From: Peter Dinda Date: Wed, 19 May 2010 16:19:03 +0000 (-0500) Subject: Merge branch 'Release-1.2' of ssh://palacios@newskysaw.cs.northwestern.edu//home... X-Git-Url: http://v3vee.org/palacios/gitweb/gitweb.cgi?p=palacios.git;a=commitdiff_plain;h=refs%2Fheads%2FRelease-1.2;hp=87b1679efe8bee960b65cfd730188ccc94966b71 Merge branch 'Release-1.2' of ssh://palacios@newskysaw.cs.northwestern.edu//home/palacios/palacios into Release-1.2 --- diff --git a/Kconfig b/Kconfig index 29926b0..a36c46b 100644 --- a/Kconfig +++ b/Kconfig @@ -3,6 +3,13 @@ mainmenu "Palacios VMM Configuration" menu "Target Configuration" +config KITTEN + bool "Set defaults needed for the Kitten OS" + default y + help + This enables the necesary options to compile Palacios with Kitten + + config CRAY_XT bool "Red Storm (Cray XT3/XT4)" help @@ -38,6 +45,15 @@ config MAX_CPUS For uniprocessor environments, set this to 1 +config CONSOLE + bool "Include Console Support" + default n + help + Enable console support in Palacios + + + + config SOCKET bool "Include Network Socket Support" default n @@ -117,7 +133,7 @@ config DEBUG_VNET config BUILT_IN_STDLIB bool "Enable Built in versions of stdlib functions" - default y + default y if KITTEN help Not all host OSes provide link targets for stdlib functions Palacios provides internal implementations of these functions, that you can select from this list @@ -176,7 +192,7 @@ config BUILT_IN_STRCMP config BUILT_IN_STRCASECMP bool "strcasecmp()" - default y + default y if KITTEN depends on BUILT_IN_STDLIB help This enables Palacios' internal implementation of strcasecmp @@ -241,7 +257,7 @@ config BUILT_IN_STRSTR config BUILT_IN_ATOI bool "atoi()" - default y + default y if KITTEN depends on BUILT_IN_STDLIB help This enables Palacios' internal implementation of atoi diff --git a/palacios/include/devices/console.h b/palacios/include/devices/console.h index 70eeff7..1040b95 100644 --- a/palacios/include/devices/console.h +++ b/palacios/include/devices/console.h @@ -28,12 +28,13 @@ struct v3_console_ops { // filled in by the backend device int (*update_screen)(uint_t x, uint_t y, uint_t length, void * private_data); int (*update_cursor)(uint_t x, uint_t y, void * private_data); - int (*scroll)(uint_t rows, void * private_data); + int (*scroll)(int rows, void * private_data); }; int v3_cons_get_fb(struct vm_device * frontend_dev, uint8_t * dst, uint_t offset, uint_t length); +int v3_console_register_cga(struct vm_device * cga_dev, struct v3_console_ops * ops, void * private_data); #endif // ! __V3VEE__ diff --git a/palacios/include/devices/lnx_virtio_pci.h b/palacios/include/devices/lnx_virtio_pci.h index 1131feb..5ecf694 100644 --- a/palacios/include/devices/lnx_virtio_pci.h +++ b/palacios/include/devices/lnx_virtio_pci.h @@ -128,6 +128,8 @@ struct virtio_queue { uint16_t queue_size; uint16_t cur_avail_idx; + uint16_t last_avail_idx; + bool idx_overflow; addr_t ring_desc_addr; addr_t ring_avail_addr; diff --git a/palacios/include/palacios/vmm.h b/palacios/include/palacios/vmm.h index b05871f..4640c01 100644 --- a/palacios/include/palacios/vmm.h +++ b/palacios/include/palacios/vmm.h @@ -21,7 +21,7 @@ #define __VMM_H__ -//#include +/*#include */ #include #include @@ -30,11 +30,11 @@ struct guest_info; #ifdef __V3VEE__ -//#include +/*#include */ #include -//#include +/*#include */ /* utility definitions */ @@ -118,7 +118,7 @@ struct guest_info; var; \ }) -// We need to check the hook structure at runtime to ensure its SAFE +/* We need to check the hook structure at runtime to ensure its SAFE */ #define V3_Free(addr) \ do { \ extern struct v3_os_hooks * os_hooks; \ @@ -127,7 +127,7 @@ struct guest_info; } \ } while (0) -// uint_t V3_CPU_KHZ(); +/* uint_t V3_CPU_KHZ(); */ #define V3_CPU_KHZ() ({ \ unsigned int khz = 0; \ extern struct v3_os_hooks * os_hooks; \ @@ -189,7 +189,7 @@ struct guest_info; typedef enum v3_vm_class {V3_INVALID_VM, V3_PC_VM, V3_CRAY_VM} v3_vm_class_t; -// Maybe make this a define.... +/* Maybe make this a define.... */ typedef enum v3_cpu_arch {V3_INVALID_CPU, V3_SVM_CPU, V3_SVM_REV3_CPU, V3_VMX_CPU, V3_VMX_EPT_CPU} v3_cpu_arch_t; @@ -210,7 +210,7 @@ v3_cpu_arch_t v3_get_cpu_type(int cpu_id); int v3_vm_enter(struct guest_info * info); -#endif //!__V3VEE__ +#endif /* !__V3VEE__ */ @@ -257,16 +257,17 @@ struct v3_os_hooks { -// -// -// This is the interrupt state that the VMM's interrupt handlers need to see -// +/* + * + * This is the interrupt state that the VMM's interrupt handlers need to see + */ struct v3_interrupt { unsigned int irq; unsigned int error; - unsigned int should_ack; // Should the vmm ack this interrupt, or will - // the host OS do it? + unsigned int should_ack; /* Should the vmm ack this interrupt, or will + * the host OS do it? + */ }; diff --git a/palacios/include/palacios/vmm_console.h b/palacios/include/palacios/vmm_console.h new file mode 100644 index 0000000..2b38c0d --- /dev/null +++ b/palacios/include/palacios/vmm_console.h @@ -0,0 +1,93 @@ +/* + * This file is part of the Palacios Virtual Machine Monitor developed + * by the V3VEE Project with funding from the United States National + * Science Foundation and the Department of Energy. + * + * The V3VEE Project is a joint project between Northwestern University + * and the University of New Mexico. You can find out more at + * http://www.v3vee.org + * + * Copyright (c) 2010, Jack Lange + * Copyright (c) 2010, Erik van der Kouwe + * Copyright (c) 2010, The V3VEE Project + * All rights reserved. + * + * Author: Jack Lange + * Author: Erik van der Kouwe + * + * This is free software. You are permitted to use, + * redistribute, and modify it as specified in the file "V3VEE_LICENSE". + */ + + +#ifndef __VMM_CONSOLE_H__ +#define __VMM_CONSOLE_H__ + +#include + + +#ifdef __V3VEE__ + +#define V3_TtyOpen(path, mode) \ + ({ \ + extern struct v3_console_hooks *console_hooks; \ + ((console_hooks) && (console_hooks)->tty_open) ? \ + (console_hooks)->tty_open((path), (mode)) : NULL; \ + }) + +#define V3_TtyCursorSet(tty, x, y) \ + ({ \ + extern struct v3_console_hooks *console_hooks; \ + ((console_hooks) && (console_hooks)->tty_cursor_set) ? \ + (console_hooks)->tty_cursor_set((tty), (x), (y)) : -1; \ + }) + +#define V3_TtyCharacterSet(tty, x, y, c, style) \ + ({ \ + extern struct v3_console_hooks *console_hooks; \ + ((console_hooks) && (console_hooks)->tty_character_set) ? \ + (console_hooks)->tty_character_set((tty), (x), (y), (c), (style)) : -1; \ + }) + +#define V3_TtyScroll(tty, lines) \ + ({ \ + extern struct v3_console_hooks *console_hooks; \ + ((console_hooks) && (console_hooks)->tty_scroll) ? \ + (console_hooks)->tty_scroll((tty), (lines)) : -1; \ + }) + +#define V3_TtyUpdate(tty) \ + ({ \ + extern struct v3_console_hooks *console_hooks; \ + ((console_hooks) && (console_hooks)->tty_update) ? \ + (console_hooks)->tty_update((tty)) : -1; \ + }) + +#endif + +#define TTY_OPEN_MODE_READ (1 << 0) +#define TTY_OPEN_MODE_WRITE (1 << 1) + +struct v3_console_hooks { + /* open console device, mode is a combination of TTY_OPEN_MODE_* flags */ + void *(*tty_open)(const char *path, int mode); + + /* set cursor position */ + int (*tty_cursor_set)(void *tty, int x, int y); + + /* output character c with specified style at (x, y) */ + int (*tty_character_set)(void *tty, int x, int y, char c, unsigned char style); + + /* scroll the console down the specified number of lines */ + int (*tty_scroll)(void *tty, int lines); + + /* force update of console display; all updates by above functions + * may be defferred until the next tty_update call + */ + int (*tty_update)(void *tty); +}; + + +extern void V3_Init_Console(struct v3_console_hooks * hooks); + +#endif diff --git a/palacios/include/palacios/vmm_host_events.h b/palacios/include/palacios/vmm_host_events.h index e115005..0635885 100644 --- a/palacios/include/palacios/vmm_host_events.h +++ b/palacios/include/palacios/vmm_host_events.h @@ -76,7 +76,7 @@ int v3_hook_host_event(struct guest_info * info, union v3_host_event_handler cb, void * private_data); -#endif // ! __V3VEE__ +#endif /* ! __V3VEE__ */ diff --git a/palacios/include/palacios/vmm_intr.h b/palacios/include/palacios/vmm_intr.h index 0781844..44de7db 100644 --- a/palacios/include/palacios/vmm_intr.h +++ b/palacios/include/palacios/vmm_intr.h @@ -84,6 +84,7 @@ struct intr_ctrl_ops { }; +void v3_clear_pending_intr(struct guest_info * core); void v3_register_intr_controller(struct guest_info * info, struct intr_ctrl_ops * ops, void * state); diff --git a/palacios/include/palacios/vmm_mem.h b/palacios/include/palacios/vmm_mem.h index f6338df..5484007 100644 --- a/palacios/include/palacios/vmm_mem.h +++ b/palacios/include/palacios/vmm_mem.h @@ -35,12 +35,13 @@ struct guest_info; -// These are the types of physical memory address regions -// from the perspective of the HOST +/* These are the types of physical memory address regions + * from the perspective of the HOST + */ typedef enum shdw_region_type { - SHDW_REGION_WRITE_HOOK, // This region is mapped as read-only (page faults on write) - SHDW_REGION_FULL_HOOK, // This region is mapped as not present (always generate page faults) - SHDW_REGION_ALLOCATED, // Region is a section of host memory + SHDW_REGION_WRITE_HOOK, /* This region is mapped as read-only (page faults on write) */ + SHDW_REGION_FULL_HOOK, /* This region is mapped as not present (always generate page faults) */ + SHDW_REGION_ALLOCATED, /* Region is a section of host memory */ } v3_shdw_region_type_t; @@ -52,12 +53,12 @@ struct v3_shadow_region { v3_shdw_region_type_t host_type; - addr_t host_addr; // This either points to a host address mapping + addr_t host_addr; /* This either points to a host address mapping */ - // Called when data is read from a memory page + /* Called when data is read from a memory page */ int (*read_hook)(addr_t guest_addr, void * dst, uint_t length, void * priv_data); - // Called when data is written to a memory page + /* Called when data is written to a memory page */ int (*write_hook)(addr_t guest_addr, void * src, uint_t length, void * priv_data); void * priv_data; @@ -131,7 +132,7 @@ int v3_handle_mem_wr_hook(struct guest_info * info, addr_t guest_va, addr_t gues int v3_handle_mem_full_hook(struct guest_info * info, addr_t guest_va, addr_t guest_pa, struct v3_shadow_region * reg, pf_error_t access_info); -#endif // ! __V3VEE__ +#endif /* ! __V3VEE__ */ #endif diff --git a/palacios/include/palacios/vmm_paging.h b/palacios/include/palacios/vmm_paging.h index d1f93f2..e494e96 100644 --- a/palacios/include/palacios/vmm_paging.h +++ b/palacios/include/palacios/vmm_paging.h @@ -370,7 +370,7 @@ typedef struct pdpe64 { } __attribute__((packed)) pdpe64_t; -// We Don't support this +/* We Don't support this */ typedef struct pdpe64_1GB { uint_t present : 1; uint_t writable : 1; @@ -445,11 +445,11 @@ typedef struct pte64 { /* *************** */ typedef struct pf_error_code { - uint_t present : 1; // if 0, fault due to page not present - uint_t write : 1; // if 1, faulting access was a write - uint_t user : 1; // if 1, faulting access was in user mode - uint_t rsvd_access : 1; // if 1, fault from reading a 1 from a reserved field (?) - uint_t ifetch : 1; // if 1, faulting access was an instr fetch (only with NX) + uint_t present : 1; /* if 0, fault due to page not present */ + uint_t write : 1; /* if 1, faulting access was a write */ + uint_t user : 1; /* if 1, faulting access was in user mode */ + uint_t rsvd_access : 1; /* if 1, fault from reading a 1 from a reserved field (?) */ + uint_t ifetch : 1; /* if 1, faulting access was an instr fetch (only with NX) */ uint_t rsvd : 27; } __attribute__((packed)) pf_error_t; @@ -592,7 +592,7 @@ void PrintHostPageTree(struct guest_info * info, addr_t virtual_addr, addr_t cr3 void PrintGuestPageTree(struct guest_info * info, addr_t virtual_addr, addr_t cr3); -#endif // !__V3VEE__ +#endif /* !