[palacios.git] / palacios / src / palacios / vmm_mem.c

/* 
 * 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) 2008, Jack Lange <jarusl@cs.northwestern.edu> 
 * Copyright (c) 2008, The V3VEE Project <http://www.v3vee.org> 
 * All rights reserved.
 *
 * Author: Jack Lange <jarusl@cs.northwestern.edu>
 *
 * This is free software.  You are permitted to use,
 * redistribute, and modify it as specified in the file "V3VEE_LICENSE".
 */

#include <palacios/vmm_mem.h>
#include <palacios/vmm.h>
#include <palacios/vmm_util.h>
#include <palacios/vmm_emulator.h>
#include <palacios/vm_guest.h>
#include <palacios/vmm_debug.h>

#include <palacios/vmm_shadow_paging.h>
#include <palacios/vmm_direct_paging.h>

#include <interfaces/vmm_numa.h>

uint64_t v3_mem_block_size = V3_CONFIG_MEM_BLOCK_SIZE;


struct v3_mem_region * v3_get_base_region(struct v3_vm_info * vm, addr_t gpa) {
    struct v3_mem_map * map = &(vm->mem_map);
    uint32_t block_index = gpa / v3_mem_block_size;

    if ((gpa >= (map->num_base_regions * v3_mem_block_size)) ||
        (block_index >= map->num_base_regions)) {
        PrintError(vm, VCORE_NONE, "Guest Address Exceeds Base Memory Size (ga=0x%p), (limit=0x%p)\n", 
                   (void *)gpa, (void *)vm->mem_size);
        v3_print_mem_map(vm);

        return NULL;
    }


    return &(map->base_regions[block_index]);
}


static int mem_offset_hypercall(struct guest_info * info, uint_t hcall_id, void * private_data) {
    /*
    PrintDebug(info->vm_info, info,"V3Vee: Memory offset hypercall (offset=%p)\n", 
               (void *)(info->vm_info->mem_map.base_region.host_addr));

    info->vm_regs.rbx = info->vm_info->mem_map.base_region.host_addr;
    */
    return -1;
}

static int unhandled_err(struct guest_info * core, addr_t guest_va, addr_t guest_pa, 
                         struct v3_mem_region * reg, pf_error_t access_info) {

    PrintError(core->vm_info, core, "Unhandled memory access error (gpa=%p, gva=%p, error_code=%d)\n",
               (void *)guest_pa, (void *)guest_va, *(uint32_t *)&access_info);

    v3_print_mem_map(core->vm_info);

    v3_print_guest_state(core);

    return -1;
}

static int gpa_to_node_from_cfg(struct v3_vm_info * vm, addr_t gpa) {
    v3_cfg_tree_t * layout_cfg = v3_cfg_subtree(vm->cfg_data->cfg, "mem_layout");
    v3_cfg_tree_t * region_desc = v3_cfg_subtree(layout_cfg, "region");

    while (region_desc) {
        char * start_addr_str = v3_cfg_val(region_desc, "start_addr");
        char * end_addr_str = v3_cfg_val(region_desc, "end_addr");
        char * node_id_str = v3_cfg_val(region_desc, "node");

        addr_t start_addr = 0;
        addr_t end_addr = 0;
        int node_id = 0;
        
        if ((!start_addr_str) || (!end_addr_str) || (!node_id_str)) {
            PrintError(vm, VCORE_NONE, "Invalid memory layout in configuration\n");
            return -1;
        }
        
        start_addr = atox(start_addr_str);
        end_addr = atox(end_addr_str);
        node_id = atoi(node_id_str);

        if ((gpa >= start_addr) && (gpa < end_addr)) {
            return node_id;
        }

