/hw/ppc_prep.c
+++ b/hw/ppc_prep.c
@@ -38,6 +38,7 @@
#include "loader.h"
#include "mc146818rtc.h"
#include "blockdev.h"
+#include "exec-memory.h"
//#define HARD_DEBUG_PPC_IO
//#define DEBUG_PPC_IO
@@ -648,7 +649,7 @@ static void ppc_prep_init (ram_addr_t ram_size,
hw_error("Only 6xx bus is supported on PREP machine\n");
}
i8259 = i8259_init(first_cpu->irq_inputs[PPC6xx_INPUT_INT]);
- pci_bus = pci_prep_init(i8259);
+ pci_bus = pci_prep_init(i8259, get_system_memory());
/* Hmm, prep has no pci-isa bridge ??? */
isa_bus_new(NULL);
isa_bus_irqs(i8259);
diff --git a/hw/ppce500_pci.c b/hw/ppce500_pci.c
index fc11af4..1344539 100644
--- a/hw/ppce500_pci.c
+++ b/hw/ppce500_pci.c
@@ -274,12 +274,15 @@ static void e500_pci_map(SysBusDevice *dev, target_phys_addr_t base)
s->reg);
}
+#include "exec-memory.h"
+
static int e500_pcihost_initfn(SysBusDevice *dev)
{
PCIHostState *h;
PPCE500PCIState *s;
PCIBus *b;
int i;
+ MemoryRegion *address_space = get_system_memory();
h = FROM_SYSBUS(PCIHostState, sysbus_from_qdev(dev));
s = DO_UPCAST(PPCE500PCIState, pci_state, h);
@@ -289,7 +292,8 @@ static int e500_pcihost_initfn(SysBusDevice *dev)
}
b = pci_register_bus(&s->pci_state.busdev.qdev, NULL, mpc85xx_pci_set_irq,
- mpc85xx_pci_map_irq, s->irq, PCI_DEVFN(0x11, 0), 4);
+ mpc85xx_pci_map_irq, s->irq, address_space,
+ PCI_DEVFN(0x11, 0), 4);
s->pci_state.bus = b;
pci_create_simple(b, 0, "e500-host-bridge");
diff --git a/hw/prep_pci.c b/hw/prep_pci.c
index f88b825..da02f0e 100644
--- a/hw/prep_pci.c
+++ b/hw/prep_pci.c
@@ -110,7 +110,7 @@ static void prep_set_irq(void *opaque, int irq_num, int level)
qemu_set_irq(pic[(irq_num & 1) ? 11 : 9] , level);
}
-PCIBus *pci_prep_init(qemu_irq *pic)
+PCIBus *pci_prep_init(qemu_irq *pic, MemoryRegion *address_space)
{
PREPPCIState *s;
PCIDevice *d;
@@ -118,7 +118,8 @@ PCIBus *pci_prep_init(qemu_irq *pic)
s = qemu_mallocz(sizeof(PREPPCIState));
s->bus = pci_register_bus(NULL, "pci",
- prep_set_irq, prep_map_irq, pic, 0, 4);
+ prep_set_irq, prep_map_irq, pic,
+ address_space, 0, 4);
pci_host_conf_register_ioport(0xcf8, s);
diff --git a/hw/prep_pci.h b/hw/prep_pci.h
index cd68512..a27368b 100644
--- a/hw/prep_pci.h
+++ b/hw/prep_pci.h
@@ -2,7 +2,8 @@
#define QEMU_PREP_PCI_H
#include "qemu-common.h"
+#include "memory.h"
-PCIBus *pci_prep_init(qemu_irq *pic);
+PCIBus *pci_prep_init(qemu_irq *pic, MemoryRegion *address_space);
#endif
diff --git a/hw/sh_pci.c b/hw/sh_pci.c
index a076cf2..0ef93a0 100644
--- a/hw/sh_pci.c
+++ b/hw/sh_pci.c
@@ -26,6 +26,7 @@
#include "pci.h"
#include "pci_host.h"
#include "bswap.h"
+#include "exec-memory.h"
typedef struct SHPCIState {
SysBusDevice busdev;
@@ -127,7 +128,8 @@ static int sh_pci_init_device(SysBusDevice *dev)
}
s->bus = pci_register_bus(&s->busdev.qdev, "pci",
sh_pci_set_irq, sh_pci_map_irq,
- s->irq, PCI_DEVFN(0, 0), 4);
+ s->irq, get_system_memory(),
+ PCI_DEVFN(0, 0), 4);
s->memconfig = cpu_register_io_memory(sh_pci_reg.r, sh_pci_reg.w,
s, DEVICE_NATIVE_ENDIAN);
sysbus_init_mmio_cb(dev, 0x224, sh_pci_map);
diff --git a/hw/unin_pci.c b/hw/unin_pci.c
index d364daa..b499523 100644
--- a/hw/unin_pci.c
+++ b/hw/unin_pci.c