__V3VEE__ */ diff --git a/palacios/include/palacios/vmm_queue.h b/palacios/include/palacios/vmm_queue.h index a4994f5..e88329f 100644 --- a/palacios/include/palacios/vmm_queue.h +++ b/palacios/include/palacios/vmm_queue.h @@ -25,7 +25,7 @@ #include #include - +#include /* IMPORTANT: @@ -43,9 +43,7 @@ struct queue_entry { struct gen_queue { uint_t num_entries; struct list_head entries; - - // We really need to implement this.... - // void * lock; + v3_lock_t lock; }; diff --git a/palacios/include/palacios/vmm_socket.h b/palacios/include/palacios/vmm_socket.h index 1e1d5f3..2108d57 100644 --- a/palacios/include/palacios/vmm_socket.h +++ b/palacios/include/palacios/vmm_socket.h @@ -204,41 +204,41 @@ struct v3_timeval { #define V3_SOCK_SETSIZE 1000 typedef struct v3_sock_set { - // This format needs to match the standard posix FD_SET format, so it can be cast + /* This format needs to match the standard posix FD_SET format, so it can be cast */ unsigned char fd_bits [(V3_SOCK_SETSIZE + 7) / 8]; } v3_sock_set; struct v3_socket_hooks { - // Socket creation routines + /* Socket creation routines */ int (*tcp_socket)(const int bufsize, const int nodelay, const int nonblocking); int (*udp_socket)(const int bufsize, const int nonblocking); - // Socket Destruction + /* Socket Destruction */ void (*close)(int sock); - // Network Server Calls + /* Network Server Calls */ int (*bind_socket)(const int sock, const int port); int (*listen)(const int sock, int backlog); int (*accept)(const int sock, unsigned int * remote_ip, unsigned int * port); - // This going to suck + /* This going to suck */ int (*select)(struct v3_sock_set * rset, \ struct v3_sock_set * wset, \ struct v3_sock_set * eset, \ struct v3_timeval tv); - // Connect calls + /* Connect calls */ int (*connect_to_ip)(const int sock, const int hostip, const int port); int (*connect_to_host)(const int sock, const char * hostname, const int port); - // TCP Data Transfer + /* TCP Data Transfer */ int (*send)(const int sock, const char * buf, const int len); int (*recv)(const int sock, char * buf, const int len); - // UDP Data Transfer + /* UDP Data Transfer */ int (*sendto_host)(const int sock, const char * hostname, const int port, const char * buf, const int len); int (*sendto_ip)(const int sock, const int ip_addr, const int port, diff --git a/palacios/include/palacios/vmm_types.h b/palacios/include/palacios/vmm_types.h index 9a3fe61..8f898dc 100644 --- a/palacios/include/palacios/vmm_types.h +++ b/palacios/include/palacios/vmm_types.h @@ -21,7 +21,7 @@ #define __VMM_TYPES_H #ifdef __V3VEE__ -//#include +/* #include */ @@ -76,6 +76,6 @@ typedef char sint8_t; typedef ulong_t addr_t; typedef ullong_t v3_reg_t; -#endif // ! __V3VEE__ +#endif /* ! __V3VEE__ */ #endif diff --git a/palacios/include/palacios/vmm_util.h b/palacios/include/palacios/vmm_util.h index c6111d2..f3319b6 100644 --- a/palacios/include/palacios/vmm_util.h +++ b/palacios/include/palacios/vmm_util.h @@ -84,7 +84,7 @@ void v3_dump_mem(uint8_t * start, int n); __rem; \ }) -//#define do_divll do_div +/*#define do_divll do_div*/ /* @@ -150,6 +150,6 @@ void v3_dump_mem(uint8_t * start, int n); -#endif // ! __V3VEE__ +#endif /* ! __V3VEE__ */ #endif diff --git a/palacios/include/palacios/vmm_vnet.h b/palacios/include/palacios/vmm_vnet.h index 90e3ca1..eef724c 100644 --- a/palacios/include/palacios/vmm_vnet.h +++ b/palacios/include/palacios/vmm_vnet.h @@ -9,8 +9,6 @@ * * Copyright (c) 2009, Lei Xia * Copyright (c) 2009, Yuan Tang - * Copyright (c) 2009, Jack Lange - * Copyright (c) 2009, Peter Dinda * All rights reserved. * @@ -40,6 +38,8 @@ typedef enum {MAC_ANY, MAC_NOT, MAC_NONE} mac_type_t; //for 'src_mac_qual' and ' typedef enum {LINK_INTERFACE, LINK_EDGE, LINK_ANY} link_type_t; //for 'type' and 'src_type' in struct routing typedef enum {TCP_TYPE, UDP_TYPE, NONE_TYPE} prot_type_t; +#define VNET_INITAB_HCALL 0xca00 + //routing table entry struct routing_entry{ char src_mac[6]; @@ -115,7 +115,6 @@ struct vnet_if_link { }__attribute__((packed)); -//link table entry struct link_entry { link_type_t type; @@ -127,6 +126,13 @@ struct link_entry { int use; }__attribute__((packed)); +struct ethernet_pkt { + uint32_t size; //size of data field + uint16_t type; + uint8_t use_header; + struct udp_link_header ext_hdr; + char data[ETHERNET_PACKET_LEN]; +}__attribute__((packed)); int v3_vnet_send_rawpkt(uchar_t *buf, int len, void *private_data); int v3_vnet_send_udppkt(uchar_t *buf, int len, void *private_data); @@ -147,7 +153,7 @@ int vnet_add_route_entry(char src_mac[6], int v3_vnet_pkt_process(); -void v3_vnet_init(struct guest_info *vm); +void v3_init_vnet(); #endif diff --git a/palacios/include/palacios/vmx_lowlevel.h b/palacios/include/palacios/vmx_lowlevel.h index 81872e0..01bca79 100644 --- a/palacios/include/palacios/vmx_lowlevel.h +++ b/palacios/include/palacios/vmx_lowlevel.h @@ -128,7 +128,7 @@ static inline int vmcs_store(addr_t vmcs_ptr) { } static inline int vmcs_read(vmcs_field_t vmcs_field, void * dst) { - uint64_t val = 0; + addr_t val = 0; uint8_t ret_valid = 0; uint8_t ret_invalid = 0; @@ -136,9 +136,9 @@ static inline int vmcs_read(vmcs_field_t vmcs_field, void * dst) { VMREAD_OPCODE EAX_ECX_MODRM "seteb %1;" // fail valid - "setnaeb %2;" // fail invalid - : "=&c"(val), "=q"(ret_valid), "=q"(ret_invalid) // Use ECX - : "a" (vmcs_field), "1"(ret_valid), "2"(ret_invalid) + "setnaeb %1;" // fail invalid + : "=c"(val), "=d"(ret_valid) //, "=r"(ret_invalid) // Use ECX + : "a" (vmcs_field), "0"(0), "1"(ret_valid) : "memory" ); @@ -171,7 +171,7 @@ static inline int vmcs_write(vmcs_field_t vmcs_field, addr_t value) { "seteb %0;" // fail valid (ZF=1) "setnaeb %1;" // fail invalid (CF=1) : "=q" (ret_valid), "=q" (ret_invalid) - : "a" (vmcs_field), "c"(value), "0"(ret_valid), "1"(ret_invalid) + : "a" (vmcs_field), "c"(value) : "memory"); CHECK_VMXFAIL(ret_valid, ret_invalid); diff --git a/palacios/src/devices/8259a.c b/palacios/src/devices/8259a.c index 443fb0a..34851dd 100644 --- a/palacios/src/devices/8259a.c +++ b/palacios/src/devices/8259a.c @@ -421,6 +421,8 @@ static int write_master_port1(ushort_t port, void * src, uint_t length, struct v return -1; } + v3_clear_pending_intr(dev->vm); + if (IS_ICW1(cw)) { PrintDebug("8259 PIC: Setting ICW1 = %x (wr_Master1)\n", cw); @@ -483,12 +485,16 @@ static int write_master_port2(ushort_t port, void * src, uint_t length, struct v return -1; } + v3_clear_pending_intr(dev->vm); + if (state->master_state == ICW2) { struct icw1 * cw1 = (struct icw1 *)&(state->master_icw1); PrintDebug("8259 PIC: Setting ICW2 = %x (wr_Master2)\n", cw); state->master_icw2 = cw; + + if (cw1->sngl == 0) { state->master_state = ICW3; } else if (cw1->ic4 == 1) { @@ -497,6 +503,8 @@ static int write_master_port2(ushort_t port, void * src, uint_t length, struct v state->master_state = READY; } + + } else if (state->master_state == ICW3) { struct icw1 * cw1 = (struct icw1 *)&(state->master_icw1); @@ -539,6 +547,8 @@ static int write_slave_port1(ushort_t port, void * src, uint_t length, struct vm return -1; } + v3_clear_pending_intr(dev->vm); + if (IS_ICW1(cw)) { PrintDebug("8259 PIC: Setting ICW1 = %x (wr_Slave1)\n", cw); state->slave_icw1 = cw; @@ -597,6 +607,9 @@ static int write_slave_port2(ushort_t port, void * src, uint_t length, struct vm return -1; } + v3_clear_pending_intr(dev->vm); + + if (state->slave_state == ICW2) { struct icw1 * cw1 = (struct icw1 *)&(state->master_icw1); diff --git a/palacios/src/devices/Kconfig b/palacios/src/devices/Kconfig index 8fc349c..30fb2b1 100644 --- a/palacios/src/devices/Kconfig +++ b/palacios/src/devices/Kconfig @@ -286,5 +286,13 @@ config TELNET_CONSOLE help Includes the virtual telnet console + +config CURSES_CONSOLE + bool "Curses Virtual Console" + default n + depends on CGA && !PASSTHROUGH_VIDEO + help + Includes the virtual curses console + endmenu diff --git a/palacios/src/devices/Makefile b/palacios/src/devices/Makefile index 1aeb3f4..b159c16 100644 --- a/palacios/src/devices/Makefile +++ b/palacios/src/devices/Makefile @@ -26,5 +26,6 @@ obj-$(CONFIG_NETDISK) += netdisk.o obj-$(CONFIG_CGA) += cga.o obj-$(CONFIG_TELNET_CONSOLE) += telnet_cons.o +obj-$(CONFIG_CURSES_CONSOLE) += curses_cons.o obj-$(CONFIG_PASSTHROUGH_PCI) += pci_passthrough.o diff --git a/palacios/src/devices/cga.c b/palacios/src/devices/cga.c index ee94f84..6c82c58 100644 --- a/palacios/src/devices/cga.c +++ b/palacios/src/devices/cga.c @@ -24,7 +24,7 @@ #include #include -#include +#include @@ -120,6 +120,7 @@ static int video_write_mem(addr_t guest_addr, void * dest, uint_t length, void * struct video_internal * state = (struct video_internal *)dev->private_data; uint_t fb_offset = guest_addr - START_ADDR; uint_t screen_byte_offset = state->screen_offset * BYTES_PER_COL; + uint_t screen_length; PrintDebug("Guest address: %p length = %d, fb_offset=%d, screen_offset=%d\n", (void *)guest_addr, length, fb_offset, screen_byte_offset); @@ -136,7 +137,11 @@ static int video_write_mem(addr_t guest_addr, void * dest, uint_t length, void * if (state->ops) { PrintDebug("\tcalling update_screen()\n"); - state->ops->update_screen(x, y, length, state->private_data); + + /* avoid updates past end of screen */ + screen_length = SCREEN_SIZE - screen_byte_offset; + if (screen_length > length) screen_length = length; + state->ops->update_screen(x, y, screen_length, state->private_data); } } diff --git a/palacios/src/devices/curses_cons.c b/palacios/src/devices/curses_cons.c new file mode 100644 index 0000000..b18b9d9 --- /dev/null +++ b/palacios/src/devices/curses_cons.c @@ -0,0 +1,223 @@ +/* + * This file is part of the Palacios Virtual Machine Monitor developed + * by the V3VEE Project with funding from the United States National + * Science Foundation and the Department of Energy. + * + * The V3VEE Project is a joint project between Northwestern University + * and the University of New Mexico. You can find out more at + * http://www.v3vee.org + * + * Copyright (c) 2009, Robert Deloatch + * Copyright (c) 2009, Steven Jaconette + * Copyright (c) 2009, The V3VEE Project + * All rights reserved. + * + * Author: Erik van der Kouwe (vdkouwe@cs.vu.nl) + * + * This is free software. You are permitted to use, + * redistribute, and modify it as specified in the file "V3VEE_LICENSE". + */ + +/* Interface between virtual video card and console */ + +#include +#include +#include +#include +#include +#include +#include + +#include + +#define NUM_ROWS 25 +#define NUM_COLS 80 +#define BYTES_PER_COL 2 +#define BYTES_PER_ROW (NUM_COLS * BYTES_PER_COL) + +#define SCREEN_SIZE (BYTES_PER_ROW * NUM_ROWS) + +struct cons_state +{ + void *tty; + struct vm_device *frontend_dev; +}; + +static int screen_update(uint_t x, uint_t y, uint_t length, void *private_data); + +static uint_t last_offset; + +static int cursor_update(uint_t x, uint_t y, void *private_data) +{ + struct vm_device *dev = (struct vm_device *) private_data; + struct cons_state *state = (struct cons_state *) dev->private_data; + uint_t offset = (x * BYTES_PER_COL) + (y * BYTES_PER_ROW); + uint_t last_x, last_y; + + /* unfortunately Palacios sometimes misses some writes, + * but if they are accompanied by a cursor move we may be able to + * detect this + */ + if (offset < last_offset) last_offset = 0; + if (offset > last_offset) { + last_x = (last_offset % BYTES_PER_ROW) / BYTES_PER_COL; + last_y = last_offset / BYTES_PER_ROW; + screen_update(last_x, last_y, offset - last_offset, private_data); + } + + /* adjust cursor */ + if (V3_TtyCursorSet(state->tty, x, y) < 0) { + PrintError("V3_TtyCursorSet(0x%p, %d, %d) failed\n", state->tty, x, y); + return -1; + } + + /* done with console update */ + if (V3_TtyUpdate(state->tty) < 0) { + PrintError("V3_TtyUpdate(0x%p) failed\n", state->tty); + return -1; + } + + return 0; +} + +static int screen_update(uint_t x, uint_t y, uint_t length, void *private_data) +{ + struct vm_device *dev = (struct vm_device *)private_data; + struct cons_state *state = (struct cons_state *)dev->private_data; + uint_t offset = (x * BYTES_PER_COL) + (y * BYTES_PER_ROW); + uint8_t fb_buf[length]; + int i; + uint_t cur_x = x; + uint_t cur_y = y; + + /* grab frame buffer */ + memset(fb_buf, 0, length); + v3_cons_get_fb(state->frontend_dev, fb_buf, offset, length); + + /* update the screen */ + for (i = 0; i < length; i += 2) + { + uint_t