* Copyright (c) 2002-2004, K A Fraser, B Dragovic
*/
-#include <linux/config.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/highmem.h>
#include <linux/pagemap.h>
#include <linux/seq_file.h>
+#include <linux/kthread.h>
+#include <asm/hypervisor.h>
#include <asm/pgalloc.h>
#include <asm/pgtable.h>
#include <asm/uaccess.h>
#include <asm/tlb.h>
-#include <asm-xen/linux-public/privcmd.h>
-#include <asm-xen/xen-public/dom0_ops.h>
-#include <asm-xen/xen_proc.h>
-
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
-#define pud_t pgd_t
-#define pud_offset(d, va) d
-#endif
+#include <asm/hypervisor.h>
+#include <xen/public/privcmd.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/dom0_ops.h>
+#include <xen/xen_proc.h>
static struct proc_dir_entry *privcmd_intf;
+static struct proc_dir_entry *capabilities_intf;
+
+#ifndef HAVE_ARCH_PRIVCMD_MMAP
+static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma);
+#endif
static int privcmd_ioctl(struct inode *inode, struct file *file,
- unsigned int cmd, unsigned long data)
+ unsigned int cmd, unsigned long data)
{
- int ret = -ENOSYS;
+ int ret = -ENOSYS;
+ void __user *udata = (void __user *) data;
- switch ( cmd )
- {
- case IOCTL_PRIVCMD_HYPERCALL:
- {
- privcmd_hypercall_t hypercall;
+ switch (cmd) {
+ case IOCTL_PRIVCMD_HYPERCALL: {
+ privcmd_hypercall_t hypercall;
- if ( copy_from_user(&hypercall, (void *)data, sizeof(hypercall)) )
- return -EFAULT;
+ if (copy_from_user(&hypercall, udata, sizeof(hypercall)))
+ return -EFAULT;
#if defined(__i386__)
- __asm__ __volatile__ (
- "pushl %%ebx; pushl %%ecx; pushl %%edx; pushl %%esi; pushl %%edi; "
- "movl 4(%%eax),%%ebx ;"
- "movl 8(%%eax),%%ecx ;"
- "movl 12(%%eax),%%edx ;"
- "movl 16(%%eax),%%esi ;"
- "movl 20(%%eax),%%edi ;"
- "movl (%%eax),%%eax ;"
- TRAP_INSTR "; "
- "popl %%edi; popl %%esi; popl %%edx; popl %%ecx; popl %%ebx"
- : "=a" (ret) : "0" (&hypercall) : "memory" );
+ if (hypercall.op >= (PAGE_SIZE >> 5))
+ break;
+ __asm__ __volatile__ (
+ "pushl %%ebx; pushl %%ecx; pushl %%edx; "
+ "pushl %%esi; pushl %%edi; "
+ "movl 8(%%eax),%%ebx ;"
+ "movl 16(%%eax),%%ecx ;"
+ "movl 24(%%eax),%%edx ;"
+ "movl 32(%%eax),%%esi ;"
+ "movl 40(%%eax),%%edi ;"
+ "movl (%%eax),%%eax ;"
+ "shll $5,%%eax ;"
+ "addl $hypercall_page,%%eax ;"
+ "call *%%eax ;"
+ "popl %%edi; popl %%esi; popl %%edx; "
+ "popl %%ecx; popl %%ebx"
+ : "=a" (ret) : "0" (&hypercall) : "memory" );
#elif defined (__x86_64__)
- __asm__ __volatile__ (
- "movq %5,%%r10; movq %6,%%r8;" TRAP_INSTR
- : "=a" (ret)
- : "a" ((unsigned long)hypercall.op),
- "D" ((unsigned long)hypercall.arg[0]),
- "S" ((unsigned long)hypercall.arg[1]),
- "d" ((unsigned long)hypercall.arg[2]),
- "g" ((unsigned long)hypercall.arg[3]),
- "g" ((unsigned long)hypercall.arg[4])
- : "r11","rcx","r8","r10","memory");
-#endif
- }
- break;
-
- case IOCTL_PRIVCMD_INITDOMAIN_EVTCHN:
- {
- extern int initdom_ctrlif_domcontroller_port;
- ret = initdom_ctrlif_domcontroller_port;
- }
- break;
-
-#if defined(CONFIG_XEN_PRIVILEGED_GUEST)
- case IOCTL_PRIVCMD_MMAP:
- {
-#define PRIVCMD_MMAP_SZ 32
- privcmd_mmap_t mmapcmd;
- privcmd_mmap_entry_t msg[PRIVCMD_MMAP_SZ], *p;
- int i, rc;
-
- if ( copy_from_user(&mmapcmd, (void *)data, sizeof(mmapcmd)) )
- return -EFAULT;
-
- p = mmapcmd.entry;
-
- for (i=0; i<mmapcmd.num; i+=PRIVCMD_MMAP_SZ, p+=PRIVCMD_MMAP_SZ)
- {
- int j, n = ((mmapcmd.num-i)>PRIVCMD_MMAP_SZ)?
