From 3285cf4fe78b4b83b70b9306cc97968b414a6e9d Mon Sep 17 00:00:00 2001 From: Anthony PERARD Date: Thu, 19 Aug 2010 12:27:56 +0100 Subject: xen: Add initialisation of Xen The xenpv machine use the common init function. Signed-off-by: Anthony PERARD Acked-by: Alexander Graf Signed-off-by: Alexander Graf --- xen-stub.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) create mode 100644 xen-stub.c (limited to 'xen-stub.c') diff --git a/xen-stub.c b/xen-stub.c new file mode 100644 index 0000000000..beb982ff5a --- /dev/null +++ b/xen-stub.c @@ -0,0 +1,15 @@ +/* + * Copyright (C) 2010 Citrix Ltd. + * + * This work is licensed under the terms of the GNU GPL, version 2. See + * the COPYING file in the top-level directory. + * + */ + +#include "qemu-common.h" +#include "hw/xen.h" + +int xen_init(void) +{ + return -ENOSYS; +} -- cgit v1.2.3-55-g7522 From 4144530012793e6cd479eaa90cceb1aef0de8158 Mon Sep 17 00:00:00 2001 From: Anthony PERARD Date: Fri, 16 Jul 2010 14:55:39 +0100 Subject: piix_pci: Introduces Xen specific call for irq. This patch introduces Xen specific call in piix_pci. The specific part for Xen is in write_config, set_irq and get_pirq. Signed-off-by: Anthony PERARD Signed-off-by: Stefano Stabellini Acked-by: Alexander Graf Signed-off-by: Alexander Graf --- hw/pc.h | 1 + hw/pc_piix.c | 6 +++++- hw/piix_pci.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++--- hw/xen.h | 6 ++++++ xen-all.c | 31 +++++++++++++++++++++++++++++++ xen-stub.c | 13 +++++++++++++ 6 files changed, 102 insertions(+), 4 deletions(-) (limited to 'xen-stub.c') diff --git a/hw/pc.h b/hw/pc.h index 6d5730b26b..0dcbee7ca5 100644 --- a/hw/pc.h +++ b/hw/pc.h @@ -176,6 +176,7 @@ 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_xen_init(PCII440FXState **pi440fx_state, int *piix3_devfn, qemu_irq *pic, 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 e814f00531..9353b91100 100644 --- a/hw/pc_piix.c +++ b/hw/pc_piix.c @@ -120,7 +120,11 @@ static void pc_init1(ram_addr_t ram_size, 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); + if (!xen_enabled()) { + pci_bus = i440fx_init(&i440fx_state, &piix3_devfn, isa_irq, ram_size); + } else { + pci_bus = i440fx_xen_init(&i440fx_state, &piix3_devfn, isa_irq, ram_size); + } } else { pci_bus = NULL; i440fx_state = NULL; diff --git a/hw/piix_pci.c b/hw/piix_pci.c index 5f0d92f10d..7f1c4cca31 100644 --- a/hw/piix_pci.c +++ b/hw/piix_pci.c @@ -29,6 +29,7 @@ #include "isa.h" #include "sysbus.h" #include "range.h" +#include "xen.h" /* * I440FX chipset data sheet. @@ -172,6 +173,13 @@ static void i440fx_write_config(PCIDevice *dev, } } +static void i440fx_write_config_xen(PCIDevice *dev, + uint32_t address, uint32_t val, int len) +{ + xen_piix_pci_write_config_client(address, val, len); + i440fx_write_config(dev, address, val, len); +} + static int i440fx_load_old(QEMUFile* f, void *opaque, int version_id) { PCII440FXState *d = opaque; @@ -239,7 +247,10 @@ static int i440fx_initfn(PCIDevice *dev) return 0; } -PCIBus *i440fx_init(PCII440FXState **pi440fx_state, int *piix3_devfn, qemu_irq *pic, ram_addr_t ram_size) +static PCIBus *i440fx_common_init(const char *device_name, + PCII440FXState **pi440fx_state, + int *piix3_devfn, + qemu_irq *pic, ram_addr_t ram_size) { DeviceState *dev; PCIBus *b; @@ -253,13 +264,13 @@ PCIBus *i440fx_init(PCII440FXState **pi440fx_state, int *piix3_devfn, qemu_irq * s->bus = b; qdev_init_nofail(dev); - d = pci_create_simple(b, 0, "i440FX"); + d = pci_create_simple(b, 0, device_name); *pi440fx_state = DO_UPCAST(PCII440FXState, dev, d); piix3 = DO_UPCAST(PIIX3State, dev, pci_create_simple_multifunction(b, -1, true, "PIIX3")); piix3->pic = pic; - pci_bus_irqs(b, piix3_set_irq, pci_slot_get_pirq, piix3, PIIX_NUM_PIRQS); + (*pi440fx_state)->piix3 = piix3; *piix3_devfn = piix3->dev.devfn; @@ -272,6 +283,30 @@ PCIBus *i440fx_init(PCII440FXState **pi440fx_state, int *piix3_devfn, qemu_irq * return b; } +PCIBus *i440fx_init(PCII440FXState **pi440fx_state, int *piix3_devfn, + qemu_irq *pic, ram_addr_t ram_size) +{ + PCIBus *b; + + b = i440fx_common_init("i440FX", pi440fx_state, piix3_devfn, pic, ram_size); + pci_bus_irqs(b, piix3_set_irq, pci_slot_get_pirq, (*pi440fx_state)->piix3, + PIIX_NUM_PIRQS); + + return