col_index = i; + uint8_t col[2]; + + col[0] = fb_buf[col_index]; // Character + col[1] = fb_buf[col_index + 1]; // Attribute + + /* update current character */ + if (V3_TtyCharacterSet(state->tty, cur_x, cur_y, col[0], col[1]) < 0) { + PrintError("V3_TtyCursorSet(0x%p, %d, %d, %d, %d) failed\n", + state->tty, cur_x, cur_y, col[1], col[0]); + return -1; + } + + // CAUTION: the order of these statements is critical + // cur_y depends on the previous value of cur_x + cur_y = cur_y + ((cur_x + 1) / NUM_COLS); + cur_x = (cur_x + 1) % NUM_COLS; + } + + /* done with console update */ + if (V3_TtyUpdate(state->tty) < 0) { + PrintError("V3_TtyUpdate(0x%p) failed\n", state->tty); + return -1; + } + + /* store offset to catch missing notifications */ + last_offset = offset + length; + + return 0; +} + +static int scroll(int rows, void *private_data) +{ + struct vm_device *dev = (struct vm_device *)private_data; + struct cons_state *state = (struct cons_state *)dev->private_data; + + if (rows < 0) { + /* simply update the screen */ + return screen_update(0, 0, SCREEN_SIZE, private_data); + } + + if (rows > 0) { + /* scroll requested number of lines*/ + if (V3_TtyScroll(state->tty, rows) < 0) { + PrintError("V3_TtyScroll(0x%p, %u) failed\n", state->tty, rows); + return -1; + } + + /* done with console update */ + if (V3_TtyUpdate(state->tty) < 0) { + PrintError("V3_TtyUpdate(0x%p) failed\n", state->tty); + return -1; + } + + last_offset = BYTES_PER_ROW * (NUM_ROWS - 1); + } + + return 0; +} + +static struct v3_console_ops cons_ops = { + .update_screen = screen_update, + .update_cursor = cursor_update, + .scroll = scroll, +}; + +static struct v3_device_ops dev_ops = { + .free = NULL, + .reset = NULL, + .start = NULL, + .stop = NULL, +}; + +static int cons_init(struct guest_info * vm, v3_cfg_tree_t * cfg) +{ + struct cons_state * state; + v3_cfg_tree_t * frontend_cfg; + const char *frontend_tag; + struct vm_device * frontend; + char *name, *ttypath; + + /* read configuration */ + frontend_cfg = v3_cfg_subtree(cfg, "frontend"); + V3_ASSERT(frontend_cfg); + frontend_tag = v3_cfg_val(frontend_cfg, "tag"); + V3_ASSERT(frontend_tag); + frontend = v3_find_dev(vm, frontend_tag); + V3_ASSERT(frontend); + name = v3_cfg_val(cfg, "name"); + + /* allocate state */ + state = (struct cons_state *) V3_Malloc(sizeof(struct cons_state)); + V3_ASSERT(state); + state->frontend_dev = frontend; + ttypath = v3_cfg_val(cfg, "tty"); + V3_ASSERT(ttypath); + + /* open tty for screen display */ + state->tty = V3_TtyOpen(ttypath, TTY_OPEN_MODE_READ | TTY_OPEN_MODE_WRITE); + if (!state->tty) { + PrintError("Could not open tty %s\n", ttypath); + V3_Free(state); + return -1; + } + + /* allocate device */ + struct vm_device *dev = v3_allocate_device(name, &dev_ops, state); + V3_ASSERT(dev); + + /* attach device to virtual machine */ + if (v3_attach_device(vm, dev) == -1) { + PrintError("Could not attach device %s\n", name); + V3_Free(state); + return -1; + } + + /* attach to front-end display adapter */ + v3_console_register_cga(frontend, &cons_ops, dev); + + return 0; +} + +device_register("CURSES_CONSOLE", cons_init) + diff --git a/palacios/src/devices/lnx_virtio_nic.c b/palacios/src/devices/lnx_virtio_nic.c index 4874b73..e83a199 100644 --- a/palacios/src/devices/lnx_virtio_nic.c +++ b/palacios/src/devices/lnx_virtio_nic.c @@ -7,14 +7,12 @@ * and the University of New Mexico. You can find out more at * http://www.v3vee.org * - * Copyright (c) 2008, Jack Lange * Copyright (c) 2008, Lei Xia * Copyright (c) 2008, Cui Zheng * Copyright (c) 2008, The V3VEE Project * All rights reserved. * - * Author: Jack Lange - * Lei Xia + * Author: Lei Xia * Cui Zheng * * @@ -28,6 +26,7 @@ #include #include #include +#include #include @@ -37,28 +36,9 @@ #define PrintDebug(fmt, args...) #endif -/* The feature bitmap for virtio net */ -#define VIRTIO_NET_F_CSUM 0 /* Host handles pkts w/ partial csum */ -#define VIRTIO_NET_F_GUEST_CSUM 1 /* Guest handles pkts w/ partial csum */ -#define VIRTIO_NET_F_MAC 5 /* Host has given MAC address. */ -#define VIRTIO_NET_F_GSO 6 /* Host handles pkts w/ any GSO type */ -#define VIRTIO_NET_F_GUEST_TSO4 7 /* Guest can handle TSOv4 in. */ -#define VIRTIO_NET_F_GUEST_TSO6 8 /* Guest can handle TSOv6 in. */ -#define VIRTIO_NET_F_GUEST_ECN 9 /* Guest can handle TSO[6] w/ ECN in. */ -#define VIRTIO_NET_F_GUEST_UFO 10 /* Guest can handle UFO in. */ -#define VIRTIO_NET_F_HOST_TSO4 11 /* Host can handle TSOv4 in. */ -#define VIRTIO_NET_F_HOST_TSO6 12 /* Host can handle TSOv6 in. */ -#define VIRTIO_NET_F_HOST_ECN 13 /* Host can handle TSO[6] w/ ECN in. */ -#define VIRTIO_NET_F_HOST_UFO 14 /* Host can handle UFO in. */ -#define VIRTIO_NET_F_MRG_RXBUF 15 /* Host can merge receive buffers. */ -#define VIRTIO_NET_F_STATUS 16 /* virtio_net_config.status available */ -#define VIRTIO_NET_F_CTRL_VQ 17 /* Control channel available */ -#define VIRTIO_NET_F_CTRL_RX 18 /* Control channel RX mode support */ -#define VIRTIO_NET_F_CTRL_VLAN 19 /* Control channel VLAN filtering */ -#define VIRTIO_NET_F_CTRL_RX_EXTRA 20 /* Extra RX mode control support */ -#define VIRTIO_NET_S_LINK_UP 1 /* Link is up */ +//#define VIRTIO_NIC_PROFILE -/* Maximum packet size we can receive from tap device: header + 64k */ +#define VIRTIO_NET_S_LINK_UP 1 /* Link is up */ #define VIRTIO_NET_MAX_BUFSIZE (sizeof(struct virtio_net_hdr) + (64 << 10)) struct virtio_net_hdr { @@ -66,7 +46,6 @@ struct virtio_net_hdr { #define VIRTIO_NET_HDR_GSO_NONE 0 /* Not a GSO frame */ uint8_t gso_type; - uint16_t hdr_len; /* Ethernet + IP + tcp/udp hdrs */ uint16_t gso_size; /* Bytes to append to hdr_len per frame */ uint16_t csum_start; /* Position to start checksumming from */ @@ -74,7 +53,7 @@ struct virtio_net_hdr { }__attribute__((packed)); -#define QUEUE_SIZE 1024 +#define QUEUE_SIZE 4096 #define CTRL_QUEUE_SIZE 64 #define ETH_ALEN 6 @@ -94,6 +73,7 @@ struct virtio_net_state { struct virtio_net_config net_cfg; struct virtio_config virtio_cfg; + struct vm_device * dev; struct pci_device * pci_dev; int io_range_size; @@ -101,6 +81,9 @@ struct virtio_net_state { struct virtio_queue tx_vq; //index 1, svq in Linux virtio driver, handle packet from guest struct virtio_queue ctrl_vq; //index 2, ctrol info from guest + ulong_t pkt_sent, pkt_recv, pkt_drop; + struct gen_queue * inpkt_q; + struct v3_dev_net_ops * net_ops; void * backend_data; @@ -108,15 +91,14 @@ struct virtio_net_state { struct list_head dev_link; }; +struct eth_pkt { + uint32_t size; //size of data + char data[ETHERNET_PACKET_LEN]; +}__attribute__((packed)); -#ifdef CONFIG_DEBUG_VIRTIO_NET - -static void print_packet(uchar_t *pkt, int size) { - PrintDebug("Virtio Nic: print_data_packet: size: %d\n", size); - v3_hexdump(pkt, size, NULL, 0); -} - -#endif +//temporary store all of the virtio states +static struct virtio_net_state *temp_net_states[4]; +static int net_idx = 0; #if 0 //Temporarly for debug @@ -161,7 +143,7 @@ static int virtio_free(struct vm_device * dev) return -1; } -static int virtio_reset(struct virtio_net_state * virtio) +static int virtio_init_state(struct virtio_net_state * virtio) { virtio->rx_vq.ring_desc_addr = 0; virtio->rx_vq.ring_avail_addr = 0; @@ -185,6 +167,11 @@ static int virtio_reset(struct virtio_net_state * virtio) //virtio->virtio_cfg.status = VIRTIO_NET_S_LINK_UP; virtio->virtio_cfg.pci_isr = 0; + virtio->pkt_sent = virtio->pkt_recv = virtio->pkt_drop = 0; + + virtio->inpkt_q = v3_create_queue(); + v3_init_queue(virtio->inpkt_q); + return 0; } @@ -203,10 +190,6 @@ static int pkt_write(struct virtio_net_state * virtio, struct vring_desc * buf_d return -1; } - PrintDebug("Length=%d\n", buf_desc->length); - //PrintDebug("Buffer Descriptor (ptr=%p) gpa=%p, len=%d, flags=%x, next=%d, buf address: %p, send address: %p\n", buf_desc, - //(void *)(buf_desc->addr_gpa), buf_desc->length, buf_desc->flags, buf_desc->next, buf, virtio->net_ops->send); - if (virtio->net_ops->send(buf, len, (void *)virtio, NULL) == -1) { return -1; } @@ -238,7 +221,6 @@ static int copy_data_to_desc(struct virtio_net_state * virtio_state, struct vrin PrintError("Could not translate buffer address\n"); return -1; } - len = (desc->length < buf_len)?desc->length:buf_len; memcpy(desc_buf, buf, len); @@ -254,83 +236,69 @@ static int send_pkt_to_guest(struct virtio_net_state * virtio, uchar_t * buf, ui uint32_t hdr_len = sizeof(struct virtio_net_hdr); uint32_t data_len = size; uint32_t offset = 0; + + PrintDebug("VIRTIO NIC: sending packet to virtio nic %p, size:%d", virtio, size); - //PrintDebug("VIRTIO NIC: Handle RX: cur_index=%d (mod=%d), avail_index=%d\n", - //q->cur_avail_idx, q->cur_avail_idx % q->queue_size, q->avail->index); - - PrintError("VIRTIO NIC: sending packet to net_state %p, size:%d", virtio, size); - + virtio->pkt_recv ++; if (!raw) { data_len -= hdr_len; } build_receive_header(&hdr, buf, 1); - //queue is not set yet if (q->ring_avail_addr == 0) { PrintError("Queue is not set\n"); return -1; } - if (q->cur_avail_idx < q->avail->index) { + if (q->last_avail_idx > q->avail->index) + q->idx_overflow = true; + q->last_avail_idx = q->avail->index; + + if (q->cur_avail_idx < q->avail->index || (q->idx_overflow && q->cur_avail_idx < q->avail->index+65536)){ addr_t hdr_addr = 0; uint16_t hdr_idx = q->avail->ring[q->cur_avail_idx % q->queue_size]; uint16_t buf_idx = 0; struct vring_desc * hdr_desc = NULL; - //PrintDebug("Descriptor index=%d\n", q->cur_avail_idx % q->queue_size); - hdr_desc = &(q->desc[hdr_idx]); - - //PrintDebug("Header Descriptor (ptr=%p) gpa=%p, len=%d, flags=%x, next=%d\n", hdr_desc, - //(void *)(hdr_desc->addr_gpa), hdr_desc->length, hdr_desc->flags, hdr_desc->next); - if (guest_pa_to_host_va(virtio->virtio_dev->vm, hdr_desc->addr_gpa, &(hdr_addr)) == -1) { PrintError("Could not translate receive buffer address\n"); return -1; } - //copy header to the header descriptor memcpy((void *)hdr_addr, &hdr, sizeof(struct virtio_net_hdr)); - - //Zheng 01/02/2010: zero payload if (offset >= data_len) { hdr_desc->flags &= ~VIRTIO_NEXT_FLAG; } - //copy data to the next descriptors - //Zheng 01/02/2010: put data into the next descriptor, rather than 0! for (buf_idx = hdr_desc->next; offset < data_len; buf_idx = q->desc[hdr_idx].next) { - // for (buf_idx = 0; offset < data_len; buf_idx = q->desc[hdr_idx].next) { struct vring_desc * buf_desc = &(q->desc[buf_idx]); uint32_t len = 0; - //Zheng 01/02/2010: commented this - we need to check - // if there still is some data left - //buf_desc->flags = VIRTIO_NEXT_FLAG; - - //PrintError("JACK: copying packet to up desc (len = %d)\n", data_len - offset); - //v3_hexdump(buf + offset, data_len - offset, NULL, 0); - - len = copy_data_to_desc(virtio, buf_desc, buf + offset, data_len - offset); - + len = copy_data_to_desc(virtio, buf_desc, buf + offset, data_len - offset); offset += len; - - //Zheng 01/02/2010: check if there still is some data left if (offset < data_len) { buf_desc->flags = VIRTIO_NEXT_FLAG; } - - buf_desc->length = len; // TODO: do we need this? - //PrintError("JACK: setting buffer descriptor length to %d)\n", buf_desc->length); + buf_desc->length = len; } - q->used->ring[q->used->index % q->queue_size].id = q->avail->ring[q->cur_avail_idx % q->queue_size]; q->used->ring[q->used->index % q->queue_size].length = data_len + hdr_len; // This should be the total length of data sent to guest (header+pkt_data) - q->used->index++; + + int last_idx = q->cur_avail_idx; q->cur_avail_idx++; + if (q->cur_avail_idx < last_idx) + q->idx_overflow = false; + } else { + virtio->pkt_drop++; + +#ifdef VIRTIO_NIC_PROFILE + PrintError("Virtio NIC: %p, one pkt dropped receieved: %ld, dropped: %ld, sent: %ld curidx: %d, avaiIdx: %d\n", + virtio, virtio->pkt_recv, virtio->pkt_drop, virtio->pkt_sent, q->cur_avail_idx, q->avail->index); +#endif } if (!