@@ -201,7 +201,7 @@ static int pci_unin_internal_init_device(SysBusDevice *dev)
return 0;
}
-PCIBus *pci_pmac_init(qemu_irq *pic)
+PCIBus *pci_pmac_init(qemu_irq *pic, MemoryRegion *address_space)
{
DeviceState *dev;
SysBusDevice *s;
@@ -215,7 +215,8 @@ PCIBus *pci_pmac_init(qemu_irq *pic)
d = FROM_SYSBUS(UNINState, s);
d->host_state.bus = pci_register_bus(&d->busdev.qdev, "pci",
pci_unin_set_irq, pci_unin_map_irq,
- pic, PCI_DEVFN(11, 0), 4);
+ pic, address_space,
+ PCI_DEVFN(11, 0), 4);
#if 0
pci_create_simple(d->host_state.bus, PCI_DEVFN(11, 0), "uni-north");
@@ -252,7 +253,7 @@ PCIBus *pci_pmac_init(qemu_irq *pic)
return d->host_state.bus;
}
-PCIBus *pci_pmac_u3_init(qemu_irq *pic)
+PCIBus *pci_pmac_u3_init(qemu_irq *pic, MemoryRegion *address_space)
{
DeviceState *dev;
SysBusDevice *s;
@@ -267,7 +268,8 @@ PCIBus *pci_pmac_u3_init(qemu_irq *pic)
d->host_state.bus = pci_register_bus(&d->busdev.qdev, "pci",
pci_unin_set_irq, pci_unin_map_irq,
- pic, PCI_DEVFN(11, 0), 4);
+ pic, address_space,
+ PCI_DEVFN(11, 0), 4);
sysbus_mmio_map(s, 0, 0xf0800000);
sysbus_mmio_map(s, 1, 0xf0c00000);
diff --git a/hw/versatile_pci.c b/hw/versatile_pci.c
index 290a900..cffe387 100644
--- a/hw/versatile_pci.c
+++ b/hw/versatile_pci.c
@@ -10,6 +10,7 @@
#include "sysbus.h"
#include "pci.h"
#include "pci_host.h"
+#include "exec-memory.h"
typedef struct {
SysBusDevice busdev;
@@ -111,6 +112,7 @@ static int pci_vpb_init(SysBusDevice *dev)
}
bus = pci_register_bus(&dev->qdev, "pci",
pci_vpb_set_irq, pci_vpb_map_irq, s->irq,
+ get_system_memory(),
PCI_DEVFN(11, 0), 4);
/* ??? Register memory space. */
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 21/23] pci: add MemoryRegion based BAR management API
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (19 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 20/23] pci: pass address space to pci bus when created Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 22/23] sysbus: add MemoryRegion based memory " Avi Kivity
` (2 subsequent siblings)
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Allow registering a BAR using a MemoryRegion. Once all users are converted,
pci_register_bar() and pci_register_bar_simple() will be removed.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
hw/pci.c | 47 +++++++++++++++++++++++++++++++++++++++--------
hw/pci.h | 3 +++
2 files changed, 42 insertions(+), 8 deletions(-)
diff --git a/hw/pci.c b/hw/pci.c
index cf16f3b..36db58b 100644
--- a/hw/pci.c
+++ b/hw/pci.c
@@ -844,10 +844,15 @@ static void pci_unregister_io_regions(PCIDevice *pci_dev)
if (r->type == PCI_BASE_ADDRESS_SPACE_IO) {
isa_unassign_ioport(r->addr, r->filtered_size);
} else {
- cpu_register_physical_memory(pci_to_cpu_addr(pci_dev->bus,
- r->addr),
- r->filtered_size,
- IO_MEM_UNASSIGNED);
+ if (r->memory) {
+ memory_region_del_subregion(pci_dev->bus->address_space,
+ r->memory);
+ } else {
+ cpu_register_physical_memory(pci_to_cpu_addr(pci_dev->bus,
+ r->addr),
+ r->filtered_size,
+ IO_MEM_UNASSIGNED);
+ }
}
}
}
@@ -893,6 +898,7 @@ void pci_register_bar(PCIDevice *pci_dev, int region_num,
r->type = type;
r->map_func = map_func;