b; +} + +PCIBus *i440fx_xen_init(PCII440FXState **pi440fx_state, int *piix3_devfn, + qemu_irq *pic, ram_addr_t ram_size) +{ + PCIBus *b; + + b = i440fx_common_init("i440FX-xen", pi440fx_state, piix3_devfn, pic, ram_size); + pci_bus_irqs(b, xen_piix3_set_irq, xen_pci_slot_get_pirq, + (*pi440fx_state)->piix3, PIIX_NUM_PIRQS); + + return b; +} + /* PIIX3 PCI to ISA bridge */ static void piix3_set_irq_pic(PIIX3State *piix3, int pic_irq) { @@ -429,6 +464,14 @@ static PCIDeviceInfo i440fx_info[] = { .no_hotplug = 1, .init = i440fx_initfn, .config_write = i440fx_write_config, + },{ + .qdev.name = "i440FX-xen", + .qdev.desc = "Host bridge", + .qdev.size = sizeof(PCII440FXState), + .qdev.vmsd = &vmstate_i440fx, + .qdev.no_user = 1, + .init = i440fx_initfn, + .config_write = i440fx_write_config_xen, },{ .qdev.name = "PIIX3", .qdev.desc = "ISA bridge", diff --git a/hw/xen.h b/hw/xen.h index bb4dcb5d19..a4096cadfd 100644 --- a/hw/xen.h +++ b/hw/xen.h @@ -8,6 +8,8 @@ */ #include +#include "qemu-common.h" + /* xen-machine.c */ enum xen_mode { XEN_EMULATE = 0, // xen emulation, using xenner (default) @@ -29,6 +31,10 @@ static inline int xen_enabled(void) #endif } +int xen_pci_slot_get_pirq(PCIDevice *pci_dev, int irq_num); +void xen_piix3_set_irq(void *opaque, int irq_num, int level); +void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len); + int xen_init(void); int xen_hvm_init(void); void xen_vcpu_init(void); diff --git a/xen-all.c b/xen-all.c index 0b984b25a6..acb051cffc 100644 --- a/xen-all.c +++ b/xen-all.c @@ -6,9 +6,40 @@ * */ +#include "hw/pci.h" #include "hw/xen_common.h" #include "hw/xen_backend.h" +/* Xen specific function for piix pci */ + +int xen_pci_slot_get_pirq(PCIDevice *pci_dev, int irq_num) +{ + return irq_num + ((pci_dev->devfn >> 3) << 2); +} + +void xen_piix3_set_irq(void *opaque, int irq_num, int level) +{ + xc_hvm_set_pci_intx_level(xen_xc, xen_domid, 0, 0, irq_num >> 2, + irq_num & 3, level); +} + +void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len) +{ + int i; + + /* Scan for updates to PCI link routes (0x60-0x63). */ + for (i = 0; i < len; i++) { + uint8_t v = (val >> (8 * i)) & 0xff; + if (v & 0x80) { + v = 0; + } + v &= 0xf; + if (((address + i) >= 0x60) && ((address + i) <= 0x63)) { + xc_hvm_set_pci_link_route(xen_xc, xen_domid, address + i - 0x60, v); + } + } +} + /* VCPU Operations, MMIO, IO ring ... */ static void xen_reset_vcpu(void *opaque) diff --git a/xen-stub.c b/xen-stub.c index beb982ff5a..dc90f108c3 100644 --- a/xen-stub.c +++ b/xen-stub.c @@ -9,6 +9,19 @@ #include "qemu-common.h" #include "hw/xen.h" +int xen_pci_slot_get_pirq(PCIDevice *pci_dev, int irq_num) +{ + return -1; +} + +void xen_piix3_set_irq(void *opaque, int irq_num, int level) +{ +} + +void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len) +{ +} + int xen_init(void) { return -ENOSYS; -- cgit v1.2.3-55-g7522 From 9c11a8ac886ccbb5f8e1b08e8ae12f045d783031 Mon Sep 17 00:00:00 2001 From: Anthony PERARD Date: Wed, 30 Jun 2010 17:50:10 +0100 Subject: xen: Introduce Xen Interrupt Controller Every set_irq call makes a Xen hypercall. Signed-off-by: Anthony PERARD Signed-off-by: Stefano Stabellini Signed-off-by: Alexander Graf --- hw/pc_piix.c | 8 ++++++-- hw/xen.h | 2 ++ xen-all.c | 12 ++++++++++++ xen-stub.c | 5 +++++ 4 files changed, 25 insertions(+), 2 deletions(-) (limited to 'xen-stub.c') diff --git a/hw/pc_piix.c b/hw/pc_piix.c index 9353b91100..62cdf71daa 100644 --- a/hw/pc_piix.c +++ b/hw/pc_piix.c @@ -110,8 +110,12 @@ static void pc_init1(ram_addr_t ram_size, below_4g_mem_size, above_4g_mem_size); } - cpu_irq = pc_allocate_cpu_irq(); - i8259 = i8259_init(cpu_irq[0]); + if (!xen_enabled()) { + cpu_irq = pc_allocate_cpu_irq(); + i8259 = i8259_init(cpu_irq[0]); + } else { + i8259 = xen_interrupt_controller_init(); + } isa_irq_state = qemu_mallocz(sizeof(*isa_irq_state)); isa_irq_state->i8259 = i8259; if (pci_enabled) { diff --git a/hw/xen.h b/hw/xen.h index a4096cadfd..9f00c0bfed 100644 --- a/hw/xen.h +++ b/hw/xen.h @@ -35,6 +35,8 @@ int xen_pci_slot_get_pirq(PCIDevice *pci_dev, int irq_num); void xen_piix3_set_irq(void *opaque, int irq_num, int level); void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len); +qemu_irq *xen_interrupt_controller_init(void); + int xen_init(void); int xen_hvm_init(void); void xen_vcpu_init(void); diff --git a/xen-all.c b/xen-all.c index acb051cffc..bb809eff16 100644 --- a/xen-all.c +++ b/xen-all.c @@ -40,6 +40,18 @@ void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len) } } +/* Xen Interrupt Controller */ + +static void xen_set_irq(void *opaque, int irq, int level) +{ + xc_hvm_set_isa_irq_level(xen_xc, xen_domid, irq, level); +} + +qemu_irq *xen_interrupt_controller_init(void) +{ + return qemu_allocate_irqs(xen_set_irq, NULL, 16); +} + /* VCPU Operations, MMIO, IO ring ... */ static void xen_reset_vcpu(void *opaque) diff --git a/xen-stub.c b/xen-stub.c index dc90f108c3..3a8449c148 100644 --- a/xen-stub.c +++ b/xen-stub.c @@ -22,6 +22,11 @@ void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len) { } +qemu_irq *xen_interrupt_controller_init(void) +{ + return NULL; +} + int xen_init(void) { return -ENOSYS; -- cgit v1.2.3-55-g7522 From 432d268c0552fd30c8be564f7ea2504a2b546101 Mon Sep 17 00:00:00 2001 From: Jun Nakajima Date: Tue, 31 Aug 2010 16:41:25 +0100 Subject: xen: Introduce the Xen mapcache On IA32 host or IA32 PAE host, at present, generally, we can't create an HVM guest with more than 2G memory, because generally it's almost impossible for Qemu to find a large enough and consecutive virtual address space to map an HVM guest's whole physical address space. The attached patch fixes this issue using dynamic mapping based on little blocks of memory. Each call to qemu_get_ram_ptr makes a call to qemu_map_cache with the lock option, so mapcache will not unmap these ram_ptr. Blocks that do not belong to the RAM, but usually to a device ROM or to a framebuffer, are handled in a separate function. So the whole RAMBlock can be map. Signed-off-by: Jun Nakajima Signed-off-by: Anthony PERARD Signed-off-by: Stefano Stabellini Signed-off-by: Alexander Graf --- Makefile.target | 3 + configure | 3 + exec.c | 48 +++++++- hw/xen.h | 13 ++ hw/xen_common.h | 9 ++ trace-events | 10 ++ xen-all.c | 66 ++++++++++ xen-mapcache-stub.c | 44 +++++++ xen-mapcache.c | 349 ++++++++++++++++++++++++++++++++++++++++++++++++++++ xen-mapcache.h | 37 ++++++ xen-stub.c | 4 + 11 files changed, 582 insertions(+), 4 deletions(-) create mode 100644 xen-mapcache-stub.c create mode 100644 xen-mapcache.c create mode 100644 xen-mapcache.h (limited to 'xen-stub.c') diff --git a/Makefile.target b/Makefile.target index fd84d467da..2e281a4588 100644 --- a/Makefile.target +++ b/Makefile.target @@ -214,8 +214,11 @@ else CONFIG_NO_XEN = y endif # xen support +CONFIG_NO_XEN_MAPCACHE = $(if $(subst n,,$(CONFIG_XEN_MAPCACHE)),n,y) obj-i386-$(CONFIG_XEN) += xen-all.o obj-$(CONFIG_NO_XEN) += xen-stub.o +obj-i386-$(CONFIG_XEN_MAPCACHE) += xen-mapcache.o +obj-$(CONFIG_NO_XEN_MAPCACHE) += xen-mapcache-stub.o # Inter-VM PCI shared memory CONFIG_IVSHMEM = diff --git a/configure b/configure index 5df84d24eb..6fc2bddc12 100755 --- a/configure +++ b/configure @@ -3299,6 +3299,9 @@ case "$target_arch2" in i386|x86_64) if test "$xen" = "yes" -a "$target_softmmu" = "yes" ; then echo "CONFIG_XEN=y" >> $config_target_mak + if test "$cpu" = "i386" -o "$cpu" = "x86_64"; then + echo "CONFIG_XEN_MAPCACHE=y" >> $config_target_mak + fi fi esac case "$target_arch2" in diff --git a/exec.c b/exec.c index 308a86dcc0..8c81cffaa7 100644 --- a/exec.c +++ b/exec.c @@ -32,6 +32,7 @@ #include "hw/qdev.h" #include "osdep.h" #include "kvm.h" +#include "hw/xen.h" #include "qemu-timer.h" #if defined(CONFIG_USER_ONLY) #include @@ -51,6 +52,8 @@ #include #endif #endif +#else /* !CONFIG_USER_ONLY */ +#include "xen-mapcache.h" #endif //#define DEBUG_TB_INVALIDATE @@ -2889,6 +2892,7 @@ ram_addr_t qemu_ram_alloc_from_ptr(DeviceState *dev, const char *name, } } + new_block->offset = find_ram_offset(size); if (host) { new_block->host = host; new_block->flags |= RAM_PREALLOC_MASK; @@ -2911,13 +2915,15 @@ ram_addr_t qemu_ram_alloc_from_ptr(DeviceState *dev, const char *name, PROT_EXEC|PROT_READ|PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0); #else - new_block->host = qemu_vmalloc(size); + if (xen_mapcache_enabled()) { + xen_ram_alloc(new_block->offset, size); + } else { + new_block->host = qemu_vmalloc(size); + } #endif qemu_madvise(new_block->host, size, QEMU_MADV_MERGEABLE); } } - - new_block->offset = find_ram_offset(size); new_block->length = size; QLIST_INSERT_HEAD(&ram_list.blocks, new_block, next); @@ -2962,7 +2968,11 @@ void qemu_ram_free(ram_addr_t addr) #if defined(TARGET_S390X) && defined(CONFIG_KVM) munmap(block->host, block->length); #else - qemu_vfree(block->host); + if (xen_mapcache_enabled()) { + qemu_invalidate_entry(block->host); + } else { + qemu_vfree(block->host); + } #endif } qemu_free(block); @@ -3051,6 +3061,16 @@ void *qemu_get_ram_ptr(ram_addr_t addr) QLIST_REMOVE(block, next); QLIST_INSERT_HEAD(&ram_list.blocks, block, next); } + if (xen_mapcache_enabled()) { + /* We need to check if the requested address is in the RAM + * because we don't want to map the entire memory in QEMU. + */ + if (block->offset == 0) { + return qemu_map_cache(addr, 0, 1); + } else if (block->host == NULL) { + block->host = xen_map_block(block->offset, block->length); + } + } return block->host + (addr - block->offset); } } @@ -3070,6 +3090,16 @@ void *qemu_safe_ram_ptr(ram_addr_t addr) QLIST_FOREACH(block, &ram_list.blocks, next) { if (addr - block->offset < block->length) { + if (xen_mapcache_enabled()) { + /* We need to check if the requested address is in the RAM + * because we don't want to map the entire memory in QEMU. + */ + if (block->offset == 0) { + return qemu_map_cache(addr, 0, 1); + } else if (block->host == NULL) { + block->host = xen_map_block(block->offset, block->length); + } + } return block->host + (addr - block->offset); } } @@ -3086,11 +3116,21 @@ int qemu_ram_addr_from_host(void *ptr, ram_addr_t *ram_addr) uint8_t *host = ptr; QLIST_FOREACH(block, &ram_list.blocks, next) { + /* This case append when the block is not mapped. */ + if (block->host == NULL) { + continue; + } if (host - block->host < block->length) { *ram_addr = block->offset + (host - block->host); return 0; } } + + if (xen_mapcache_enabled()) { + *ram_addr = qemu_ram_addr_from_mapcache(ptr); + return 0; + } + return -1; } diff --git a/hw/xen.h b/hw/xen.h index 9f00c0bfed..6245b38153 100644 --- a/hw/xen.h +++ b/hw/xen.h @@ -31,6 +31,15 @@ static inline int xen_enabled(void) #endif } +static inline int xen_mapcache_enabled(void) +{ +#ifdef CONFIG_XEN_MAPCACHE + return xen_enabled(); +#else + return 0; +#endif +} + int xen_pci_slot_get_pirq(PCIDevice *pci_dev, int irq_num); void xen_piix3_set_irq(void *opaque, int irq_num, int level); void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len); @@ -41,6 +50,10 @@ int xen_init(void); int xen_hvm_init(void); void xen_vcpu_init(void); +#if defined(NEED_CPU_H) && !defined(CONFIG_USER_ONLY) +void xen_ram_alloc(ram_addr_t ram_addr, ram_addr_t size); +#endif + #if defined(CONFIG_XEN) && CONFIG_XEN_CTRL_INTERFACE_VERSION < 400 # define HVM_MAX_VCPUS 32 #endif diff --git a/hw/xen_common.h b/hw/xen_common.h index 47b8afea02..dd3e896e94 100644 --- a/hw/xen_common.h +++ b/hw/xen_common.h @@ -63,6 +63,15 @@ static inline int xc_fd(int xen_xc) } +static inline int xc_domain_populate_physmap_exact + (XenXC xc_handle, uint32_t domid, unsigned long nr_extents, + unsigned int extent_order, unsigned int mem_flags, xen_pfn_t *extent_start) +{ + return xc_domain_memory_populate_physmap + (xc_handle, domid, nr_extents, extent_order, mem_flags, extent_start); +} + + /* Xen 4.1 */ #else diff --git a/trace-events b/trace-events index 4f965e2ebd..d703347d40 100644 --- a/trace-events +++ b/trace-events @@ -361,3 +361,13 @@ disable milkymist_uart_pulse_irq_tx(void) "Pulse IRQ TX" # hw/milkymist-vgafb.c disable milkymist_vgafb_memory_read(uint32_t addr, uint32_t value) "addr %08x value %08x" disable milkymist_vgafb_memory_write(uint32_t addr, uint32_t value) "addr %08x value %08x" + +# xen-all.c +disable xen_ram_alloc(unsigned long ram_addr, unsigned long size) "requested: %#lx, size %#lx" + +# xen-mapcache.c +disable qemu_map_cache(uint64_t phys_addr) "want %#"PRIx64"" +disable qemu_remap_bucket(uint64_t