(q->avail->flags & VIRTIO_NO_IRQ_FLAG)) { @@ -339,6 +307,14 @@ static int send_pkt_to_guest(struct virtio_net_state * virtio, uchar_t * buf, ui virtio->virtio_cfg.pci_isr = 0x1; } +#ifdef VIRTIO_NIC_PROFILE + if ((virtio->pkt_recv % 10000) == 0){ + PrintError("Virtio NIC: %p, receieved: %ld, dropped: %ld, sent: %ld\n", + virtio, virtio->pkt_recv, virtio->pkt_drop, virtio->pkt_sent); + } +#endif + + return offset; } @@ -348,19 +324,74 @@ int virtio_send(struct guest_info * vm, char *dev_name, uchar_t * buf, uint_t si struct vm_device * virtio_dev = v3_find_dev(vm, dev_name); // TODO: how to get virtio net state from device?? - // this is not right now struct virtio_net_state * virtio_state = (struct virtio_net_state *)virtio_dev->private_data; return send_pkt_to_guest(virtio_state, buf, size, 1, NULL); } +int __virtio_sendto_buf(struct virtio_net_state * net_state, uchar_t * buf, uint_t size) { + struct eth_pkt *pkt; + + pkt = (struct eth_pkt *)V3_Malloc(sizeof(struct eth_pkt)); + if(pkt == NULL){ + PrintError("Virtio NIC: Memory allocate fails\n"); + return -1; + } + + pkt->size = size; + memcpy(pkt->data, buf, size); + v3_enqueue(net_state->inpkt_q, (addr_t)pkt); + + PrintDebug("Virtio NIC: __virtio_sendto_buf: transmitting packet: (size:%d)\n", (int)pkt->size); + + return pkt->size; +} + + static int __virtio_dev_send(uchar_t * buf, uint32_t size, void *private_data) { struct virtio_net_state *virtio_state = (struct virtio_net_state *)private_data; - return send_pkt_to_guest(virtio_state, buf, size, 1, NULL); + return __virtio_sendto_buf(virtio_state, buf, size); +} + +/* + *called in svm/vmx handler + *iteative handled the unsent packet in incoming packet queues for + *all virtio nic devices in this guest + */ +int v3_virtionic_pktprocess(struct guest_info * info) +{ + struct eth_pkt *pkt = NULL; + struct virtio_net_state *net_state; + int i; + + for (i = 0; i < net_idx; i++) { + while (1) { + net_state = temp_net_states[i]; + if(net_state->dev->vm != info) + break; + + pkt = (struct eth_pkt *)v3_dequeue(net_state->inpkt_q); + if(pkt == NULL) + break; + + if (send_pkt_to_guest(net_state, pkt->data, pkt->size, 1, NULL)) { + PrintDebug("Virtio NIC: %p In pkt_handle: send one packet! pt length %d\n", + net_state, (int)pkt->size); + } else { + PrintDebug("Virtio NIC: %p In pkt_handle: Fail to send one packet, pt length %d, discard it!\n", + net_state, (int)pkt->size); + } + + V3_Free(pkt); + } + } + + return 0; } + static int get_desc_count(struct virtio_queue * q, int index) { struct vring_desc * tmp_desc = &(q->desc[index]); int cnt = 1; @@ -381,15 +412,15 @@ static int handle_ctrl(struct virtio_net_state * dev) { //get packet from guest static int handle_pkt_tx(struct virtio_net_state * virtio_state) { - //struct virtio_net_state *virtio = (struct virtio_net_state *)dev->private_data; struct virtio_queue * q = &(virtio_state->tx_vq); - - //PrintDebug("VIRTIO NIC pkt_tx: cur_index=%d (mod=%d), avail_index=%d\n", - // q->cur_avail_idx, q->cur_avail_idx % q->queue_size, q->avail->index); - struct virtio_net_hdr * hdr = NULL; - while (q->cur_avail_idx < q->avail->index) { + if (q->avail->index < q->last_avail_idx) + q->idx_overflow = true; + q->last_avail_idx = q->avail->index; + + while (q->cur_avail_idx < q->avail->index || + (q->idx_overflow && q->cur_avail_idx < (q->avail->index + 65536))) { struct vring_desc * hdr_desc = NULL; addr_t hdr_addr = 0; uint16_t desc_idx = q->avail->ring[q->cur_avail_idx % q->queue_size]; @@ -397,30 +428,17 @@ static int handle_pkt_tx(struct virtio_net_state * virtio_state) uint32_t req_len = 0; int i = 0; - //PrintDebug("Descriptor Count=%d, index=%d\n", desc_cnt, q->cur_avail_idx % q->queue_size); - hdr_desc = &(q->desc[desc_idx]); - - //PrintDebug("Header Descriptor (ptr=%p) gpa=%p, len=%d, flags=%x, next=%d\n", hdr_desc, - //(void *)(hdr_desc->addr_gpa), hdr_desc->length, hdr_desc->flags, hdr_desc->next); - if (guest_pa_to_host_va(virtio_state->virtio_dev->vm, hdr_desc->addr_gpa, &(hdr_addr)) == -1) { PrintError("Could not translate block header address\n"); return -1; } hdr = (struct virtio_net_hdr*)hdr_addr; - - //PrintDebug("NIC Op Hdr (ptr=%p) header len =%x\n", (void *)hdr_addr, (int)hdr->hdr_len); - desc_idx = hdr_desc->next; for (i = 0; i < desc_cnt - 1; i++) { struct vring_desc * buf_desc = &(q->desc[desc_idx]); - - //PrintDebug("Buffer Descriptor (ptr=%p) gpa=%p, len=%d, flags=%x, next=%d\n", buf_desc, - //(void *)(buf_desc->addr_gpa), buf_desc->length, buf_desc->flags, buf_desc->next); - if (pkt_write(virtio_state, buf_desc) == -1) { PrintError("Error handling nic operation\n"); return -1; @@ -429,24 +447,31 @@ static int handle_pkt_tx(struct virtio_net_state * virtio_state) req_len += buf_desc->length; desc_idx = buf_desc->next; } + virtio_state->pkt_sent ++; q->used->ring[q->used->index % q->queue_size].id = q->avail->ring[q->cur_avail_idx % q->queue_size]; q->used->ring[q->used->index % q->queue_size].length = req_len; // What do we set this to???? - q->used->index++; - q->cur_avail_idx++; + + int last_idx = q->cur_avail_idx; + q->cur_avail_idx ++; + if (q->cur_avail_idx < last_idx) + q->idx_overflow = false; } if (!(q->avail->flags & VIRTIO_NO_IRQ_FLAG)) { - //PrintDebug("Raising IRQ %d\n", virtio_state->pci_dev->config_header.intr_line); v3_pci_raise_irq(virtio_state->virtio_dev->pci_bus, 0, virtio_state->pci_dev); virtio_state->virtio_cfg.pci_isr = 0x1; } +#ifdef VIRTIO_NIC_PROFILE + if(virtio_state->pkt_sent % 10000 == 0) + PrintError("Virtio NIC: %p, pkt_sent: %ld\n", virtio_state, virtio_state->pkt_sent); +#endif + return 0; } - static int virtio_setup_queue(struct virtio_net_state * virtio_state, struct virtio_queue * queue, addr_t pfn, addr_t page_addr) { queue->pfn = pfn; @@ -458,19 +483,16 @@ static int virtio_setup_queue(struct virtio_net_state * virtio_state, struct vir // round up to next page boundary. queue->ring_used_addr = (queue->ring_used_addr + 0xfff) & ~0xfff; - if (guest_pa_to_host_va(virtio_state->virtio_dev->vm, queue->ring_desc_addr, (addr_t *)&(queue->desc)) == -1) { PrintError("Could not translate ring descriptor address\n"); return -1; } - if (guest_pa_to_host_va(virtio_state->virtio_dev->vm, queue->ring_avail_addr, (addr_t *)&(queue->avail)) == -1) { PrintError("Could not translate ring available address\n"); return -1; } - if (guest_pa_to_host_va(virtio_state->virtio_dev->vm, queue->ring_used_addr, (addr_t *)&(queue->used)) == -1) { PrintError("Could not translate ring used address\n"); return -1; @@ -487,41 +509,33 @@ static int virtio_setup_queue(struct virtio_net_state * virtio_state, struct vir return 0; } - - static int virtio_io_write(uint16_t port, void * src, uint_t length, void * private_data) { //struct vm_device * dev = (struct vm_device *)private_data; struct virtio_net_state * virtio = (struct virtio_net_state *)private_data; int port_idx = port % virtio->io_range_size; - PrintDebug("VIRTIO NIC %p Write for port %d (index=%d) len=%d, value=%x\n", private_data, port, port_idx, length, *(uint32_t *)src); - switch (port_idx) { case GUEST_FEATURES_PORT: if (length != 4) { PrintError("Illegal write length for guest features\n"); return -1; - } - + } virtio->virtio_cfg.guest_features = *(uint32_t *)src; PrintDebug("Setting Guest Features to %x\n", virtio->virtio_cfg.guest_features); - break; + case VRING_PG_NUM_PORT: - if (length != 4) { PrintError("Illegal write length for page frame number\n"); return -1; } - addr_t pfn = *(uint32_t *)src; addr_t page_addr = (pfn << VIRTIO_PAGE_SHIFT); uint16_t queue_idx = virtio->virtio_cfg.vring_queue_selector; - switch (queue_idx) { case 0: virtio_setup_queue(virtio, &virtio->rx_vq, pfn, page_addr); @@ -531,29 +545,24 @@ static int virtio_io_write(uint16_t port, void * src, uint_t length, void * priv break; case 2: virtio_setup_queue(virtio, &virtio->ctrl_vq, pfn, page_addr); - break; - + break; default: break; } - break; + case VRING_Q_SEL_PORT: virtio->virtio_cfg.vring_queue_selector = *(uint16_t *)src; - if (virtio->virtio_cfg.vring_queue_selector > 2) { PrintError("Virtio NIC device only uses 3 queue, selected %d\n", virtio->virtio_cfg.vring_queue_selector); return -1; } - break; + case VRING_Q_NOTIFY_PORT: { - uint16_t queue_idx = *(uint16_t *)src; - - //PrintDebug("Handling Kick\n"); - + uint16_t queue_idx = *(uint16_t *)src; if (queue_idx == 0){ PrintDebug("receive queue notification 0, packet get by Guest\n"); } else if (queue_idx == 1){ @@ -569,18 +578,16 @@ static int virtio_io_write(uint16_t port, void * src, uint_t length, void * priv } else { PrintError("Virtio NIC device only uses 3 queue, selected %d\n", queue_idx); - } - + } break; } + case VIRTIO_STATUS_PORT: virtio->virtio_cfg.status = *(uint8_t *)src; - if (virtio->virtio_cfg.status == 0) { PrintDebug("Resetting device\n"); - virtio_reset(virtio); + virtio_init_state(virtio); } - break; case VIRTIO_ISR_PORT: @@ -594,7 +601,6 @@ static int virtio_io_write(uint16_t port, void * src, uint_t length, void * priv return length; } - static int virtio_io_read(uint16_t port, void * dst, uint_t length, void * private_data) { //struct vm_device * dev = (struct vm_device *)private_data; @@ -611,19 +617,15 @@ static int virtio_io_read(uint16_t port, void * dst, uint_t length, void * priva PrintError("Illegal read length for host features\n"); return -1; } - *(uint32_t *)dst = virtio->virtio_cfg.host_features; - PrintDebug("value=0x%x\n", *(uint32_t *)dst); - break; - case VRING_PG_NUM_PORT: + case VRING_PG_NUM_PORT: if (length != 4) { PrintError("Illegal read length for page frame number\n"); return -1; } - switch (queue_idx) { case 0: *(uint32_t *)dst = virtio->rx_vq.pfn; @@ -637,17 +639,14 @@ static int virtio_io_read(uint16_t port, void * dst, uint_t length, void * priva default: break; } - PrintDebug(", value=0x%x\n", *(uint32_t *)dst); - break; - case VRING_SIZE_PORT: + case VRING_SIZE_PORT: if (length != 2) { PrintError("Illegal read length for vring size\n"); return -1; } - switch (queue_idx) { case 0: *(uint16_t *)dst = virtio->rx_vq.queue_size; @@ -661,29 +660,23 @@ static int virtio_io_read(uint16_t port, void * dst, uint_t length, void * priva default: break; } - PrintDebug("queue index: %d, value=0x%x\n", (int)queue_idx, *(uint16_t *)dst); - break; - case VIRTIO_STATUS_PORT: + case VIRTIO_STATUS_PORT: if (length != 1) { PrintError("Illegal read length for status\n"); return -1; } - *(uint8_t *)dst = virtio->virtio_cfg.status; - PrintDebug(", value=0x%x\n", *(uint8_t *)dst); break; - + case VIRTIO_ISR_PORT: *(uint8_t *)dst = virtio->virtio_cfg.pci_isr; virtio->virtio_cfg.pci_isr = 0; v3_pci_lower_irq(virtio->virtio_dev->pci_bus, 0, virtio->pci_dev); - - PrintDebug(", value=0x%x\n", *(uint8_t *)dst); - + PrintDebug(", value=0x%x\n", *(uint8_t *)dst); break; default: @@ -702,7 +695,6 @@ static struct v3_device_ops dev_ops = { .stop = NULL, }; - static int register_dev(struct virtio_dev_state * virtio, struct virtio_net_state * net_state) { struct pci_device * pci_dev = NULL; @@ -713,7 +705,6 @@ static int register_dev(struct virtio_dev_state * virtio, struct virtio_net_stat // This gets the number of ports, rounded up to a power of 2 net_state->io_range_size = 1; // must be a power of 2 - while (tmp_ports > 0) { tmp_ports >>= 1; net_state->io_range_size <<= 1; @@ -725,7 +716,6 @@ static int register_dev(struct virtio_dev_state * virtio, struct virtio_net_stat net_state->io_range_size >>= 1; } - for (i = 0; i < 6; i++) { bars[i].type = PCI_BAR_NONE; } @@ -735,7 +725,6 @@ static int register_dev(struct virtio_dev_state * virtio, struct virtio_net_stat bars[0].type = PCI_BAR_IO; bars[0].default_base_port = -1; bars[0].num_ports = net_state->io_range_size; - bars[0].io_read = virtio_io_read; bars[0].io_write = virtio_io_write; bars[0].private_data = net_state; @@ -758,33 +747,23 @@ static int register_dev(struct virtio_dev_state * virtio, struct virtio_net_stat pci_dev->config_header.device_id = VIRTIO_NET_DEV_ID; pci_dev->config_header.class = PCI_CLASS_NETWORK; - pci_dev->config_header.subclass = PCI_NET_SUBCLASS_OTHER; - - pci_dev->config_header.subsystem_id = VIRTIO_NET_SUBDEVICE_ID;; - + pci_dev->config_header.subclass = PCI_NET_SUBCLASS_OTHER; + pci_dev->config_header.subsystem_id = VIRTIO_NET_SUBDEVICE_ID; pci_dev->config_header.intr_pin = 1; - pci_dev->config_header.max_latency = 1; // ?? (qemu does it...) - - net_state->pci_dev = pci_dev; - //net_state->pci_bus = pci_bus; - + net_state->pci_dev = pci_dev; net_state->virtio_cfg.host_features = 0; //no features support now - net_state->rx_vq.queue_size = QUEUE_SIZE; net_state->tx_vq.queue_size = QUEUE_SIZE; net_state->ctrl_vq.queue_size = CTRL_QUEUE_SIZE; - net_state->virtio_dev = virtio; - - virtio_reset(net_state); + + virtio_init_state(net_state); return 0; } - - static int connect_fn(struct guest_info * info, void * frontend_data, struct v3_dev_net_ops * ops, @@ -810,13 +789,8 @@ static int connect_fn(struct guest_info * info, //used when virtio_nic get a packet from guest and send it to the backend // send packet to all of the virtio nic devices other than the sender static int vnet_send(uint8_t * buf, uint32_t len, void * private_data, struct vm_device *dest_dev){ - PrintDebug("Virito NIC: In vnet_send: guest net state %p\n", private_data); -#ifdef CONFIG_DEBUG_VIRTIO_NET - print_packet(buf, len); -#endif - v3_vnet_send_rawpkt(buf, len, private_data); return 0; } @@ -830,10 +804,6 @@ static int vnet_receive(uint8_t * buf, uint32_t count, void * private_data, stru static int virtio_input(uchar_t * buf, uint_t len, void * private_data){ PrintDebug("Virito NIC: In virtio_input: guest net state %p\n", private_data); -#ifdef CONFIG_DEBUG_VIRTIO_NET - print_packet(buf, len); -#endif - return __virtio_dev_send(buf, len, private_data); } @@ -842,36 +812,35 @@ static int virtio_input(uchar_t * buf, uint_t len, void * private_data){ void register_virtio_to_vnet(struct vm_device *dev, char *dev_name, uchar_t mac[6]){ - struct virtio_net_state * net_state = (struct virtio_net_state *)V3_Malloc(sizeof(struct virtio_net_state)); - memset(net_state, 0, sizeof(struct virtio_net_state)); - + struct virtio_net_state * net_state; struct virtio_dev_state *virtio_state = (struct virtio_dev_state *)dev->private_data; + //uchar_t tapmac[6] = {0x00,0x02,0x55,0x67,0x42,0x39}; //for Intel-VT test HW + uchar_t tapmac[6] = {0x6e,0xa8,0x75,0xf4,0x82,0x95}; + uchar_t dstmac[6] = {0xff,0xff,0xff,0xff,0xff,0xff}; + uchar_t zeromac[6] = {0,0,0,0,0,0}; - net_state->net_ops = (struct v3_dev_net_ops *)V3_Malloc(sizeof(struct v3_dev_net_ops)); + net_state = (struct virtio_net_state *)V3_Malloc(sizeof(struct virtio_net_state)); + memset(net_state, 0, sizeof(struct virtio_net_state)); + net_state->net_ops = (struct v3_dev_net_ops *)V3_Malloc(sizeof(struct v3_dev_net_ops)); net_state->net_ops->send = &vnet_send; net_state->net_ops->receive = &vnet_receive; + net_state->dev = dev; register_dev(virtio_state, net_state); - - PrintDebug("Virtio NIC After register Device %s: queue size: %d, %d\n", dev->name, - net_state->rx_vq.queue_size, net_state->tx_vq.queue_size); - PrintDebug("VNET: connect virtio nic state %p to vnet\n", net_state); - - //add a device link to link table + PrintDebug("Virtio NIC register Device %s: queue size: %d, %d\n", dev->name, + net_state->rx_vq.queue_size, net_state->tx_vq.queue_size); + PrintDebug("Virtio NIC: connect virtio device %s, state %p, to vnet\n", dev->name, net_state); + int idx = vnet_register_device(dev, dev_name, mac, &virtio_input, net_state); - - uchar_t srcmac[6] = {0x00,0x02,0x55,0x67,0x42,0x39}; - uchar_t dstmac[6] = {0xff,0xff,0xff,0xff,0xff,0xff}; - uchar_t zeromac[6] = {0,0,0,0,0,0}; - vnet_add_route_entry(zeromac, dstmac, MAC_ANY, MAC_NONE, idx, LINK_INTERFACE, -1, LINK_INTERFACE); - if (idx == 0) - vnet_add_route_entry(zeromac, srcmac, MAC_ANY, MAC_NONE, idx, LINK_INTERFACE, -1, LINK_INTERFACE); - if (idx == 1) - vnet_add_route_entry(srcmac, zeromac, MAC_NONE, MAC_ANY, idx, LINK_INTERFACE, -1, LINK_INTERFACE); - + if (!strcmp(dev_name, "net_virtiodom0")) + vnet_add_route_entry(zeromac, tapmac, MAC_ANY, MAC_NONE, idx, LINK_INTERFACE, -1, LINK_INTERFACE); + if (!strcmp(dev_name, "net_virtio")) + vnet_add_route_entry(tapmac, zeromac, MAC_NONE, MAC_ANY, idx, LINK_INTERFACE, -1, LINK_INTERFACE); + + temp_net_states[net_idx ++] = net_state; } #endif @@ -906,8 +875,7 @@ static int virtio_init(struct guest_info * vm, v3_cfg_tree_t * cfg) { return -1; } - - //for temporary testing, add a backend + //for temporary testing, virtio bridge in Palacios #if 0 struct virtio_net_state * net_state = (struct virtio_net_state *)V3_Malloc(sizeof(struct virtio_net_state)); @@ -925,15 +893,21 @@ static int virtio_init(struct guest_info * vm, v3_cfg_tree_t * cfg) { temp_net_states[net_idx ++] = net_state; - PrintDebug("Net_states: 0: %p, 1: %p, 2: %p\n", temp_net_states[0], temp_net_states[1], temp_net_states[2]); + PrintDebug("Net_states: 0: %p, 1: %p, 2: %p\n", + temp_net_states[0], + temp_net_states[1], + temp_net_states[2]); #endif #if 1 //test interface between vnet & virtio-nic - uchar_t mac[6] = {0,0,0,0,0,0}; - register_virtio_to_vnet(dev, name,mac); + register_virtio_to_vnet(dev, name, mac); + PrintDebug("Net_states: 0: %p, 1: %p, 2: %p\n", + temp_net_states[0], + temp_net_states[1], + temp_net_states[2]); #endif return 0; diff --git a/palacios/src/devices/netdisk.c b/palacios/src/devices/netdisk.c index 3052a9b..006776a 100644 --- a/palacios/src/devices/netdisk.c +++ b/palacios/src/devices/netdisk.c @@ -261,7 +261,7 @@ static int socket_init(struct disk_state * disk) { return -1; } - PrintDebug("Capacity: %p\n", (void *)(disk->capacity)); + PrintDebug("Capacity: %p\n", (void *)(addr_t)disk->capacity); } diff --git a/palacios/src/devices/telnet_cons.c b/palacios/src/devices/telnet_cons.c index 30e07d1..d0c2b1f 100644 --- a/palacios/src/devices/telnet_cons.c +++ b/palacios/src/devices/telnet_cons.c @@ -29,8 +29,10 @@ #include #include -#include +#include +#if 0 #include +#endif #define NUM_ROWS 25 #define NUM_COLS 80 @@ -354,7 +356,7 @@ static int screen_update(uint_t x, uint_t y, uint_t length, void * private_data) return ret; } -static int scroll(uint_t rows, void * private_data) { +static int scroll(int rows, void * private_data) { struct vm_device * dev = (struct vm_device *)private_data; struct cons_state * state = (struct cons_state *)dev->private_data; addr_t irq_state = 0; @@ -516,6 +518,8 @@ static int cons_server(void * arg) { state->connected = 0; V3_Close_Socket(state->client_fd); } + + return -1; } diff --git a/palacios/src/palacios/Makefile b/palacios/src/palacios/Makefile index c669cd0..edd13a0 100644 --- a/palacios/src/palacios/Makefile +++ b/palacios/src/palacios/Makefile @@ -56,6 +56,7 @@ obj-$(CONFIG_VMX) += vmx.o \ obj-$(CONFIG_INSTRUMENT_VMM) += vmm_instrument.o obj-$(CONFIG_TELEMETRY) += vmm_telemetry.o obj-$(CONFIG_SOCKET) += vmm_socket.o +obj-$(CONFIG_CONSOLE) += vmm_console.o obj-$(CONFIG_VNET) += vmm_vnet.o obj-$(CONFIG_SYMBIOTIC) += vmm_sym_iface.o diff --git a/palacios/src/palacios/svm.c b/palacios/src/palacios/svm.c index 9f72514..ca8fc18 100644 --- a/palacios/src/palacios/svm.c +++ b/palacios/src/palacios/svm.c @@ -309,7 +309,7 @@ static int update_irq_exit_state(struct guest_info * info) { // Interrupt was taken fully vectored info->intr_state.irq_started = 0; - } else { + } else if ((info->intr_state.irq_started == 1) && (guest_ctrl->exit_int_info.valid == 1)) { #ifdef CONFIG_DEBUG_INTERRUPTS PrintDebug("EXIT INT INFO is set (vec=%d)\n", guest_ctrl->exit_int_info.vector); #endif @@ -322,6 +322,12 @@ static int update_irq_exit_state(struct guest_info * info) { static int update_irq_entry_state(struct guest_info * info) { vmcb_ctrl_t * guest_ctrl = GET_VMCB_CTRL_AREA((vmcb_t*)(info->vmm_data)); + + if (info->intr_state.irq_pending == 0) { + guest_ctrl->guest_ctrl.V_IRQ = 0; + guest_ctrl->guest_ctrl.V_INTR_VECTOR = 0; + } + if (v3_excp_pending(info)) { uint_t excp = v3_get_excp_number(info); diff --git a/palacios/src/palacios/svm_handler.c b/palacios/src/palacios/svm_handler.c index bd19242..cab4dc6 100644 --- a/palacios/src/palacios/svm_handler.c +++ b/palacios/src/palacios/svm_handler.c @@ -34,15 +34,17 @@ #include #include - -#ifdef CONFIG_TELEMETRY -#include -#endif - #ifdef CONFIG_VNET #include #endif +#ifdef CONFIG_LINUX_VIRTIO_NET + extern int v3_virtionic_pktprocess(struct guest_info * info); +#endif + +#ifdef CONFIG_TELEMETRY +#include +#endif int v3_handle_svm_exit(struct guest_info * info, addr_t exit_code, addr_t exit_info1, addr_t exit_info2) { @@ -52,7 +54,6 @@ int v3_handle_svm_exit(struct guest_info * info, addr_t exit_code, addr_t exit_i } #endif - //PrintDebug("SVM Returned: Exit Code: %x\n",exit_code); switch (exit_code) { @@ -285,10 +286,14 @@ int v3_handle_svm_exit(struct guest_info * info, addr_t exit_code, addr_t exit_i } // END OF SWITCH (EXIT_CODE) + #ifdef CONFIG_VNET v3_vnet_pkt_process(info); #endif +#ifdef CONFIG_LINUX_VIRTIO_NET + v3_virtionic_pktprocess(info); +#endif #ifdef CONFIG_TELEMETRY if (info->enable_telemetry) { @@ -300,6 +305,7 @@ int v3_handle_svm_exit(struct guest_info * info, addr_t exit_code, addr_t exit_i if (exit_code == VMEXIT_INTR) { //PrintDebug("INTR ret IP = %x\n", guest_state->rip); } + return 0; } diff --git a/palacios/src/palacios/svm_lowlevel.S b/palacios/src/palacios/svm_lowlevel.S index 6dcd39b..45c5575 100644 --- a/palacios/src/palacios/svm_lowlevel.S +++ b/palacios/src/palacios/svm_lowlevel.S @@ -74,12 +74,15 @@ v3_svm_launch: addl $4, %esp; + popa; + popf; movl 16(%ebp), %eax; vmload; - popa; - popf; + // We don't detect failures here, so just return 0 + xorl %eax, %eax + pop %ebp; ret diff --git a/palacios/src/palacios/vmcs.c b/palacios/src/palacios/vmcs.c index 9156e23..a77de77 100644 --- a/palacios/src/palacios/vmcs.c +++ b/palacios/src/palacios/vmcs.c @@ -270,24 +270,45 @@ int v3_update_vmcs_host_state(struct guest_info * info) { struct vmx_data * arch_data = (struct vmx_data *)(info->vmm_data); struct v3_msr tmp_msr; +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%cr0, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%cr0, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_CR0, tmp); +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%cr3, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%cr3, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_CR3, tmp); +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%cr4, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%cr4, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_CR4, tmp); @@ -309,40 +330,82 @@ int v3_update_vmcs_host_state(struct guest_info * info) { +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%cs, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%cs, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_CS_SELECTOR, tmp); +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%ss, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%ss, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_SS_SELECTOR, tmp); +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%ds, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%ds, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_DS_SELECTOR, tmp); +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%es, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%es, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_ES_SELECTOR, tmp); +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%fs, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%fs, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_FS_SELECTOR, tmp); +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%gs, %0; " : "=q"(tmp) : ); +#else + __asm__ __volatile__ ( "movl %%gs, %0; " + : "=q"(tmp) + : + ); +#endif vmx_ret |= check_vmcs_write(VMCS_HOST_GS_SELECTOR, tmp); vmx_ret |= check_vmcs_write(VMCS_HOST_TR_SELECTOR, arch_data->host_state.tr.selector); diff --git a/palacios/src/palacios/vmm.c b/palacios/src/palacios/vmm.c index c56356d..402cedf 100644 --- a/palacios/src/palacios/vmm.c +++ b/palacios/src/palacios/vmm.c @@ -25,6 +25,7 @@ #include #include #include +#include #ifdef CONFIG_SVM #include @@ -98,6 +99,10 @@ void Init_V3(struct v3_os_hooks * hooks, int num_cpus) { hooks->call_on_cpu(i, &init_cpu, (void *)(addr_t)i); } } + +#ifdef CONFIG_VNET + v3_init_vnet(); +#endif } diff --git a/palacios/src/palacios/vmm_config.c b/palacios/src/palacios/vmm_config.c index 002cf75..81a37d3 100644 --- a/palacios/src/palacios/vmm_config.c +++ b/palacios/src/palacios/vmm_config.c @@ -41,10 +41,6 @@ #endif -#ifdef CONFIG_VNET -#include -#endif - #include #include @@ -272,10 +268,6 @@ static int pre_config_guest(struct guest_info * info, struct v3_config * config_ v3_init_sym_swap(info); #endif -#ifdef CONFIG_VNET - v3_vnet_init(info); -#endif - if (schedule_hz_str) { sched_hz = atoi(schedule_hz_str); } diff --git a/palacios/src/palacios/vmm_console.c b/palacios/src/palacios/vmm_console.c new file mode 100644 index 0000000..ef142b9 --- /dev/null +++ b/palacios/src/palacios/vmm_console.c @@ -0,0 +1,36 @@ +/* + * This file is part of the Palacios Virtual Machine Monitor developed + * by the V3VEE Project with funding from the United States National + * Science Foundation and the Department of Energy. + * + * The V3VEE Project is a joint project between Northwestern University + * and the University of New Mexico. You can find out more at + * http://www.v3vee.org + * + * Copyright (c) 2010, Jack Lange + * Copyright (c) 2010, Erik van der Kouwe + * Copyright (c) 2010, The V3VEE Project + * All rights reserved. + * + * Author: Jack Lange + * Author: Erik van der Kouwe + * + * This is free software. You are permitted to use, + * redistribute, and modify it as specified in the file "V3VEE_LICENSE". + */ + + +#include +#include +#include +#include + + +struct v3_console_hooks * console_hooks = 0; + +void V3_Init_Console(struct v3_console_hooks * hooks) { + console_hooks = hooks; + PrintDebug("V3 console inited\n"); + + return; +} diff --git a/palacios/src/palacios/vmm_intr.c b/palacios/src/palacios/vmm_intr.c index 656fa1e..bc36d05 100644 --- a/palacios/src/palacios/vmm_intr.c +++ b/palacios/src/palacios/vmm_intr.c @@ -200,6 +200,13 @@ int v3_raise_irq(struct guest_info * info, int irq) { } +void v3_clear_pending_intr(struct guest_info * core) { + struct v3_intr_state * intr_state = &(core->intr_state); + + intr_state->irq_pending = 0; + +} + v3_intr_type_t v3_intr_pending(struct guest_info * info) { struct v3_intr_state * intr_state = &(info->intr_state); diff --git a/palacios/src/palacios/vmm_queue.c b/palacios/src/palacios/vmm_queue.c index 2e1f72f..b06ff73 100644 --- a/palacios/src/palacios/vmm_queue.c +++ b/palacios/src/palacios/vmm_queue.c @@ -22,9 +22,9 @@ void v3_init_queue(struct gen_queue * queue) { queue->num_entries = 0; INIT_LIST_HEAD(&(queue->entries)); + v3_lock_init(&queue->lock); } - struct gen_queue * v3_create_queue() { struct gen_queue * tmp_queue = V3_Malloc(sizeof(struct gen_queue)); v3_init_queue(tmp_queue); @@ -34,15 +34,18 @@ struct gen_queue * v3_create_queue() { void v3_enqueue(struct gen_queue * queue, addr_t entry) { struct queue_entry * q_entry = V3_Malloc(sizeof(struct queue_entry)); + v3_lock(queue->lock); q_entry->entry = entry; list_add_tail(&(q_entry->entry_list), &(queue->entries)); queue->num_entries++; + v3_unlock(queue->lock); } addr_t v3_dequeue(struct gen_queue * queue) { addr_t entry_val = 0; + v3_lock(queue->lock); if (!list_empty(&(queue->entries))) { struct list_head * q_entry = queue->entries.next; struct queue_entry * tmp_entry = list_entry(q_entry, struct queue_entry, entry_list); @@ -51,6 +54,7 @@ addr_t v3_dequeue(struct gen_queue * queue) { list_del(q_entry); V3_Free(tmp_entry); } + v3_unlock(queue->lock); return entry_val; } diff --git a/palacios/src/palacios/vmm_vnet.c b/palacios/src/palacios/vmm_vnet.c index d2daa7e..dffb6d4 100644 --- a/palacios/src/palacios/vmm_vnet.c +++ b/palacios/src/palacios/vmm_vnet.c @@ -8,16 +8,12 @@ * http://www.v3vee.org * * Copyright (c) 2009, Lei Xia - * Copyright (c) 2009, Yuan Tang - * Copyright (c) 2009, Jack Lange - * Copyright (c) 2009, Peter Dinda + * Copyright (c) 2009, Yuan Tang * Copyright (c) 2009, The V3VEE Project * All rights reserved. * * Author: Lei Xia * Yuan Tang - * Jack Lange - * Peter Dinda #include #include +#include #ifndef CONFIG_DEBUG_VNET #undef PrintDebug #define PrintDebug(fmt, args...) #endif - -struct ethernet_pkt { - uint32_t size; //size of data - uint16_t type; - struct udp_link_header ext_hdr; //possible externel header to applied to data before sent - char data[ETHERNET_PACKET_LEN]; -}__attribute__((packed)); - - // 14 (ethernet frame) + 20 bytes struct in_pkt_header { uchar_t ethernetdest[6]; @@ -49,44 +37,40 @@ struct in_pkt_header { char ip[20]; }__attribute__((packed)); -#define VNET_INITAB_HCALL 0xca00 // inital hypercall id - #define MAX_LINKS 10 #define MAX_ROUTES 10 #define HASH_KEY_LEN 16 #define MIN_CACHE_SIZE 100 -static const uint_t hash_key_size = 16; +#define HASH_KEY_SIZE 16 struct link_table { struct link_entry * links[MAX_LINKS]; uint16_t size; + v3_lock_t lock; }__attribute__((packed)); struct routing_table { struct routing_entry * routes[MAX_ROUTES]; uint16_t size; + v3_lock_t lock; }__attribute__((packed)); -static struct link_table g_links; -static struct routing_table g_routes; -static struct gen_queue * g_inpkt_q; - -/* Hash key format: - * 0-5: src_eth_addr - * 6-11: dest_eth_addr - * 12: src type - * 13-16: src index - */ typedef char * route_hashkey_t; -// This is the hash value, Format: 0: num_matched_routes, 1...n: matches[] -- TY struct route_cache_entry { int num_matched_routes; int * matches; }; -// the route cache -static struct hashtable * g_route_cache; +struct vnet_state_t { + struct link_table g_links; + struct routing_table g_routes; + struct gen_queue * g_inpkt_q; + struct hashtable * g_route_cache; + v3_lock_t cache_lock; +}; + +static struct vnet_state_t g_vnet_state;//global state for vnet static uint16_t ip_xsum(struct ip_header *ip_hdr, int hdr_len){ long sum = 0; @@ -108,27 +92,30 @@ static uint16_t ip_xsum(struct ip_header *ip_hdr, int hdr_len){ return (uint16_t)~sum; } -static inline void ethernet_packet_init(struct ethernet_pkt *pt, uchar_t *data, const size_t size) { +static void ethernet_packet_init(struct ethernet_pkt *pt, + uchar_t *data, + const size_t size) { pt->size = size; + pt->use_header = 0; + memset(&pt->ext_hdr, 0, sizeof(struct udp_link_header)); memcpy(pt->data, data, size); } -static uint_t hash_from_key_fn(addr_t hashkey) { - uint8_t * key = (uint8_t *)hashkey; - return v3_hash_buffer(key, HASH_KEY_LEN); +static inline uint_t hash_from_key_fn(addr_t hashkey) { + return v3_hash_buffer((uint8_t *)hashkey, HASH_KEY_LEN); } -static int hash_key_equal(addr_t key1, addr_t key2) { - uint8_t * buf1 = (uint8_t *)key1; - uint8_t * buf2 = (uint8_t *)key2; - return (memcmp(buf1, buf2, HASH_KEY_LEN) == 0); +static inline int hash_key_equal(addr_t key1, addr_t key2) { + return (memcmp((uint8_t *)key1, (uint8_t *)key2, HASH_KEY_LEN) == 0); } -static int init_route_cache() { - g_route_cache = v3_create_htable(MIN_CACHE_SIZE, &hash_from_key_fn, &hash_key_equal); +static int init_route_cache(struct vnet_state_t *vnet_state) { + vnet_state->g_route_cache = v3_create_htable(MIN_CACHE_SIZE, + &hash_from_key_fn, + &hash_key_equal); - if (g_route_cache == NULL) { - PrintError("Vnet: Route Cache Initiate Failurely\n"); + if (vnet_state->g_route_cache == NULL) { + PrintError("Vnet: Route Cache Init Fails\n"); return -1; } @@ -136,10 +123,10 @@ static int init_route_cache() { } static void make_hash_key(route_hashkey_t hashkey, - char src_addr[6], - char dest_addr[6], - char src_type, - int src_index) { + char src_addr[6], + char dest_addr[6], + char src_type, + int src_index) { int j; for (j = 0; j < 6; j++) { @@ -152,8 +139,11 @@ static void make_hash_key(route_hashkey_t hashkey, *(int *)(hashkey + 12) = src_index; } -static int add_route_to_cache(route_hashkey_t hashkey, int num_matched_r, int * matches) { +static int add_route_to_cache(route_hashkey_t hashkey, + int num_matched_r, + int * matches) { struct route_cache_entry * new_entry = NULL; + struct vnet_state_t *vnet_state = &g_vnet_state; int i; new_entry = (struct route_cache_entry *)V3_Malloc(sizeof(struct route_cache_entry)); @@ -163,7 +153,6 @@ static int add_route_to_cache(route_hashkey_t hashkey, int num_matched_r, int * } new_entry->num_matched_routes = num_matched_r; - new_entry->matches = (int *)V3_Malloc(num_matched_r * sizeof(int)); if (new_entry->matches == NULL) { @@ -175,9 +164,8 @@ static int add_route_to_cache(route_hashkey_t hashkey, int num_matched_r, int * new_entry->matches[i] = matches[i]; } - // here, when v3_htable_insert return 0, it means insert fails - if (v3_htable_insert(g_route_cache, (addr_t)hashkey, (addr_t)new_entry) == 0) { - PrintError("Vnet: Insert new route entry to cache failed\n"); + if (v3_htable_insert(vnet_state->g_route_cache, (addr_t)hashkey, (addr_t)new_entry) == 0) { + PrintError("Vnet: Failed to insert new route entry to the cache\n"); V3_Free(new_entry->matches); V3_Free(new_entry); } @@ -186,11 +174,14 @@ static int add_route_to_cache(route_hashkey_t hashkey, int num_matched_r, int * } static int clear_hash_cache() { - v3_free_htable(g_route_cache, 1, 1); - - g_route_cache = v3_create_htable(MIN_CACHE_SIZE, hash_from_key_fn, hash_key_equal); + struct vnet_state_t *vnet_state = &g_vnet_state; + + v3_free_htable(vnet_state->g_route_cache, 1, 1); + vnet_state->g_route_cache = v3_create_htable(MIN_CACHE_SIZE, + hash_from_key_fn, + hash_key_equal); - if (g_route_cache == NULL) { + if (vnet_state->g_route_cache == NULL) { PrintError("Vnet: Route Cache Create Failurely\n"); return -1; } @@ -202,8 +193,10 @@ static int look_into_cache(route_hashkey_t hashkey, int * matches) { struct route_cache_entry * found = NULL; int n_matches = -1; int i = 0; + struct vnet_state_t *vnet_state = &g_vnet_state; - found = (struct route_cache_entry *)v3_htable_search(g_route_cache, (addr_t)hashkey); + found = (struct route_cache_entry *)v3_htable_search(vnet_state->g_route_cache, + (addr_t)hashkey); if (found != NULL) { n_matches = found->num_matched_routes; @@ -216,10 +209,11 @@ static int look_into_cache(route_hashkey_t hashkey, int * matches) { return n_matches; } + #ifdef CONFIG_DEBUG_VNET static void print_packet(uchar_t *pkt, int size) { - PrintDebug("Vnet: print_data_packet: size: %d\n", size); + PrintDebug("Vnet: data_packet: size: %d\n", size); v3_hexdump(pkt, size, NULL, 0); } @@ -252,64 +246,122 @@ static inline void mac_to_string(char mac[6], char * buf) { mac[3], mac[4], mac[5]); } -static void dump_routes(struct routing_entry **route_tables) { + +static void print_route(struct routing_entry *route){ char dest_str[18]; char src_str[18]; - struct routing_entry *route = NULL; + + mac_to_string(route->src_mac, src_str); + mac_to_string(route->dest_mac, dest_str); + + PrintDebug("SRC(%s), DEST(%s), src_mac_qual(%d), dst_mac_qual(%d)\n", + src_str, + dest_str, + route->src_mac_qual, + route->dest_mac_qual); + PrintDebug("Src_Link(%d), src_type(%d), dst_link(%d), dst_type(%d)\n\n", + route->src_link_idx, + route->src_type, + route->link_idx, + route->link_type); +} + + +static void dump_routes(struct routing_entry **route_tables) { int i; PrintDebug("\nVnet: route table dump start =====\n"); for(i = 0; i < MAX_ROUTES; i++) { if (route_tables[i] != NULL){ - route = route_tables[i]; - - mac_to_string(route->src_mac, src_str); - mac_to_string(route->dest_mac, dest_str); - - PrintDebug("route: %d\n", i); - PrintDebug("SRC(%s), DEST(%s), src_mac_qual(%d), dst_mac_qual(%d)\n", src_str, dest_str, route->src_mac_qual, route->dest_mac_qual); - PrintDebug("Src_Link(%d), src_type(%d), dst_link(%d), dst_type(%d)\n\n", route->src_link_idx, route->src_type, route->link_idx, route->link_type); + print_route(route_tables[i]); } } PrintDebug("\nVnet: route table dump end =====\n"); } +static void dump_dom0_routes(struct routing_entry routes[], int size) { + int i; + + PrintDebug("\nVnet: route table from dom0 guest =====\n"); + + for(i = 0; i dest_ip, + link->dest_port, + link->pro_type); + PrintDebug("vnet_header:\n"); + v3_hexdump(&link->vnet_header, sizeof(struct udp_link_header), NULL, 0); + } + + PrintDebug("\nVnet: link table dom0 guest end =====\n"); +} + #endif static int __add_link_entry(struct link_entry * link) { int idx; - + struct vnet_state_t *vnet_state = &g_vnet_state; + + v3_lock(vnet_state->g_links.lock); for (idx = 0; idx < MAX_LINKS; idx++) { - if (g_links.links[idx] == NULL) { - g_links.links[idx] = link; - g_links.size++; - - return idx; + if (vnet_state->g_links.links[idx] == NULL) { + vnet_state->g_links.links[idx] = link; + vnet_state->g_links.size++; + break; } } + v3_unlock(vnet_state->g_links.lock); - PrintError("No available Link entry\n"); - return -1; + if (idx == MAX_LINKS) { + PrintDebug("VNET: No available Link entry for new link\n"); + return -1; + } + + return idx; } static int __add_route_entry(struct routing_entry * route) { int idx; - + struct vnet_state_t *vnet_state = &g_vnet_state; + + v3_lock(vnet_state->g_routes.lock); for (idx = 0; idx < MAX_ROUTES; idx++) { - if (g_routes.routes[idx] == NULL) { - g_routes.routes[idx] = route; - g_routes.size++; - - return idx; + if (vnet_state->g_routes.routes[idx] == NULL) { + vnet_state->g_routes.routes[idx] = route; + vnet_state->g_routes.size++; + break; } } + v3_unlock(vnet_state->g_routes.lock); - PrintError("No available route entry\n"); - return -1; -} + if(idx == MAX_LINKS){ + PrintDebug("VNET: No available route entry for new route\n"); + return -1; + } +#ifdef CONFIG_DEBUG_VNET + dump_routes(vnet_state->g_routes.routes); +#endif + + return idx; +} int vnet_add_route_entry(char src_mac[6], char dest_mac[6], @@ -322,8 +374,9 @@ int vnet_add_route_entry(char src_mac[6], struct routing_entry * new_route = (struct routing_entry *)V3_Malloc(sizeof(struct routing_entry)); int idx = -1; + PrintDebug("Vnet: vnet_add_route_entry\n"); + memset(new_route, 0, sizeof(struct routing_entry)); - if ((src_mac_qual != MAC_ANY)) { memcpy(new_route->src_mac, src_mac, 6); } @@ -345,143 +398,33 @@ int vnet_add_route_entry(char src_mac[6], } clear_hash_cache(); - - return idx; -} - -#if 0 -static void * __delete_link_entry(int index) { - struct link_entry * link = NULL; - void * ret = NULL; - link_type_t type; - - if ((index >= MAX_LINKS) || (g_links.links[index] == NULL)) { - return NULL; - } - - link = g_links.links[index]; - type = g_links.links[index]->type; - - if (type == LINK_INTERFACE) { - ret = (void *)g_links.links[index]->dst_dev; - } else if (type == LINK_EDGE) { - ret = (void *)g_links.links[index]->dst_link; - } - - g_links.links[index] = NULL; - g_links.size--; - - V3_Free(link); - - return ret; -} - -static int find_route_entry(char src_mac[6], - char dest_mac[6], - int src_mac_qual, - int dest_mac_qual, - int link_idx, - link_type_t link_type, - int src, - link_type_t src_type) { - int i; - char temp_src_mac[6]; - char temp_dest_mac[6]; - - if ((src_mac_qual != MAC_ANY) && (src_mac_qual != MAC_NONE)) { - memcpy(temp_src_mac, src_mac, 6); - } else { - memset(temp_src_mac, 0, 6); - } - - if ((dest_mac_qual != MAC_ANY) && (dest_mac_qual != MAC_NONE)) { - memcpy(temp_dest_mac, dest_mac, 6); - } else { - memset(temp_dest_mac, 0, 6); - } - - for (i = 0; i < MAX_ROUTES; i++) { - if (g_routes.routes[i] != NULL) { - if ((memcmp(temp_src_mac, g_routes.routes[i]->src_mac, 6) == 0) && - (memcmp(temp_dest_mac, g_routes.routes[i]->dest_mac, 6) == 0) && - (g_routes.routes[i]->src_mac_qual == src_mac_qual) && - (g_routes.routes[i]->dest_mac_qual == dest_mac_qual) && - ( (link_type == LINK_ANY) || - ((link_type == g_routes.routes[i]->link_type) && (g_routes.routes[i]->link_idx == link_idx))) && - ( (src_type == LINK_ANY) || - ((src_type == g_routes.routes[i]->src_type) && (g_routes.routes[i]->src_link_idx == src)))) { - return i; - } - } - } - - return -1; -} - -static int __delete_route_entry(int index) { - struct routing_entry * route; - - if ((index >= MAX_ROUTES) || (g_routes.routes[index] == NULL)) { - PrintDebug("VNET: wrong index in delete route entry %d\n", index); - return -1; - } - route = g_routes.routes[index]; - g_routes.routes[index] = NULL; - g_routes.size--; - - V3_Free(route); - - clear_hash_cache(); - - return 0; + return idx; } -static int vnet_delete_route_entry_by_addr(char src_mac[6], - char dest_mac[6], - int src_mac_qual, - int dest_mac_qual, - int link_idx, - link_type_t type, - int src, - link_type_t src_type) { - int index = find_route_entry(src_mac, dest_mac, src_mac_qual, - dest_mac_qual, link_idx, type, src, src_type); - - if (index == -1) { - PrintDebug("VNET: wrong in delete route entry %d\n", index); - return -1; - } - - return __delete_route_entry(index); -} -#endif - static int match_route(uint8_t * src_mac, uint8_t * dst_mac, link_type_t src_type, int src_index, int * matches) { + struct routing_entry *route = NULL; + struct vnet_state_t *vnet_state = &g_vnet_state; int matched_routes[MAX_ROUTES]; int num_matches = 0; int i; - struct routing_entry *route = NULL; #ifdef CONFIG_DEBUG_VNET char dest_str[18]; char src_str[18]; - + mac_to_string(src_mac, src_str); mac_to_string(dst_mac, dest_str); - PrintDebug("Vnet: match_route. pkt: SRC(%s), DEST(%s)\n", src_str, dest_str); - - dump_routes(g_routes.routes); #endif for(i = 0; i < MAX_ROUTES; i++) { - if (g_routes.routes[i] != NULL){ - route = g_routes.routes[i]; + if (vnet_state->g_routes.routes[i] != NULL){ + route = vnet_state->g_routes.routes[i]; if(src_type == LINK_ANY && src_index == -1) { if ((route->dest_mac_qual == MAC_ANY) && @@ -554,17 +497,15 @@ static int match_route(uint8_t * src_mac, static int handle_one_pkt(struct ethernet_pkt *pkt) { int src_link_index = -1; //the value of src_link_index of udp always is 0 - int i; char src_mac[6]; char dst_mac[6]; - int matches[MAX_ROUTES]; int num_matched_routes = 0; - struct in_pkt_header header; + char hash_key[HASH_KEY_SIZE]; + struct vnet_state_t *vnet_state = &g_vnet_state; + int i; - char hash_key[hash_key_size]; - // get the ethernet and ip headers from the packet memcpy((void *)&header, (void *)pkt->data, sizeof(header)); memcpy(src_mac, header.ethernetsrc, 6); @@ -573,22 +514,17 @@ static int handle_one_pkt(struct ethernet_pkt *pkt) { #ifdef CONFIG_DEBUG_VNET char dest_str[18]; char src_str[18]; - + mac_to_string(src_mac, src_str); mac_to_string(dst_mac, dest_str); - PrintDebug("Vnet: HandleDataOverLink. SRC(%s), DEST(%s)\n", src_str, dest_str); #endif - // link_edge -> pt->type??? make_hash_key(hash_key, src_mac, dst_mac, LINK_EDGE, src_link_index); - num_matched_routes = look_into_cache((route_hashkey_t)hash_key, matches); if (num_matched_routes == -1) { - // no match in the cache - num_matched_routes = match_route(src_mac, dst_mac, LINK_ANY, src_link_index, matches); - + num_matched_routes = match_route(src_mac, dst_mac, LINK_ANY, src_link_index, matches); if (num_matched_routes > 0) { add_route_to_cache(hash_key, num_matched_routes,matches); } @@ -607,14 +543,14 @@ static int handle_one_pkt(struct ethernet_pkt *pkt) { struct link_entry * link = NULL; route_index = matches[i]; - link_index = g_routes.routes[route_index]->link_idx; + link_index = vnet_state->g_routes.routes[route_index]->link_idx; if ((link_index < 0) || (link_index > MAX_LINKS) || - (g_links.links[link_index] == NULL)) { + (vnet_state->g_links.links[link_index] == NULL)) { continue; } - link = g_links.links[link_index]; + link = vnet_state->g_links.links[link_index]; pkt_len = pkt->size; if (link->type == LINK_EDGE) { @@ -623,9 +559,9 @@ static int handle_one_pkt(struct ethernet_pkt *pkt) { struct udp_link_header *hdr = &(link->dst_link->vnet_header); struct ip_header *ip = &hdr->ip_hdr; struct udp_header *udp = &hdr->udp_hdr; - udp->len = pkt_len + sizeof(struct udp_header); - ip->total_len = pkt_len + sizeof(struct udp_header) + sizeof(struct ip_header); - ip->cksum = ip_xsum(ip, sizeof(struct ip_header)); + udp->len = pkt_len + sizeof(struct udp_header); + ip->total_len = pkt_len + sizeof(struct udp_header) + sizeof(struct ip_header); + ip->cksum = ip_xsum(ip, sizeof(struct ip_header)); int hdr_size = sizeof(struct udp_link_header); memcpy(&pkt->ext_hdr, hdr, hdr_size); @@ -655,52 +591,54 @@ static int handle_one_pkt(struct ethernet_pkt *pkt) { return 0; } -static int send_ethernet_pkt(uchar_t *data, int len) { +static int send_ethernet_pkt(uchar_t *data, int len, void *private_data) { struct ethernet_pkt *pkt; + struct vnet_state_t *vnet_state = &g_vnet_state; pkt = (struct ethernet_pkt *)V3_Malloc(sizeof(struct ethernet_pkt)); - memset(pkt, 0, sizeof(struct ethernet_pkt)); - if(pkt == NULL){ PrintError("VNET: Memory allocate fails\n"); return -1; } - - ethernet_packet_init(pkt, data, len); //====here we copy sending data once - - PrintDebug("VNET: vm_send_pkt: transmitting packet: (size:%d)\n", (int)pkt->size); + memset(pkt, 0, sizeof(struct ethernet_pkt)); + ethernet_packet_init(pkt, data, len); + v3_enqueue(vnet_state->g_inpkt_q, (addr_t)pkt); + #ifdef CONFIG_DEBUG_VNET + PrintDebug("VNET: send_pkt: transmitting packet: (size:%d)\n", (int)pkt->size); print_packet((char *)data, len); #endif - - v3_enqueue(g_inpkt_q, (addr_t)pkt); return 0; } -//send raw ethernet packet -int v3_vnet_send_rawpkt(uchar_t * buf, int len, void *private_data) { - PrintDebug("VNET: In V3_Send_pkt: pkt length %d\n", len); +int v3_vnet_send_rawpkt(uchar_t * buf, + int len, + void *private_data) { + PrintDebug("VNET: In v3_vnet_send_rawpkt: pkt length %d\n", len); - return send_ethernet_pkt(buf, len); + return send_ethernet_pkt(buf, len, private_data); } //sending the packet from Dom0, should remove the link header -int v3_vnet_send_udppkt(uchar_t * buf, int len, void *private_data) { - PrintDebug("VNET: In V3_Send_pkt: pkt length %d\n", len); - +int v3_vnet_send_udppkt(uchar_t * buf, + int len, + void *private_data) { uint_t hdr_len = sizeof(struct udp_link_header); + + PrintDebug("VNET: In v3_vnet_send_udppkt: pkt length %d\n", len); - return send_ethernet_pkt((uchar_t *)(buf+hdr_len), len - hdr_len); + return send_ethernet_pkt((uchar_t *)(buf+hdr_len), len - hdr_len, private_data); } static int search_device(char * device_name) { + struct vnet_state_t *vnet_state = &g_vnet_state; int i; for (i = 0; i < MAX_LINKS; i++) { - if ((g_links.links[i] != NULL) && (g_links.links[i]->type == LINK_INTERFACE)) { - if (strcmp(device_name, g_links.links[i]->dst_dev->name) == 0) { + if ((vnet_state->g_links.links[i] != NULL) && (vnet_state->g_links.links[i]->type == LINK_INTERFACE)) { + if (strcmp(device_name, vnet_state->g_links.links[i]->dst_dev->name) == 0) { return i; } } @@ -743,38 +681,13 @@ int vnet_register_device(struct vm_device * vdev, return idx; } -#if 0 -static int vnet_unregister_device(char * dev_name) { - int idx; - - idx = search_device(dev_name); - - if (idx == -1) { - PrintDebug("VNET: No device with name %s found\n", dev_name); - return -1; - } - - struct vnet_if_device * device = (struct vnet_if_device *)__delete_link_entry(idx); - if (device == NULL) { - PrintError("VNET: Device %s not in the link table %d, something may be wrong in link table\n", dev_name, idx); - return -1; - } - - V3_Free(device); - - return idx; -} - -#endif - int v3_vnet_pkt_process() { struct ethernet_pkt * pkt; + struct vnet_state_t *vnet_state = &g_vnet_state; - while ((pkt = (struct ethernet_pkt *)v3_dequeue(g_inpkt_q)) != NULL) { - PrintDebug("VNET: In vnet_check: pt length %d, pt type %d\n", (int)pkt->size, (int)pkt->type); - + while ((pkt = (struct ethernet_pkt *)v3_dequeue(vnet_state->g_inpkt_q))!= NULL) { if (handle_one_pkt(pkt) != -1) { - PrintDebug("VNET: vnet_check: handle one packet!\n"); + PrintDebug("VNET: vnet_check: handle one packet! pt length %d, pt type %d\n", (int)pkt->size, (int)pkt->type); } else { PrintDebug("VNET: vnet_check: Fail to forward one packet, discard it!