r->ram_addr = IO_MEM_UNASSIGNED;
+ r->memory = NULL;
wmask = ~(size - 1);
addr = pci_bar(pci_dev, region_num);
@@ -918,6 +924,16 @@ static void pci_simple_bar_mapfunc(PCIDevice *pci_dev, int region_num,
pci_dev->io_regions[region_num].ram_addr);
}
+static void pci_simple_bar_mapfunc_region(PCIDevice *pci_dev, int region_num,
+ pcibus_t addr, pcibus_t size,
+ int type)
+{
+ memory_region_add_subregion_overlap(pci_dev->bus->address_space,
+ addr,
+ pci_dev->io_regions[region_num].memory,
+ 1);
+}
+
void pci_register_bar_simple(PCIDevice *pci_dev, int region_num,
pcibus_t size, uint8_t attr, ram_addr_t ram_addr)
{
@@ -927,6 +943,15 @@ void pci_register_bar_simple(PCIDevice *pci_dev, int region_num,
pci_dev->io_regions[region_num].ram_addr = ram_addr;
}
+void pci_register_bar_region(PCIDevice *pci_dev, int region_num,
+ uint8_t attr, MemoryRegion *memory)
+{
+ pci_register_bar(pci_dev, region_num, memory_region_size(memory),
+ PCI_BASE_ADDRESS_SPACE_MEMORY | attr,
+ pci_simple_bar_mapfunc_region);
+ pci_dev->io_regions[region_num].memory = memory;
+}
+
static void pci_bridge_filter(PCIDevice *d, pcibus_t *addr, pcibus_t *size,
uint8_t type)
{
@@ -1065,10 +1090,16 @@ static void pci_update_mappings(PCIDevice *d)
isa_unassign_ioport(r->addr, r->filtered_size);
}
} else {
- cpu_register_physical_memory(pci_to_cpu_addr(d->bus, r->addr),
- r->filtered_size,
- IO_MEM_UNASSIGNED);
- qemu_unregister_coalesced_mmio(r->addr, r->filtered_size);
+ if (r->memory) {
+ memory_region_del_subregion(d->bus->address_space,
+ r->memory);
+ } else {
+ cpu_register_physical_memory(pci_to_cpu_addr(d->bus,
+ r->addr),
+ r->filtered_size,
+ IO_MEM_UNASSIGNED);
+ qemu_unregister_coalesced_mmio(r->addr, r->filtered_size);
+ }
}
}
r->addr = new_addr;
diff --git a/hw/pci.h b/hw/pci.h
index cfeb042..c51156d 100644
--- a/hw/pci.h
+++ b/hw/pci.h
@@ -94,6 +94,7 @@ typedef struct PCIIORegion {
uint8_t type;
PCIMapIORegionFunc *map_func;
ram_addr_t ram_addr;
+ MemoryRegion *memory;
} PCIIORegion;
#define PCI_ROM_SLOT 6
@@ -204,6 +205,8 @@ void pci_register_bar(PCIDevice *pci_dev, int region_num,
PCIMapIORegionFunc *map_func);
void pci_register_bar_simple(PCIDevice *pci_dev, int region_num,
pcibus_t size, uint8_t attr, ram_addr_t ram_addr);
+void pci_register_bar_region(PCIDevice *pci_dev, int region_num,
+ uint8_t attr, MemoryRegion *memory);
int pci_add_capability(PCIDevice *pdev, uint8_t cap_id,
uint8_t offset, uint8_t size);
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 22/23] sysbus: add MemoryRegion based memory management API
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (20 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 21/23] pci: add MemoryRegion based BAR management API Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 23/23] usb-ohci: convert to MemoryRegion Avi Kivity
2011-07-29 14:18 ` [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Anthony Liguori
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Allow registering sysbus device memory using a MemoryRegion. Once all users
are converted, sysbus_init_mmio() and sysbus_init_mmio_cb() will be removed.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