index) "index %#"PRIx64"" +disable qemu_map_cache_return(void* ptr) "%p" +disable xen_map_block(uint64_t phys_addr, uint64_t size) "%#"PRIx64", size %#"PRIx64"" +disable xen_unmap_block(void* addr, unsigned long size) "%p, size %#lx" diff --git a/xen-all.c b/xen-all.c index bb809eff16..cb01ab9aad 100644 --- a/xen-all.c +++ b/xen-all.c @@ -10,6 +10,9 @@ #include "hw/xen_common.h" #include "hw/xen_backend.h" +#include "xen-mapcache.h" +#include "trace.h" + /* Xen specific function for piix pci */ int xen_pci_slot_get_pirq(PCIDevice *pci_dev, int irq_num) @@ -52,6 +55,65 @@ qemu_irq *xen_interrupt_controller_init(void) return qemu_allocate_irqs(xen_set_irq, NULL, 16); } +/* Memory Ops */ + +static void xen_ram_init(ram_addr_t ram_size) +{ + RAMBlock *new_block; + ram_addr_t below_4g_mem_size, above_4g_mem_size = 0; + + new_block = qemu_mallocz(sizeof (*new_block)); + pstrcpy(new_block->idstr, sizeof (new_block->idstr), "xen.ram"); + new_block->host = NULL; + new_block->offset = 0; + new_block->length = ram_size; + + QLIST_INSERT_HEAD(&ram_list.blocks, new_block, next); + + ram_list.phys_dirty = qemu_realloc(ram_list.phys_dirty, + new_block->length >> TARGET_PAGE_BITS); + memset(ram_list.phys_dirty + (new_block->offset >> TARGET_PAGE_BITS), + 0xff, new_block->length >> TARGET_PAGE_BITS); + + if (ram_size >= 0xe0000000 ) { + above_4g_mem_size = ram_size - 0xe0000000; + below_4g_mem_size = 0xe0000000; + } else { + below_4g_mem_size = ram_size; + } + + cpu_register_physical_memory(0, below_4g_mem_size, new_block->offset); +#if TARGET_PHYS_ADDR_BITS > 32 + if (above_4g_mem_size > 0) { + cpu_register_physical_memory(0x100000000ULL, above_4g_mem_size, + new_block->offset + below_4g_mem_size); + } +#endif +} + +void xen_ram_alloc(ram_addr_t ram_addr, ram_addr_t size) +{ + unsigned long nr_pfn; + xen_pfn_t *pfn_list; + int i; + + trace_xen_ram_alloc(ram_addr, size); + + nr_pfn = size >> TARGET_PAGE_BITS; + pfn_list = qemu_malloc(sizeof (*pfn_list) * nr_pfn); + + for (i = 0; i < nr_pfn; i++) { + pfn_list[i] = (ram_addr >> TARGET_PAGE_BITS) + i; + } + + if (xc_domain_populate_physmap_exact(xen_xc, xen_domid, nr_pfn, 0, 0, pfn_list)) { + hw_error("xen: failed to populate ram at %lx", ram_addr); + } + + qemu_free(pfn_list); +} + + /* VCPU Operations, MMIO, IO ring ... */ static void xen_reset_vcpu(void *opaque) @@ -86,5 +148,9 @@ int xen_init(void) int xen_hvm_init(void) { + /* Init RAM management */ + qemu_map_cache_init(); + xen_ram_init(ram_size); + return 0; } diff --git a/xen-mapcache-stub.c b/xen-mapcache-stub.c new file mode 100644 index 0000000000..7c14b3d141 --- /dev/null +++ b/xen-mapcache-stub.c @@ -0,0 +1,44 @@ +/* + * Copyright (C) 2011 Citrix Ltd. + * + * This work is licensed under the terms of the GNU GPL, version 2. See + * the COPYING file in the top-level directory. + * + */ + +#include "config.h" + +#include "exec-all.h" +#include "qemu-common.h" +#include "cpu-common.h" +#include "xen-mapcache.h" + +void qemu_map_cache_init(void) +{ +} + +uint8_t *qemu_map_cache(target_phys_addr_t phys_addr, target_phys_addr_t size, uint8_t lock) +{ + return qemu_get_ram_ptr(phys_addr); +} + +void qemu_map_cache_unlock(void *buffer) +{ +} + +ram_addr_t qemu_ram_addr_from_mapcache(void *ptr) +{ + return -1; +} + +void qemu_invalidate_map_cache(void) +{ +} + +void qemu_invalidate_entry(uint8_t *buffer) +{ +} +uint8_t *xen_map_block(target_phys_addr_t phys_addr, target_phys_addr_t size) +{ + return NULL; +} diff --git a/xen-mapcache.c b/xen-mapcache.c new file mode 100644 index 0000000000..a539358fb7 --- /dev/null +++ b/xen-mapcache.c @@ -0,0 +1,349 @@ +/* + * Copyright (C) 2011 Citrix Ltd. + * + * This work is licensed under the terms of the GNU GPL, version 2. See + * the COPYING file in the top-level directory. + * + */ + +#include "config.h" + +#include + +#include "hw/xen_backend.h" +#include "blockdev.h" + +#include +#include + +#include "xen-mapcache.h" +#include "trace.h" + + +//#define MAPCACHE_DEBUG + +#ifdef MAPCACHE_DEBUG +# define DPRINTF(fmt, ...) do { \ + fprintf(stderr, "xen_mapcache: " fmt, ## __VA_ARGS__); \ +} while (0) +#else +# define DPRINTF(fmt, ...) do { } while (0) +#endif + +#if