        region_desc = v3_cfg_next_branch(region_desc);
    }

    return -1;
}

//
// This code parallels that in vmm_shadow_paging.c:v3_init_shdw_impl() 
// and vmm_config.c:determine_paging_mode.   The determination of which
// paging mode will be used is determined much later than the allocation of
// the guest memory regions, so we need to do this here to decide if they
// need to be below 4 GB or not.
static int will_use_shadow_paging(struct v3_vm_info *vm)
{
    v3_cfg_tree_t * pg_cfg = v3_cfg_subtree(vm->cfg_data->cfg, "paging");
    char * pg_mode = v3_cfg_val(pg_cfg, "mode");
   
    if (pg_mode == NULL) { 
        return 1; // did not ask, get shadow
    } else {
        if (strcasecmp(pg_mode, "nested") == 0) {
            extern v3_cpu_arch_t v3_mach_type;
            if ((v3_mach_type == V3_SVM_REV3_CPU) || 
                (v3_mach_type == V3_VMX_EPT_CPU) ||
                (v3_mach_type == V3_VMX_EPT_UG_CPU)) {
                return 0; // ask for nested, get nested
            } else { 
                return 1; // ask for nested, get shadow
            }
        } else if (strcasecmp(pg_mode, "shadow") != 0) { 
            return 1;     // ask for shadow, get shadow
        } else {
            return 1;     // ask for something else, get shadow
        }
    }
}


int v3_init_mem_map(struct v3_vm_info * vm) {
    struct v3_mem_map * map = &(vm->mem_map);
    addr_t block_pages = v3_mem_block_size >> 12;
    int i = 0;

    map->num_base_regions = (vm->mem_size / v3_mem_block_size) + \
        ((vm->mem_size % v3_mem_block_size) > 0);


    map->mem_regions.rb_node = NULL;

    map->base_regions = V3_Malloc(sizeof(struct v3_mem_region) * map->num_base_regions);

    if (map->base_regions == NULL) {
        PrintError(vm, VCORE_NONE, "Could not allocate base region array\n");
        return -1;
    }

    memset(map->base_regions, 0, sizeof(struct v3_mem_region) * map->num_base_regions);
  

    for (i = 0; i < map->num_base_regions; i++) {
        struct v3_mem_region * region = &(map->base_regions[i]);
        int node_id = -1;

        // 2MB page alignment needed for 2MB hardware nested paging
        region->guest_start = v3_mem_block_size * i;
        region->guest_end = region->guest_start + v3_mem_block_size;

        // We assume that the xml config was smart enough to align the layout to the block size
        // If they didn't we're going to ignore their settings 
        //     and use whatever node the first byte of the block is assigned to
        node_id = gpa_to_node_from_cfg(vm, region->guest_start);
        
        V3_Print(vm, VCORE_NONE, "Allocating block %d on node %d\n", i, node_id);

        region->host_addr = (addr_t)V3_AllocPagesExtended(block_pages,
                                                          PAGE_SIZE_4KB,
                                                          node_id,
                                                          will_use_shadow_paging(vm) ? 
                                                          V3_ALLOC_PAGES_CONSTRAINT_4GB : 0 ); 
                                                             
        if ((void *)region->host_addr == NULL) { 
            PrintError(vm, VCORE_NONE, "Could not allocate guest memory\n");
            return -1;
        }

        // Clear the memory...
        memset(V3_VAddr((void *)region->host_addr), 0, v3_mem_block_size);
        
        // Note assigned numa ID could be different than our request... 
        region->numa_id = v3_numa_hpa_to_node(region->host_addr);

        region->flags.read = 1;
        region->flags.write = 1;
        region->flags.exec = 1;
        region->flags.base = 1;
        region->flags.alloced = 1;
        region->flags.limit32 = will_use_shadow_paging(vm);
        

        region->unhandled = unhandled_err;
    }

    v3_register_hypercall(vm, MEM_OFFSET_HCALL, mem_offset_hypercall, NULL);

    return 0;
}


void v3_delete_mem_map(struct v3_vm_info * vm) {
    struct v3_mem_map * map = &(vm->mem_map);
    struct rb_node * node = v3_rb_first(&(map->mem_regions));
    struct v3_mem_region * reg;
    struct rb_node * tmp_node = NULL;
    addr_t block_pages = v3_mem_block_size >> 12;
    int i = 0;

    while (node) {
        reg = rb_entry(node, struct v3_mem_region, tree_node);
        tmp_node = node;
        node = v3_rb_next(node);

        v3_delete_mem_region(vm, reg);
    }

    for (i = 0; i < map->num_base_regions; i++) {
        struct v3_mem_region * region = &(map->base_regions[i]);
        V3_FreePages((void *)(region->host_addr), block_pages);
    }