hw/sysbus.c | 27 ++++++++++++++++++++++++---
hw/sysbus.h | 3 +++
2 files changed, 27 insertions(+), 3 deletions(-)
diff --git a/hw/sysbus.c b/hw/sysbus.c
index 2e22be7..ea442ac 100644
--- a/hw/sysbus.c
+++ b/hw/sysbus.c
@@ -19,6 +19,7 @@
#include "sysbus.h"
#include "monitor.h"
+#include "exec-memory.h"
static void sysbus_dev_print(Monitor *mon, DeviceState *dev, int indent);
static char *sysbus_get_fw_dev_path(DeviceState *dev);
@@ -49,11 +50,20 @@ void sysbus_mmio_map(SysBusDevice *dev, int n, target_phys_addr_t addr)
}
if (dev->mmio[n].addr != (target_phys_addr_t)-1) {
/* Unregister previous mapping. */
- cpu_register_physical_memory(dev->mmio[n].addr, dev->mmio[n].size,
- IO_MEM_UNASSIGNED);
+ if (dev->mmio[n].memory) {
+ memory_region_del_subregion(get_system_memory(),
+ dev->mmio[n].memory);
+ } else {
+ cpu_register_physical_memory(dev->mmio[n].addr, dev->mmio[n].size,
+ IO_MEM_UNASSIGNED);
+ }
}
dev->mmio[n].addr = addr;
- if (dev->mmio[n].cb) {
+ if (dev->mmio[n].memory) {
+ memory_region_add_subregion(get_system_memory(),
+ addr,
+ dev->mmio[n].memory);
+ } else if (dev->mmio[n].cb) {
dev->mmio[n].cb(dev, addr);
} else {
cpu_register_physical_memory(addr, dev->mmio[n].size,
@@ -107,6 +117,17 @@ void sysbus_init_mmio_cb(SysBusDevice *dev, target_phys_addr_t size,
dev->mmio[n].cb = cb;
}
+void sysbus_init_mmio_region(SysBusDevice *dev, MemoryRegion *memory)
+{
+ int n;
+
+ assert(dev->num_mmio < QDEV_MAX_MMIO);
+ n = dev->num_mmio++;
+ dev->mmio[n].addr = -1;
+ dev->mmio[n].size = memory_region_size(memory);
+ dev->mmio[n].memory = memory;
+}
+
void sysbus_init_ioports(SysBusDevice *dev, pio_addr_t ioport, pio_addr_t size)
{
pio_addr_t i;
diff --git a/hw/sysbus.h b/hw/sysbus.h
index 4e8cb16..5f62e2d 100644
--- a/hw/sysbus.h
+++ b/hw/sysbus.h
@@ -4,6 +4,7 @@
/* Devices attached directly to the main system bus. */
#include "qdev.h"
+#include "memory.h"
#define QDEV_MAX_MMIO 32
#define QDEV_MAX_PIO 32
@@ -23,6 +24,7 @@ struct SysBusDevice {
target_phys_addr_t size;
mmio_mapfunc cb;
ram_addr_t iofunc;
+ MemoryRegion *memory;
} mmio[QDEV_MAX_MMIO];
int num_pio;
pio_addr_t pio[QDEV_MAX_PIO];
@@ -46,6 +48,7 @@ void sysbus_init_mmio(SysBusDevice *dev, target_phys_addr_t size,
ram_addr_t iofunc);
void sysbus_init_mmio_cb(SysBusDevice *dev, target_phys_addr_t size,
mmio_mapfunc cb);
+void sysbus_init_mmio_region(SysBusDevice *dev, MemoryRegion *memory);
void sysbus_init_irq(SysBusDevice *dev, qemu_irq *p);
void sysbus_pass_irq(SysBusDevice *dev, SysBusDevice *target);
void sysbus_init_ioports(SysBusDevice *dev, pio_addr_t ioport, pio_addr_t size);
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* [Qemu-devel] [PATCH v2 23/23] usb-ohci: convert to MemoryRegion
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (21 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 22/23] sysbus: add MemoryRegion based memory " Avi Kivity
@ 2011-07-26 11:26 ` Avi Kivity
2011-07-29 14:18 ` [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Anthony Liguori
23 siblings, 0 replies; 31+ messages in thread
From: Avi Kivity @ 2011-07-26 11:26 UTC (permalink / raw)