defined(__i386__) +# define MCACHE_BUCKET_SHIFT 16 +#elif defined(__x86_64__) +# define MCACHE_BUCKET_SHIFT 20 +#endif +#define MCACHE_BUCKET_SIZE (1UL << MCACHE_BUCKET_SHIFT) + +#define BITS_PER_LONG (sizeof(long) * 8) +#define BITS_TO_LONGS(bits) (((bits) + BITS_PER_LONG - 1) / BITS_PER_LONG) +#define DECLARE_BITMAP(name, bits) unsigned long name[BITS_TO_LONGS(bits)] + +typedef struct MapCacheEntry { + target_phys_addr_t paddr_index; + uint8_t *vaddr_base; + DECLARE_BITMAP(valid_mapping, MCACHE_BUCKET_SIZE >> XC_PAGE_SHIFT); + uint8_t lock; + struct MapCacheEntry *next; +} MapCacheEntry; + +typedef struct MapCacheRev { + uint8_t *vaddr_req; + target_phys_addr_t paddr_index; + QTAILQ_ENTRY(MapCacheRev) next; +} MapCacheRev; + +typedef struct MapCache { + MapCacheEntry *entry; + unsigned long nr_buckets; + QTAILQ_HEAD(map_cache_head, MapCacheRev) locked_entries; + + /* For most cases (>99.9%), the page address is the same. */ + target_phys_addr_t last_address_index; + uint8_t *last_address_vaddr; + unsigned long max_mcache_size; + unsigned int mcache_bucket_shift; +} MapCache; + +static MapCache *mapcache; + +static inline int test_bit(unsigned int bit, const unsigned long *map) +{ + return !!((map)[(bit) / BITS_PER_LONG] & (1UL << ((bit) % BITS_PER_LONG))); +} + +void qemu_map_cache_init(void) +{ + unsigned long size; + struct rlimit rlimit_as; + + mapcache = qemu_mallocz(sizeof (MapCache)); + + QTAILQ_INIT(&mapcache->locked_entries); + mapcache->last_address_index = -1; + + getrlimit(RLIMIT_AS, &rlimit_as); + rlimit_as.rlim_cur = rlimit_as.rlim_max; + setrlimit(RLIMIT_AS, &rlimit_as); + mapcache->max_mcache_size = rlimit_as.rlim_max; + + mapcache->nr_buckets = + (((mapcache->max_mcache_size >> XC_PAGE_SHIFT) + + (1UL << (MCACHE_BUCKET_SHIFT - XC_PAGE_SHIFT)) - 1) >> + (MCACHE_BUCKET_SHIFT - XC_PAGE_SHIFT)); + + size = mapcache->nr_buckets * sizeof (MapCacheEntry); + size = (size + XC_PAGE_SIZE - 1) & ~(XC_PAGE_SIZE - 1); + DPRINTF("qemu_map_cache_init, nr_buckets = %lx size %lu\n", mapcache->nr_buckets, size); + mapcache->entry = qemu_mallocz(size); +} + +static void qemu_remap_bucket(MapCacheEntry *entry, + target_phys_addr_t size, + target_phys_addr_t address_index) +{ + uint8_t *vaddr_base; + xen_pfn_t *pfns; + int *err; + unsigned int i, j; + target_phys_addr_t nb_pfn = size >> XC_PAGE_SHIFT; + + trace_qemu_remap_bucket(address_index); + + pfns = qemu_mallocz(nb_pfn * sizeof (xen_pfn_t)); + err = qemu_mallocz(nb_pfn * sizeof (int)); + + if (entry->vaddr_base != NULL) { + if (munmap(entry->vaddr_base, size) != 0) { + perror("unmap fails"); + exit(-1); + } + } + + for (i = 0; i < nb_pfn; i++) { + pfns[i] = (address_index << (MCACHE_BUCKET_SHIFT-XC_PAGE_SHIFT)) + i; + } + + vaddr_base = xc_map_foreign_bulk(xen_xc, xen_domid, PROT_READ|PROT_WRITE, + pfns, err, nb_pfn); + if (vaddr_base == NULL) { + perror("xc_map_foreign_bulk"); + exit(-1); + } + + entry->vaddr_base = vaddr_base; + entry->paddr_index = address_index; + + for (i = 0; i < nb_pfn; i += BITS_PER_LONG) { + unsigned long word = 0; + if ((i + BITS_PER_LONG) > nb_pfn) { + j = nb_pfn % BITS_PER_LONG; + } else { + j = BITS_PER_LONG; + } + while (j > 0) { + word = (word << 1) | !err[i + --j]; + } + entry->valid_mapping[i / BITS_PER_LONG] = word; + } + + qemu_free(pfns); + qemu_free(err); +} + +uint8_t *qemu_map_cache(target_phys_addr_t phys_addr, target_phys_addr_t size, uint8_t lock) +{ + MapCacheEntry *entry, *pentry = NULL; + target_phys_addr_t address_index = phys_addr >> MCACHE_BUCKET_SHIFT; + target_phys_addr_t address_offset = phys_addr & (MCACHE_BUCKET_SIZE - 1); + + trace_qemu_map_cache(phys_addr); + + if (address_index == mapcache->last_address_index && !lock) { + trace_qemu_map_cache_return(mapcache->last_address_vaddr + address_offset); + return mapcache->last_address_vaddr + address_offset; + } + + entry = &mapcache->entry[address_index % mapcache->nr_buckets]; + + while (entry && entry->lock && entry->paddr_index != address_index && entry->vaddr_base) { + pentry = entry; + entry = entry->next; + } + if (!entry) { + entry = qemu_mallocz(sizeof (MapCacheEntry)); + pentry->next = entry; + qemu_remap_bucket(entry, size ? : MCACHE_BUCKET_SIZE, address_index); + } else if (!entry->lock) { + if (!entry->vaddr_base || entry->paddr_index != address_index || + !test_bit(address_offset >> XC_PAGE_SHIFT, entry->valid_mapping)) { + qemu_remap_bucket(entry, size ? : MCACHE_BUCKET_SIZE, address_index); + } + } + + if (!test_bit(address_offset >> XC_PAGE_SHIFT, entry->valid_mapping)) { + mapcache->last_address_index = -1; + trace_qemu_map_cache_return(NULL); + return NULL; + } + + mapcache->last_address_index = address_index; + mapcache->last_address_vaddr = entry->vaddr_base; + if (lock) { + MapCacheRev *reventry = qemu_mallocz(sizeof(MapCacheRev)); + entry->lock++; + reventry->vaddr_req = mapcache->last_address_vaddr + address_offset; + reventry->paddr_index = mapcache->last_address_index; + QTAILQ_INSERT_HEAD(&mapcache->locked_entries, reventry, next); + } + + trace_qemu_map_cache_return(mapcache->last_address_vaddr + address_offset); + return mapcache->last_address_vaddr + address_offset; +} + +ram_addr_t qemu_ram_addr_from_mapcache(void *ptr) +{ + MapCacheRev *reventry; + target_phys_addr_t paddr_index; + int found = 0; + + QTAILQ_FOREACH(reventry, &mapcache->locked_entries, next) { + if (reventry->vaddr_req == ptr) { + paddr_index = reventry->paddr_index; + found = 1; + break; + } + } + if (!found) { + fprintf(stderr, "qemu_ram_addr_from_mapcache, could not find %p\n", ptr); + QTAILQ_FOREACH(reventry, &mapcache->locked_entries, next) { + DPRINTF(" "TARGET_FMT_plx" -> %p is present\n", reventry->paddr_index, + reventry->vaddr_req); + } + abort(); + return 0; + } + + return paddr_index << MCACHE_BUCKET_SHIFT; +} + +void qemu_invalidate_entry(uint8_t *buffer) +{ + MapCacheEntry *entry = NULL, *pentry = NULL; + MapCacheRev *reventry; + target_phys_addr_t paddr_index; + int found = 0; + + if (mapcache->last_address_vaddr == buffer) { + mapcache->last_address_index = -1; + } + + QTAILQ_FOREACH(reventry, &mapcache->locked_entries, next) { + if (reventry->vaddr_req == buffer) { + paddr_index = reventry->paddr_index; + found = 1; + break; + } + } + if (!found) { + DPRINTF("qemu_invalidate_entry, could not find %p\n", buffer); + QTAILQ_FOREACH(reventry, &mapcache->locked_entries, next) { + DPRINTF(" "TARGET_FMT_plx" -> %p is present\n", reventry->paddr_index, reventry->vaddr_req); + } + return; + } + QTAILQ_REMOVE(&mapcache->locked_entries, reventry, next); + qemu_free(reventry); + + entry = &mapcache->entry[paddr_index % mapcache->nr_buckets]; + while (entry && entry->paddr_index != paddr_index) { + pentry = entry; + entry = entry->next; + } + if (!entry) { + DPRINTF("Trying to unmap address %p that is not in the mapcache!\n", buffer); + return; + } + entry->lock--; + if (entry->lock > 0 || pentry == NULL) { + return; + } + + pentry->next = entry->next; + if (munmap(entry->vaddr_base, MCACHE_BUCKET_SIZE) != 0) { + perror("unmap fails"); + exit(-1); + } + qemu_free(entry); +} + +void qemu_invalidate_map_cache(void) +{ + unsigned long i; + MapCacheRev *reventry; + + /* Flush pending AIO before destroying the mapcache */ + qemu_aio_flush(); + + QTAILQ_FOREACH(reventry, &mapcache->locked_entries, next) { + DPRINTF("There should be no locked mappings at this time, " + "but "TARGET_FMT_plx" -> %p is present\n", + reventry->paddr_index, reventry->vaddr_req); + } + + mapcache_lock(); + + for (i = 0; i < mapcache->nr_buckets; i++) { + MapCacheEntry *entry = &mapcache->entry[i]; + + if (entry->vaddr_base == NULL) { + continue; + } + + if (munmap(entry->vaddr_base, MCACHE_BUCKET_SIZE) != 0) { + perror("unmap fails"); + exit(-1); + } + + entry->paddr_index = 0; + entry->vaddr_base = NULL; + } + + mapcache->last_address_index = -1; + mapcache->last_address_vaddr = NULL; + + mapcache_unlock(); +} + +uint8_t *xen_map_block(target_phys_addr_t phys_addr, target_phys_addr_t size) +{ + uint8_t *vaddr_base; + xen_pfn_t *pfns; + int *err; + unsigned int i; + target_phys_addr_t nb_pfn = size >> XC_PAGE_SHIFT; + + trace_xen_map_block(phys_addr, size); + phys_addr >>= XC_PAGE_SHIFT; + + pfns = qemu_mallocz(nb_pfn * sizeof (xen_pfn_t)); + err = qemu_mallocz(nb_pfn * sizeof (int)); + + for (i = 0; i < nb_pfn; i++) { + pfns[i] = phys_addr + i; + } + + vaddr_base = xc_map_foreign_bulk(xen_xc, xen_domid, PROT_READ|PROT_WRITE, + pfns, err, nb_pfn); + if (vaddr_base == NULL) { + perror("xc_map_foreign_bulk"); + exit(-1); + } + + qemu_free(pfns); + qemu_free(err); + + return vaddr_base; +} diff --git a/xen-mapcache.h b/xen-mapcache.h new file mode 100644 index 0000000000..339444c94e --- /dev/null +++ b/xen-mapcache.h @@ -0,0 +1,37 @@ +/* + * Copyright (C) 2011 Citrix Ltd. + * + * This work is licensed under the terms of the GNU GPL, version 2. See + * the COPYING file in the top-level directory. + * + */ + +#ifndef XEN_MAPCACHE_H +#define XEN_MAPCACHE_H + +#include +#include "trace.h" + +void qemu_map_cache_init(void); +uint8_t *qemu_map_cache(target_phys_addr_t phys_addr, target_phys_addr_t size, uint8_t lock); +void qemu_map_cache_unlock(void *phys_addr); +ram_addr_t qemu_ram_addr_from_mapcache(void *ptr); +void qemu_invalidate_entry(uint8_t *buffer); +void qemu_invalidate_map_cache(void); + +uint8_t *xen_map_block(target_phys_addr_t phys_addr, target_phys_addr_t size); + +static inline void xen_unmap_block(void *addr, ram_addr_t size) +{ + trace_xen_unmap_block(addr, size); + + if (munmap(addr, size) != 0) { + hw_error("xen_unmap_block: %s", strerror(errno)); + } +} + + +#define mapcache_lock() ((void)0) +#define mapcache_unlock() ((void)0) + +#endif /* !XEN_MAPCACHE_H */ diff --git a/xen-stub.c b/xen-stub.c index 3a8449c148..8d2fa54b4b 100644 --- a/xen-stub.c +++ b/xen-stub.c @@ -22,6 +22,10 @@ void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len) { } +void xen_ram_alloc(ram_addr_t ram_addr, ram_addr_t size) +{ +} + qemu_irq *xen_interrupt_controller_init(void) { return NULL; -- cgit v1.2.3-55-g7522 From c962247883ffd957dd7f3bccb519803e2775ced2 Mon Sep 17 00:00:00 2001 From: Anthony PERARD Date: Tue, 5 Oct 2010 16:40:22 +0100 Subject: xen: Add Xen hypercall for sleep state in the cmos_s3 callback. Signed-off-by: Anthony PERARD Signed-off-by: Alexander Graf --- hw/pc_piix.c | 6 +++++- hw/xen.h | 1 + xen-all.c | 9 +++++++++ xen-stub.c | 4 ++++ 4 files changed, 19 insertions(+), 1 deletion(-) (limited to 'xen-stub.c') diff --git a/hw/pc_piix.c b/hw/pc_piix.c index 62cdf71daa..9a22a8afc8 100644 --- a/hw/pc_piix.c +++ b/hw/pc_piix.c @@ -179,7 +179,11 @@ static void pc_init1(ram_addr_t ram_size, if (pci_enabled && acpi_enabled) { i2c_bus *smbus; - cmos_s3 = qemu_allocate_irqs(pc_cmos_set_s3_resume, rtc_state, 1); + if (!xen_enabled()) { + cmos_s3 = qemu_allocate_irqs(pc_cmos_set_s3_resume, rtc_state, 1); + } else { + cmos_s3 = qemu_allocate_irqs(xen_cmos_set_s3_resume, rtc_state, 1); + } smi_irq = qemu_allocate_irqs(pc_acpi_smi_interrupt, first_cpu, 1); /* TODO: Populate SPD eeprom data. */ smbus = piix4_pm_init(pci_bus, piix3_devfn + 3, 0xb100, diff --git a/hw/xen.h b/hw/xen.h index 6245b38153..d435ca0ce5 100644 --- a/hw/xen.h +++ b/hw/xen.h @@ -43,6 +43,7 @@ static inline int xen_mapcache_enabled(void) int xen_pci_slot_get_pirq(PCIDevice *pci_dev, int irq_num); void xen_piix3_set_irq(void *opaque, int irq_num, int level); void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len); +void xen_cmos_set_s3_resume(void *opaque, int irq, int level); qemu_irq *xen_interrupt_controller_init(void); diff --git a/xen-all.c b/xen-all.c index 19c2fe1e0f..0eac202d4e 100644 --- a/xen-all.c +++ b/xen-all.c @@ -9,6 +9,7 @@ #include #include "hw/pci.h" +#include "hw/pc.h" #include "hw/xen_common.h" #include "hw/xen_backend.h" @@ -99,6 +100,14 @@ void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len) } } +void xen_cmos_set_s3_resume(void *opaque, int irq, int level) +{ + pc_cmos_set_s3_resume(opaque, irq, level); + if (level) { + xc_set_hvm_param(xen_xc, xen_domid, HVM_PARAM_ACPI_S_STATE, 3); + } +} + /* Xen Interrupt Controller */ static void xen_set_irq(void *opaque, int irq, int level) diff --git a/xen-stub.c b/xen-stub.c index 8d2fa54b4b..a4f35a19fb 100644 --- a/xen-stub.c +++ b/xen-stub.c @@ -22,6 +22,10 @@ void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int len) { } +void xen_cmos_set_s3_resume(void *opaque, int irq, int level) +{ +} + void xen_ram_alloc(ram_addr_t ram_addr, ram_addr_t size) { } -- cgit v1.2.3-55-g7522