    V3_Free(map->base_regions);

}


struct v3_mem_region * v3_create_mem_region(struct v3_vm_info * vm, uint16_t core_id, 
                                               addr_t guest_addr_start, addr_t guest_addr_end) {
    struct v3_mem_region * entry = NULL;

    if (guest_addr_start >= guest_addr_end) {
        PrintError(vm, VCORE_NONE, "Region start is after region end\n");
        return NULL;
    }

    entry = (struct v3_mem_region *)V3_Malloc(sizeof(struct v3_mem_region));

    if (!entry) {
        PrintError(vm, VCORE_NONE, "Cannot allocate in creating a memory region\n");
        return NULL;
    }

    memset(entry, 0, sizeof(struct v3_mem_region));

    entry->guest_start = guest_addr_start;
    entry->guest_end = guest_addr_end;
    entry->core_id = core_id;
    entry->unhandled = unhandled_err;

    return entry;
}




int v3_add_shadow_mem( struct v3_vm_info * vm, uint16_t core_id,
                       addr_t               guest_addr_start,
                       addr_t               guest_addr_end,
                       addr_t               host_addr)
{
    struct v3_mem_region * entry = NULL;

    entry = v3_create_mem_region(vm, core_id, 
                                 guest_addr_start, 
                                 guest_addr_end);

    entry->host_addr = host_addr;

    entry->flags.read = 1;
    entry->flags.write = 1;
    entry->flags.exec = 1;
    entry->flags.alloced = 1;

    if (v3_insert_mem_region(vm, entry) == -1) {
        V3_Free(entry);
        return -1;
    }

    return 0;
}



static inline 
struct v3_mem_region * __insert_mem_region(struct v3_vm_info * vm, 
                                           struct v3_mem_region * region) {
    struct rb_node ** p = &(vm->mem_map.mem_regions.rb_node);
    struct rb_node * parent = NULL;
    struct v3_mem_region * tmp_region;

    while (*p) {
        parent = *p;
        tmp_region = rb_entry(parent, struct v3_mem_region, tree_node);

        if (region->guest_end <= tmp_region->guest_start) {
            p = &(*p)->rb_left;
        } else if (region->guest_start >= tmp_region->guest_end) {
            p = &(*p)->rb_right;
        } else {
            if ((region->guest_end != tmp_region->guest_end) ||
                (region->guest_start != tmp_region->guest_start)) {
                PrintError(vm, VCORE_NONE, "Trying to map a partial overlapped core specific page...\n");
                return tmp_region; // This is ugly... 
            } else if (region->core_id == tmp_region->core_id) {
                PrintError(vm, VCORE_NONE, "Trying to map a core-overlapping page\n");
                return tmp_region;
            } else if (region->core_id < tmp_region->core_id) {
                p = &(*p)->rb_left;
            } else { 
                p = &(*p)->rb_right;
            }
        }
    }

    rb_link_node(&(region->tree_node), parent, p);
  
    return NULL;
}



int v3_insert_mem_region(struct v3_vm_info * vm, struct v3_mem_region * region) {
    struct v3_mem_region * ret;
    int i = 0;
    int rc;

    if ((ret = __insert_mem_region(vm, region))) {
        PrintError(vm, VCORE_NONE, "Internal insert failed returned region is from 0x%p to 0x%p on vcore %d\n", (void*)(ret->guest_start), (void*)(ret->guest_end), ret->core_id);
        return -1;
    }

    v3_rb_insert_color(&(region->tree_node), &(vm->mem_map.mem_regions));


    rc = 0;

    for (i = 0; i < vm->num_cores; i++) {
        struct guest_info * info = &(vm->cores[i]);