\n"); } @@ -785,37 +698,36 @@ int v3_vnet_pkt_process() { return 0; } -static void init_empty_link_table() { +static void vnet_state_init(struct vnet_state_t *vnet_state) { int i; + /*initial links table */ for (i = 0; i < MAX_LINKS; i++) { - g_links.links[i] = NULL; + vnet_state->g_links.links[i] = NULL; } + vnet_state->g_links.size = 0; + if(v3_lock_init(&(vnet_state->g_links.lock)) == -1){ + PrintError("VNET: Failure to init lock for links table\n"); + } + PrintDebug("VNET: Links table initiated\n"); - g_links.size = 0; -} - -static void init_empty_route_table() { - int i; - + /*initial routes table */ for (i = 0; i < MAX_ROUTES; i++) { - g_routes.routes[i] = NULL; + vnet_state->g_routes.routes[i] = NULL; } + vnet_state->g_routes.size = 0; + if(v3_lock_init(&(vnet_state->g_routes.lock)) == -1){ + PrintError("VNET: Failure to init lock for routes table\n"); + } + PrintDebug("VNET: Routes table initiated\n"); - g_links.size = 0; -} - -static void init_tables() { - init_empty_link_table(); - init_empty_route_table(); - init_route_cache(); -} - -static void init_pkt_queue() { - PrintDebug("VNET Init package receiving queue\n"); - - g_inpkt_q = v3_create_queue(); - v3_init_queue(g_inpkt_q); + /*initial pkt receiving queue */ + vnet_state->g_inpkt_q = v3_create_queue(); + v3_init_queue(vnet_state->g_inpkt_q); + PrintDebug("VNET: Receiving queue initiated\n"); + + /*initial routing cache */ + init_route_cache(vnet_state); } static void free_link_mem(struct link_entry *link){ @@ -823,7 +735,8 @@ static void free_link_mem(struct link_entry *link){ V3_Free(link); } -// TODO: + +// TODO: static int addto_routing_link_tables(struct routing_entry *route_tab, uint16_t num_routes, struct link_entry *link_tab, @@ -907,15 +820,13 @@ static int addto_routing_link_tables(struct routing_entry *route_tab, } new_route = (struct routing_entry *)V3_Malloc(sizeof(struct routing_entry)); - if (new_route == NULL){ PrintError("VNET: Memory allocate fails\n"); return -1; } memcpy(new_route, route, sizeof(struct routing_entry)); - new_route->link_idx = link_idxs[new_route->link_idx]; - + new_route->link_idx = link_idxs[new_route->link_idx]; if (route->src_link_idx != -1) new_route->src_link_idx = link_idxs[new_route->src_link_idx]; @@ -929,53 +840,62 @@ static int addto_routing_link_tables(struct routing_entry *route_tab, return 0; } -struct table_init_info { - addr_t routing_table_start; - uint16_t routing_table_size; - addr_t link_table_start; - uint16_t link_table_size; -}; - //add the guest specified routes and links to the tables static int handle_init_tables_hcall(struct guest_info * info, uint_t hcall_id, void * priv_data) { - addr_t guest_addr = (addr_t)info->vm_regs.rcx; - addr_t info_addr, route_addr, link_addr; - struct table_init_info *init_info; - struct link_entry *link_array; - struct routing_entry *route_array; + struct vnet_if_link *link_array; + struct routing_entry *route_array; + addr_t routes_addr, routes_addr_gva; + addr_t links_addr, links_addr_gva; + uint16_t route_size, link_size; - PrintDebug("Vnet: In handle_init_tables_hcall\n"); + routes_addr_gva = (addr_t)info->vm_regs.rbx; + route_size = (uint16_t)info->vm_regs.rcx; + links_addr_gva = (addr_t)info->vm_regs.rdx; + link_size = (uint16_t)info->vm_regs.rsi; - if (guest_va_to_host_va(info, guest_addr, &info_addr) == -1) { + PrintDebug("Vnet: In handle_init_tables_hcall\n"); + PrintDebug("Vnet: route_table_start: %x, route size: %d\n", (int)routes_addr_gva, route_size); + PrintDebug("Vnet: link_table_start: %x, link size: %d\n", (int)links_addr_gva, link_size); + + if (guest_va_to_host_va(info, routes_addr_gva, &routes_addr) == -1) { PrintError("VNET: Could not translate guest address\n"); return -1; } - init_info = (struct table_init_info *)info_addr; + route_array = (struct routing_entry *)routes_addr; - if (guest_va_to_host_va(info, init_info->routing_table_start, &route_addr) == -1) { - PrintError("VNET: Could not translate guest address\n"); - return -1; - } - route_array = (struct routing_entry *)route_addr; +#ifdef CONFIG_DEBUG_VNET + dump_dom0_routes(route_array, route_size); +#endif - if (guest_va_to_host_va(info, init_info->link_table_start, &link_addr) == -1) { + if (guest_va_to_host_va(info, links_addr_gva, &links_addr) == -1) { PrintError("VNET: Could not translate guest address\n"); return -1; } - link_array = (struct link_entry *)link_addr; + link_array = (struct vnet_if_link *)links_addr; + +#ifdef CONFIG_DEBUG_VNET + dump_dom0_links(link_array, link_size); +#endif + + +// TODO: + if (0)addto_routing_link_tables(route_array, route_size, NULL, 0); - addto_routing_link_tables(route_array, init_info->routing_table_size, link_array, init_info->link_table_size); - return 0; } -void v3_vnet_init(struct guest_info * vm) { - init_tables(); - init_pkt_queue(); +void v3_init_vnet() { + vnet_state_init(&g_vnet_state); - v3_register_hypercall(vm, VNET_INITAB_HCALL, handle_init_tables_hcall, NULL); - PrintDebug("VNET Initialized\n"); } +//only need to called in config dom0 guest +int v3_register_guest_vnet(struct guest_info *vm){ + return v3_register_hypercall(vm, + VNET_INITAB_HCALL, + handle_init_tables_hcall, + NULL); +} + diff --git a/palacios/src/palacios/vmx.c b/palacios/src/palacios/vmx.c index aba5656..efc8e24 100644 --- a/palacios/src/palacios/vmx.c +++ b/palacios/src/palacios/vmx.c @@ -494,7 +494,7 @@ static int update_irq_entry_state(struct guest_info * info) { int_info.valid = 1; #ifdef CONFIG_DEBUG_INTERRUPTS - PrintDebug("Injecting exception %d (EIP=%p)\n", int_info.vector, (void *)info->rip); + PrintDebug("Injecting exception %d (EIP=%p)\n", int_info.vector, (void *)(addr_t)info->rip); #endif check_vmcs_write(VMCS_ENTRY_INT_INFO, int_info.value); @@ -536,7 +536,7 @@ static int update_irq_entry_state(struct guest_info * info) { PrintDebug("Injecting Interrupt %d at exit %u(EIP=%p)\n", info->intr_state.irq_vector, (uint32_t)info->num_exits, - (void *)info->rip); + (void *)(addr_t)info->rip); #endif check_vmcs_write(VMCS_ENTRY_INT_INFO, ent_int.value); @@ -794,7 +794,7 @@ void v3_init_vmx_cpu(int cpu_id) { uint64_t ret = 0; v3_get_msr(VMX_CR4_FIXED0_MSR,&(tmp_msr.hi),&(tmp_msr.lo)); - +#ifdef __V3_64BIT__ __asm__ __volatile__ ( "movq %%cr4, %%rbx;" "orq $0x00002000, %%rbx;" @@ -823,6 +823,38 @@ void v3_init_vmx_cpu(int cpu_id) { : : "%rbx" ); +#elif __V3_32BIT__ + __asm__ __volatile__ ( + "movl %%cr4, %%ecx;" + "orl $0x00002000, %%ecx;" + "movl %%ecx, %0;" + : "=m"(ret) + : + : "%ecx" + ); + + if ((~ret & tmp_msr.value) == 0) { + __asm__ __volatile__ ( + "movl %0, %%cr4;" + : + : "q"(ret) + ); + } else { + PrintError("Invalid CR4 Settings!\n"); + return; + } + + __asm__ __volatile__ ( + "movl %%cr0, %%ecx; " + "orl $0x00000020,%%ecx; " + "movl %%ecx, %%cr0;" + : + : + : "%ecx" + ); + +#endif + // // Should check and return Error here.... diff --git a/palacios/src/palacios/vmx_assist.c b/palacios/src/palacios/vmx_assist.c index f926c42..892a69e 100644 --- a/palacios/src/palacios/vmx_assist.c +++ b/palacios/src/palacios/vmx_assist.c @@ -105,7 +105,7 @@ static void load_segment(struct vmx_assist_segment * vmx_assist_seg, struct v3_s static void vmx_save_world_ctx(struct guest_info * info, struct vmx_assist_context * ctx) { struct vmx_data * vmx_info = (struct vmx_data *)(info->vmm_data); - PrintDebug("Writing from RIP: 0x%p\n", (void *)info->rip); + PrintDebug("Writing from RIP: 0x%p\n", (void *)(addr_t)info->rip); ctx->eip = info->rip; ctx->esp = info->vm_regs.rsp; diff --git a/palacios/src/palacios/vmx_ctrl_regs.c b/palacios/src/palacios/vmx_ctrl_regs.c index c7bf6e3..eb12760 100644 --- a/palacios/src/palacios/vmx_ctrl_regs.c +++ b/palacios/src/palacios/vmx_ctrl_regs.c @@ -157,10 +157,10 @@ static int handle_mov_to_cr0(struct guest_info * info, v3_reg_t * new_cr0, struc } if (vmx_info->state == VMXASSIST_ENABLED) { - PrintDebug("Loading VMXASSIST at RIP: %p\n", (void *)info->rip); + PrintDebug("Loading VMXASSIST at RIP: %p\n", (void *)(addr_t)info->rip); } else { PrintDebug("Leaving VMXASSIST and entering protected mode at RIP: %p\n", - (void *)info->rip); + (void *)(addr_t)info->rip); } // PE switches modify the RIP directly, so we clear the instr_len field to avoid catastrophe diff --git a/palacios/src/palacios/vmx_handler.c b/palacios/src/palacios/vmx_handler.c index 67f03a4..7b738ca 100644 --- a/palacios/src/palacios/vmx_handler.c +++ b/palacios/src/palacios/vmx_handler.c @@ -32,16 +32,17 @@ #include #include -#ifdef CONFIG_TELEMETRY -#include -#endif - - #ifdef CONFIG_VNET #include #endif +#ifdef CONFIG_LINUX_VIRTIO_NET + extern int v3_virtionic_pktprocess(struct guest_info * info); +#endif +#ifdef CONFIG_TELEMETRY +#include +#endif /* At this point the GPRs are already copied into the guest_info state */ int v3_handle_vmx_exit(struct guest_info * info, struct vmx_exit_info * exit_info) { @@ -61,12 +62,20 @@ int v3_handle_vmx_exit(struct guest_info * info, struct vmx_exit_info * exit_inf } #endif +#ifdef VNET_PROFILE + uint64_t exit_start_time, vnet_start_time; + uint64_t exit_end_time, vnet_end_time; + rdtscll(exit_start_time); + num_exit ++; + if (last_exit_time > 0) + guest_time += exit_start_time - last_exit_time; +#endif + switch (exit_info->exit_reason) { case VMEXIT_INFO_EXCEPTION_OR_NMI: { pf_error_t error_code = *(pf_error_t *)&(exit_info->int_err); - - // JRL: Change "0x0e" to a macro value + // JRL: Change "0x0e" to a macro value if ((uint8_t)exit_info->int_info == 0x0e) { #ifdef CONFIG_DEBUG_SHADOW_PAGING PrintDebug("Page Fault at %p error_code=%x\n", (void *)exit_info->exit_qual, *(uint32_t *)&error_code); @@ -226,16 +235,22 @@ int v3_handle_vmx_exit(struct guest_info * info, struct vmx_exit_info * exit_inf return -1; } + #ifdef CONFIG_VNET v3_vnet_pkt_process(info); #endif +#ifdef CONFIG_LINUX_VIRTIO_NET + v3_virtionic_pktprocess(info); +#endif + #ifdef CONFIG_TELEMETRY if (info->enable_telemetry) { v3_telemetry_end_exit(info, exit_info->exit_reason); } #endif + return 0; } diff --git a/palacios/src/palacios/vmx_lowlevel.S b/palacios/src/palacios/vmx_lowlevel.S index aaae943..757d835 100644 --- a/palacios/src/palacios/vmx_lowlevel.S +++ b/palacios/src/palacios/vmx_lowlevel.S @@ -201,7 +201,7 @@ vmx_launch_ret: #else -#define save_resgisters(location) \ +#define save_registers(location) \ pushl %eax; \ movl location, %eax; \ movl %edi, (%eax); \ @@ -229,6 +229,25 @@ vmx_launch_ret: movl 48(%eax), %ecx; \ popl %eax; + +#define save_ctrl_regs(location) \ + push %eax; \ + push %ebx; \ + movl location, %eax; \ + movl %cr2, %ebx; \ + movl %ebx, 8(%eax); \ + popl %ebx; \ + popl %eax + +#define restore_ctrl_regs(location) \ + push %eax; \ + push %ebx; \ + movl location, %eax; \ + movl 8(%eax), %ebx; \ + movl %ebx, %cr2; \ + popl %ebx; \ + popl %eax + #define PUSHA \ push %eax; \ push %ebx; \ @@ -247,82 +266,86 @@ vmx_launch_ret: pop %ebx; \ pop %eax; -.align 8 -.globl v3_vmx_exit_handler -v3_vmx_exit_handler: - save_registers(4(%esp)) - addl $8, %esp - POPA - popf - pushl %edi - call v3_handle_vmx_exit - andl %eax, %eax - jnz .Lvmm_failure +#define PRE_LAUNCH(return_target) \ + pushf; \ + PUSHA; \ + pushl %edi; \ + pushl %edx; \ + \ + movl %esp, %eax; \ + movl $VMCS_HOST_RSP, %ebx; \ + vmwrite %eax, %ebx; \ + jz .Lfail_valid; \ + jc .Lfail_invalid; \ + \ + movl return_target, %eax; \ + movl $VMCS_HOST_RIP, %ebx; \ + vmwrite %eax, %ebx; \ + jz .Lfail_valid; \ + jc .Lfail_invalid; \ + \ + restore_ctrl_regs(%edx); \ + restore_registers(%edi); -v3_vmx_vmresume: - popl %edi - pushf - PUSHA - pushl %edi - restore_registers(%rdi) + +.align 4 +.globl v3_vmx_resume +v3_vmx_resume: + + PRE_LAUNCH($vmx_resume_ret); vmresume - addl $8, %esp +vmx_resume_ret: jz .Lfail_valid jc .Lfail_invalid - jmp .Lreturn + jmp .Lnormal_exit -.align 8 -.globl v3_vmx_vmlaunch +.align 4 +.globl v3_vmx_launch // vm_regs = %edi -v3_vmx_vmlaunch: - cli - pushf - PUSHA - pushl %edi - - movl %esp, %eax - movl $VMCS_HOST_RSP, %ebx - vmwrite %eax, %ebx - jz .Lfail_valid - jc .Lfail_invalid - - movl $v3_vmx_exit_handler, %eax - movl $VMCS_HOST_RIP, %ebx - vmwrite %eax, %ebx - jz .Lfail_valid - jc .Lfail_invalid +v3_vmx_launch: - restore_registers(%edi) + PRE_LAUNCH($vmx_launch_ret); vmlaunch + +vmx_launch_ret: jz .Lfail_valid jc .Lfail_invalid - jmp .Lreturn + jmp .Lnormal_exit .Lfail_valid: addl $8, %esp POPA + popf movl $VMX_FAIL_VALID, %eax jmp .Lreturn .Lfail_invalid: - addq $8, %esp + addl $8, %esp POPA - movl $MVX_FAIL_INVALID, %eax + popf + movl $VMX_FAIL_INVALID, %eax jmp .Lreturn .Lvmm_failure: - addq $8, %esp - POPA + addl $12, %esp movl $VMM_FAILURE, %eax jmp .Lreturn -.Lreturn: - sti + +.Lnormal_exit: + save_registers(8(%esp)); + save_ctrl_regs(8(%esp)); + addl $8, %esp + POPA popf + xorl %eax, %eax + jmp .Lreturn + +.Lreturn: ret #endif diff --git a/utils/guest_creator/default.xml b/utils/guest_creator/default.xml index d8576e7..ee1b098 100644 --- a/utils/guest_creator/default.xml +++ b/utils/guest_creator/default.xml @@ -27,7 +27,7 @@ - + @@ -108,12 +108,12 @@ 0 - +