To: Anthony Liguori, qemu-devel; +Cc: kvm
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
---
hw/usb-ohci.c | 42 +++++++++++++++++-------------------------
1 files changed, 17 insertions(+), 25 deletions(-)
diff --git a/hw/usb-ohci.c b/hw/usb-ohci.c
index 8491d59..337b250 100644
--- a/hw/usb-ohci.c
+++ b/hw/usb-ohci.c
@@ -62,7 +62,7 @@ typedef struct OHCIPort {
typedef struct {
USBBus bus;
qemu_irq irq;
- int mem;
+ MemoryRegion mem;
int num_ports;
const char *name;
@@ -1440,13 +1440,13 @@ static void ohci_port_set_status(OHCIState *ohci, int portnum, uint32_t val)
return;
}
-static uint32_t ohci_mem_read(void *ptr, target_phys_addr_t addr)
+static uint64_t ohci_mem_read(void *opaque,
+ target_phys_addr_t addr,
+ unsigned size)
{
- OHCIState *ohci = ptr;
+ OHCIState *ohci = opaque;
uint32_t retval;
- addr &= 0xff;
-
/* Only aligned reads are allowed on OHCI */
if (addr & 3) {
fprintf(stderr, "usb-ohci: Mis-aligned read\n");
@@ -1563,11 +1563,12 @@ static uint32_t ohci_mem_read(void *ptr, target_phys_addr_t addr)
return retval;
}
-static void ohci_mem_write(void *ptr, target_phys_addr_t addr, uint32_t val)
+static void ohci_mem_write(void *opaque,
+ target_phys_addr_t addr,
+ uint64_t val,
+ unsigned size)
{
- OHCIState *ohci = ptr;
-
- addr &= 0xff;
+ OHCIState *ohci = opaque;
/* Only aligned reads are allowed on OHCI */
if (addr & 3) {
@@ -1697,18 +1698,10 @@ static void ohci_async_cancel_device(OHCIState *ohci, USBDevice *dev)
}
}
-/* Only dword reads are defined on OHCI register space */
-static CPUReadMemoryFunc * const ohci_readfn[3]={
- ohci_mem_read,
- ohci_mem_read,
- ohci_mem_read
-};
-
-/* Only dword writes are defined on OHCI register space */
-static CPUWriteMemoryFunc * const ohci_writefn[3]={
- ohci_mem_write,
- ohci_mem_write,
- ohci_mem_write
+static const MemoryRegionOps ohci_mem_ops = {
+ .read = ohci_mem_read,
+ .write = ohci_mem_write,
+ .endianness = DEVICE_LITTLE_ENDIAN,
};
static USBPortOps ohci_port_ops = {
@@ -1764,8 +1757,7 @@ static int usb_ohci_init(OHCIState *ohci, DeviceState *dev,
}
}
- ohci->mem = cpu_register_io_memory(ohci_readfn, ohci_writefn, ohci,
- DEVICE_LITTLE_ENDIAN);
+ memory_region_init_io(&ohci->mem, &ohci_mem_ops, ohci, "ohci", 256);
ohci->localmem_base = localmem_base;
ohci->name = dev->info->name;
@@ -1799,7 +1791,7 @@ static int usb_ohci_initfn_pci(struct PCIDevice *dev)
ohci->state.irq = ohci->pci_dev.irq[0];
/* TODO: avoid cast below by using dev */
- pci_register_bar_simple(&ohci->pci_dev, 0, 256, 0, ohci->state.mem);
+ pci_register_bar_region(&ohci->pci_dev, 0, 0, &ohci->state.mem);
return 0;
}
@@ -1822,7 +1814,7 @@ static int ohci_init_pxa(SysBusDevice *dev)
/* Cannot fail as we pass NULL for masterbus */
usb_ohci_init(&s->ohci, &dev->qdev, s->num_ports, s->dma_offset, NULL, 0);
sysbus_init_irq(dev, &s->ohci.irq);
- sysbus_init_mmio(dev, 0x1000, s->ohci.mem);
+ sysbus_init_mmio_region(dev, &s->ohci.mem);
return 0;
}
--
1.7.5.3
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1
2011-07-26 11:25 [Qemu-devel] [PATCH v2 00/23] Memory API, batch 1 Avi Kivity
` (22 preceding siblings ...)