        // flush virtual page tables 
        // 3 cases shadow, shadow passthrough, and nested

        if (info->shdw_pg_mode == SHADOW_PAGING) {
            v3_mem_mode_t mem_mode = v3_get_vm_mem_mode(info);
            
            if (mem_mode == PHYSICAL_MEM) {
                rc |= v3_invalidate_passthrough_addr_range(info, region->guest_start, region->guest_end-1);
            } else {
                rc |= v3_invalidate_shadow_pts(info);
            }
            
        } else if (info->shdw_pg_mode == NESTED_PAGING) {
            rc |= v3_invalidate_nested_addr_range(info, region->guest_start, region->guest_end-1);
        }
    }

    return rc;
}
                                                 



struct v3_mem_region * v3_get_mem_region(struct v3_vm_info * vm, uint16_t core_id, addr_t guest_addr) {
    struct rb_node * n = vm->mem_map.mem_regions.rb_node;
    struct v3_mem_region * reg = NULL;

    while (n) {

        reg = rb_entry(n, struct v3_mem_region, tree_node);

        if (guest_addr < reg->guest_start) {
            n = n->rb_left;
        } else if (guest_addr >= reg->guest_end) {
            n = n->rb_right;
        } else {
            if (reg->core_id == V3_MEM_CORE_ANY) {
                // found relevant region, it's available on all cores
                return reg;
            } else if (core_id == reg->core_id) { 
                // found relevant region, it's available on the indicated core
                return reg;
            } else if (core_id < reg->core_id) { 
                // go left, core too big
                n = n->rb_left;
            } else if (core_id > reg->core_id) { 
                // go right, core too small
                n = n->rb_right;
            } else {
                PrintDebug(vm, VCORE_NONE, "v3_get_mem_region: Impossible!\n");
                return NULL;
            }
        }
    }


    // There is not registered region, so we check if its a valid address in the base region

    return v3_get_base_region(vm, guest_addr);
}



/* This returns the next memory region based on a given address. 
 * If the address falls inside a sub region, that region is returned. 
 * If the address falls outside a sub region, the next sub region is returned
 * NOTE that we have to be careful about core_ids here...
 */
static struct v3_mem_region * get_next_mem_region( struct v3_vm_info * vm, uint16_t core_id, addr_t guest_addr) {
    struct rb_node * n = vm->mem_map.mem_regions.rb_node;
    struct v3_mem_region * reg = NULL;
    struct v3_mem_region * parent = NULL;

    if (n == NULL) {
        return NULL;
    }

    while (n) {

        reg = rb_entry(n, struct v3_mem_region, tree_node);

        if (guest_addr < reg->guest_start) {
            n = n->rb_left;
        } else if (guest_addr >= reg->guest_end) {
            n = n->rb_right;
        } else {
            if (reg->core_id == V3_MEM_CORE_ANY) {
                // found relevant region, it's available on all cores
                return reg;
            } else if (core_id == reg->core_id) { 
                // found relevant region, it's available on the indicated core
                return reg;
            } else if (core_id < reg->core_id) { 
                // go left, core too big
                n = n->rb_left;
            } else if (core_id > reg->core_id) { 
                // go right, core too small
                n = n->rb_right;
            } else {
                PrintError(vm, VCORE_NONE, "v3_get_mem_region: Impossible!\n");
                return NULL;
            }
        }

        if ((reg->core_id == core_id) || (reg->core_id == V3_MEM_CORE_ANY)) {
            parent = reg;
        }
    }


    if (parent->guest_start > guest_addr) {
        return parent;
    } else if (parent->guest_end < guest_addr) {
        struct rb_node * node = &(parent->tree_node);

        while ((node = v3_rb_next(node)) != NULL) {
            struct v3_mem_region * next_reg = rb_entry(node, struct v3_mem_region, tree_node);

            if ((next_reg->core_id == V3_MEM_CORE_ANY) ||
                (next_reg->core_id == core_id)) {

                // This check is not strictly necessary, but it makes it clearer
                if (next_reg->guest_start > guest_addr) {
                    return next_reg;
                }
            }
        }
    }

    return NULL;
}




/* Given an address region of memory, find if there are any regions that overlap with it. 
 * This checks that the range lies in a single region, and returns that region if it does, 
 * this can be either the base region or a sub region. 
 * IF there are multiple regions in the range then it returns NULL
 */
static struct v3_mem_region * get_overlapping_region(struct v3_vm_info * vm, uint16_t core_id, 
                                                     addr_t start_gpa, addr_t end_gpa) {
    struct v3_mem_region * start_region = v3_get_mem_region(vm, core_id, start_gpa);

    if (start_region == NULL) {
        PrintError(vm, VCORE_NONE, "No overlapping region for core=%d, start_gpa=%p\n", core_id, (void*)start_gpa);
        v3_print_mem_map(vm);
        return NULL;
    }


    if (start_region->guest_end < end_gpa) {
        // Region ends before range
        return NULL;
    } else if (start_region->flags.base == 0) {
        // sub region overlaps range
        return start_region;
    } else {
        // Base region, now we have to scan forward for the next sub region
        struct v3_mem_region * next_reg = get_next_mem_region(vm, core_id, start_gpa);
        
        if (next_reg == NULL) {
            // no sub regions after start_addr, base region is ok
            return start_region;
        } else if (next_reg->guest_start >= end_gpa) {
            // Next sub region begins outside range
            return start_region;
        } else {
            return NULL;
        }
    }


    // Should never get here
    return NULL;
}





void v3_delete_mem_region(struct v3_vm_info * vm, struct v3_mem_region * reg) {
    int i = 0;
    int rc;

    if (reg == NULL) {
        return;
    }


    v3_rb_erase(&(reg->tree_node), &(vm->mem_map.mem_regions));



    // If the guest isn't running then there shouldn't be anything to invalidate. 
    // Page tables should __always__ be created on demand during execution
    // NOTE: This is a sanity check, and can be removed if that assumption changes
    if (vm->run_state != VM_RUNNING) {
        V3_Free(reg);
        return;
    }

    rc = 0;

    for (i = 0; i < vm->num_cores; i++) {
        struct guest_info * info = &(vm->cores[i]);

        // flush virtual page tables 
        // 3 cases shadow, shadow passthrough, and nested

        if (info->shdw_pg_mode == SHADOW_PAGING) {
            v3_mem_mode_t mem_mode = v3_get_vm_mem_mode(info);
            
            if (mem_mode == PHYSICAL_MEM) {
              rc |= v3_invalidate_passthrough_addr_range(info,reg->guest_start, reg->guest_end-1);
            } else {
              rc |= v3_invalidate_shadow_pts(info);
            }
            
        } else if (info->shdw_pg_mode == NESTED_PAGING) {
          rc |= v3_invalidate_nested_addr_range(info,reg->guest_start, reg->guest_end-1);
        }
    }

    V3_Free(reg);

    // flush virtual page tables 
    // 3 cases shadow, shadow passthrough, and nested

    if (rc) { PrintError(vm, VCORE_NONE, "Error in deleting memory region\n"); }
}

// Determine if a given address can be handled by a large page of the requested size
uint32_t v3_get_max_page_size(struct guest_info * core, addr_t page_addr, v3_cpu_mode_t mode) {
    addr_t pg_start = 0;
    addr_t pg_end = 0; 
    uint32_t page_size = PAGE_SIZE_4KB;
    struct v3_mem_region * reg = NULL;
    
    switch (mode) {
        case PROTECTED:
            if (core->use_large_pages == 1) {
                pg_start = PAGE_ADDR_4MB(page_addr);
                pg_end = (pg_start + PAGE_SIZE_4MB);

                reg = get_overlapping_region(core->vm_info, core->vcpu_id, pg_start, pg_end); 