2011-07-26 11:26 ` [Qemu-devel] [PATCH v2 23/23] usb-ohci: convert to MemoryRegion Avi Kivity
@ 2011-07-29 14:18 ` Anthony Liguori
23 siblings, 0 replies; 31+ messages in thread
From: Anthony Liguori @ 2011-07-29 14:18 UTC (permalink / raw)
To: Avi Kivity; +Cc: qemu-devel, kvm
On 07/26/2011 06:25 AM, Avi Kivity wrote:
> This patchset contains the core of the memory API, with one device
> (usb-ohci) coverted for reference. The API is currently implemented on
> top of the old ram_addr_t/cpu_register_physical_memory() API, but the plan
> is to make it standalone later.
>
> The goals of the API are:
> - correctness: by modelling the memory hierarchy, things like the 440FX PAM
> registers and movable, overlapping PCI BARs can be modelled accurately.
> - efficiency: by maintaining an object tree describing guest memory, we
> can eventually get rid of the page descriptor array
> - security: by having more information available declaratively, we reduce
> coding errors that may be exploited by malicious guests
Applied all. Thanks.
Regards,
Anthony Liguori
>
> Also available from
>
> git://git.kernel.org/pub/scm/virt/kvm/qemu-kvm.git
> refs/tags/memory-region-batch-1-v2
>
> Changes from v1:
> - switched to gtk-doc
> - more copyright blurbs
> - simplified flatview_simplify()
> - use assert() instead of abort() for invariant checks
> (but keep abort() for runtime errors)
> - commit log fixups
>
> Avi Kivity (23):
> Add memory API documentation
> Hierarchical memory region API
> memory: implement dirty tracking
> memory: merge adjacent segments of a single memory region
> Internal interfaces for memory API
> memory: abstract address space operations
> memory: rename MemoryRegion::has_ram_addr to ::terminates
> memory: late initialization of ram_addr
> memory: I/O address space support
> memory: add backward compatibility for old portio registration
> memory: add backward compatibility for old mmio registration
> memory: add ioeventfd support
> memory: separate building the final memory map into two steps
> memory: transaction API
> exec.c: initialize memory map
> ioport: register ranges by byte aligned addresses always
> pc: grab system_memory
> pc: convert pc_memory_init() to memory API
> pc: move global memory map out of pc_init1() and into its callers
> pci: pass address space to pci bus when created
> pci: add MemoryRegion based BAR management API
> sysbus: add MemoryRegion based memory management API
> usb-ohci: convert to MemoryRegion
>
> Makefile.target | 1 +
> docs/memory.txt | 172 ++++++++
> exec-memory.h | 39 ++
> exec.c | 19 +
> hw/apb_pci.c | 2 +
> hw/bonito.c | 4 +-
> hw/grackle_pci.c | 5 +-
> hw/gt64xxx.c | 4 +-
> hw/pc.c | 62 ++-
> hw/pc.h | 9 +-
> hw/pc_piix.c | 20 +-
> hw/pci.c | 63 +++-
> hw/pci.h | 15 +-
> hw/pci_host.h | 1 +
> hw/pci_internals.h | 1 +
> hw/piix_pci.c | 13 +-
> hw/ppc4xx_pci.c | 5 +-
> hw/ppc_mac.h | 9 +-
> hw/ppc_newworld.c | 5 +-
> hw/ppc_oldworld.c | 3 +-
> hw/ppc_prep.c | 3 +-
> hw/ppce500_pci.c | 6 +-
> hw/prep_pci.c | 5 +-
> hw/prep_pci.h | 3 +-
> hw/sh_pci.c | 4 +-
> hw/sysbus.c | 27 ++-
> hw/sysbus.h | 3 +
> hw/unin_pci.c | 10 +-
> hw/usb-ohci.c | 42 +--
> hw/versatile_pci.c | 2 +
> ioport.c | 4 +-
> memory.c | 1141 ++++++++++++++++++++++++++++++++++++++++++++++++++++
> memory.h | 469 +++++++++++++++++++++
> 33 files changed, 2072 insertions(+), 99 deletions(-)
> create mode 100644 docs/memory.txt
> create mode 100644 exec-memory.h
> create mode 100644 memory.c
> create mode 100644 memory.h
>
^ permalink raw reply [flat|nested] 31+ messages in thread