                if ((reg) && ((reg->host_addr % PAGE_SIZE_4MB) == 0)) {
                    page_size = PAGE_SIZE_4MB;
                }
            }
            break;
        case PROTECTED_PAE:
            if (core->use_large_pages == 1) {
                pg_start = PAGE_ADDR_2MB(page_addr);
                pg_end = (pg_start + PAGE_SIZE_2MB);

                reg = get_overlapping_region(core->vm_info, core->vcpu_id, pg_start, pg_end);

                if ((reg) && ((reg->host_addr % PAGE_SIZE_2MB) == 0)) {
                    page_size = PAGE_SIZE_2MB;
                }
            }
            break;
        case LONG:
        case LONG_32_COMPAT:
        case LONG_16_COMPAT:
            if (core->use_giant_pages == 1) {
                pg_start = PAGE_ADDR_1GB(page_addr);
                pg_end = (pg_start + PAGE_SIZE_1GB);
                
                reg = get_overlapping_region(core->vm_info, core->vcpu_id, pg_start, pg_end);
                
                if ((reg) && ((reg->host_addr % PAGE_SIZE_1GB) == 0)) {
                    page_size = PAGE_SIZE_1GB;
                    break;
                }
            }

            if (core->use_large_pages == 1) {
                pg_start = PAGE_ADDR_2MB(page_addr);
                pg_end = (pg_start + PAGE_SIZE_2MB);

                reg = get_overlapping_region(core->vm_info, core->vcpu_id, pg_start, pg_end);
                
                if ((reg) && ((reg->host_addr % PAGE_SIZE_2MB) == 0)) {
                    page_size = PAGE_SIZE_2MB;
                }
            }
            break;
        default:
            PrintError(core->vm_info, core, "Invalid CPU mode: %s\n", v3_cpu_mode_to_str(v3_get_vm_cpu_mode(core)));
            return -1;
    }

    return page_size;
}



void v3_print_mem_map(struct v3_vm_info * vm) {
    struct v3_mem_map * map = &(vm->mem_map);
    struct rb_node * node = v3_rb_first(&(vm->mem_map.mem_regions));
    struct v3_mem_region * reg = NULL;
    int i = 0;

    V3_Print(vm, VCORE_NONE, "Memory Layout (all cores):\n");
    
    V3_Print(vm, VCORE_NONE, "Base Memory: (%d regions)\n", map->num_base_regions);

    for (i = 0; i < map->num_base_regions; i++) {
        reg = &(map->base_regions[i]);

        V3_Print(vm, VCORE_NONE, "Base Region[%d] (all cores):  0x%p - 0x%p -> 0x%p\n", 
                 i, 
                 (void *)(reg->guest_start), 
                 (void *)(reg->guest_end - 1), 
                 (void *)(reg->host_addr));
    
    }

    // If the memory map is empty, don't print it
    if (node == NULL) {
        return;
    }

    do {
        reg = rb_entry(node, struct v3_mem_region, tree_node);

        V3_Print(vm, VCORE_NONE, "%d:  0x%p - 0x%p -> 0x%p\n", i, 
                   (void *)(reg->guest_start), 
                   (void *)(reg->guest_end - 1), 
                   (void *)(reg->host_addr));

        V3_Print(vm, VCORE_NONE, "\t(flags=0x%x) (core=0x%x) (unhandled = 0x%p)\n", 
                 reg->flags.value, 
                 reg->core_id,
                 reg->unhandled);
    
        i++;
    } while ((node = v3_rb_next(node)));
}


void v3_init_mem()
{
    char *arg = v3_lookup_option("mem_block_size");

    if (arg) { 
        v3_mem_block_size = atoi(arg);
        V3_Print(VM_NONE,VCORE_NONE,"memory block size set to %llu bytes\n",v3_mem_block_size);
    } else {
        V3_Print(VM_NONE,VCORE_NONE,"default memory block size of %llu bytes is in use\n",v3_mem_block_size);
    }
}

void v3_deinit_mem()
{
    // currently nothing
}