- PRIVCMD_MMAP_SZ:(mmapcmd.num-i);
-
-
- if ( copy_from_user(&msg, p, n*sizeof(privcmd_mmap_entry_t)) )
- return -EFAULT;
-
- for ( j = 0; j < n; j++ )
- {
- struct vm_area_struct *vma =
- find_vma( current->mm, msg[j].va );
-
- if ( !vma )
- return -EINVAL;
-
- if ( msg[j].va > PAGE_OFFSET )
- return -EINVAL;
-
- if ( (msg[j].va + (msg[j].npages<<PAGE_SHIFT)) > vma->vm_end )
- return -EINVAL;
-
- if ( (rc = direct_remap_area_pages(vma->vm_mm,
- msg[j].va&PAGE_MASK,
- msg[j].mfn<<PAGE_SHIFT,
- msg[j].npages<<PAGE_SHIFT,
- vma->vm_page_prot,
- mmapcmd.dom)) < 0 )
- return rc;
- }
- }
- ret = 0;
- }
- break;
-
- case IOCTL_PRIVCMD_MMAPBATCH:
- {
- mmu_update_t u;
- privcmd_mmapbatch_t m;
- struct vm_area_struct *vma = NULL;
- unsigned long *p, addr;
- unsigned long mfn;
- int i;
-
- if ( copy_from_user(&m, (void *)data, sizeof(m)) )
- { ret = -EFAULT; goto batch_err; }
-
- vma = find_vma( current->mm, m.addr );
-
- if ( !vma )
- { ret = -EINVAL; goto batch_err; }
-
- if ( m.addr > PAGE_OFFSET )
- { ret = -EFAULT; goto batch_err; }
-
- if ( (m.addr + (m.num<<PAGE_SHIFT)) > vma->vm_end )
- { ret = -EFAULT; goto batch_err; }
-
- p = m.arr;
- addr = m.addr;
- for ( i = 0; i < m.num; i++, addr += PAGE_SIZE, p++ )
- {
- if ( get_user(mfn, p) )
- return -EFAULT;
-
- u.val = (mfn << PAGE_SHIFT) | pgprot_val(vma->vm_page_prot);
-
- __direct_remap_area_pages(vma->vm_mm,
- addr,
- PAGE_SIZE,
- &u);
-
- if ( unlikely(HYPERVISOR_mmu_update(&u, 1, NULL, m.dom) < 0) )
- put_user(0xF0000000 | mfn, p);
- }
-
- ret = 0;
- break;
-
- batch_err:
- printk("batch_err ret=%d vma=%p addr=%lx num=%d arr=%p %lx-%lx\n",
- ret, vma, m.addr, m.num, m.arr,
- vma ? vma->vm_start : 0, vma ? vma->vm_end : 0);
- break;
- }
- break;
+ if (hypercall.op < (PAGE_SIZE >> 5)) {
+ long ign1, ign2, ign3;
+ __asm__ __volatile__ (
+ "movq %8,%%r10; movq %9,%%r8;"
+ "shll $5,%%eax ;"
+ "addq $hypercall_page,%%rax ;"
+ "call *%%rax"
+ : "=a" (ret), "=D" (ign1),
+ "=S" (ign2), "=d" (ign3)
+ : "0" ((unsigned int)hypercall.op),
+ "1" (hypercall.arg[0]),
+ "2" (hypercall.arg[1]),
+ "3" (hypercall.arg[2]),
+ "g" (hypercall.arg[3]),
+ "g" (hypercall.arg[4])
+ : "r8", "r10", "memory" );
+ }
+#elif defined (__ia64__)
+ __asm__ __volatile__ (
+ ";; mov r14=%2; mov r15=%3; "
+ "mov r16=%4; mov r17=%5; mov r18=%6;"
+ "mov r2=%1; break 0x1000;; mov %0=r8 ;;"
+ : "=r" (ret)
+ : "r" (hypercall.op),
+ "r" (hypercall.arg[0]),
+ "r" (hypercall.arg[1]),
+ "r" (hypercall.arg[2]),
+ "r" (hypercall.arg[3]),
+ "r" (hypercall.arg[4])
+ : "r14","r15","r16","r17","r18","r2","r8","memory");
#endif
+ }
+ break;
+
+ case IOCTL_PRIVCMD_MMAP: {
+ privcmd_mmap_t mmapcmd;
+ privcmd_mmap_entry_t msg;
+ privcmd_mmap_entry_t __user *p;
+ struct mm_struct *mm = current->mm;
+ struct vm_area_struct *vma;
+ unsigned long va;
+ int i, rc;
+
+ if (!is_initial_xendomain())
+ return -EPERM;
+
+ if (copy_from_user(&mmapcmd, udata, sizeof(mmapcmd)))
+ return -EFAULT;
+
+ p = mmapcmd.entry;
+ if (copy_from_user(&msg, p, sizeof(msg)))
+ return -EFAULT;
+
+ down_read(&mm->mmap_sem);
+
+ vma = find_vma(mm, msg.va);
+ rc = -EINVAL;
+ if (!vma || (msg.va != vma->vm_start) ||
+ !privcmd_enforce_singleshot_mapping(vma))
+ goto mmap_out;
+
+ va = vma->vm_start;
+
+ for (i = 0; i < mmapcmd.num; i++) {
+ rc = -EFAULT;
+ if (copy_from_user(&msg, p, sizeof(msg)))
+ goto mmap_out;
+
+ /* Do not allow range to wrap the address space. */
+ rc = -EINVAL;
+ if ((msg.npages > (LONG_MAX >> PAGE_SHIFT)) ||
+ ((unsigned long)(msg.npages << PAGE_SHIFT) >= -va))
+ goto mmap_out;
+
+ /* Range chunks must be contiguous in va space. */
+ if ((msg.va != va) ||
+ ((msg.va+(msg.npages<<PAGE_SHIFT)) > vma->vm_end))
+ goto mmap_out;
+
+ if ((rc = direct_remap_pfn_range(
+ vma,
+ msg.va & PAGE_MASK,
+ msg.mfn,
+ msg.npages << PAGE_SHIFT,
+ vma->vm_page_prot,
+ mmapcmd.dom)) < 0)
+ goto mmap_out;
+
+ p++;
+ va += msg.npages << PAGE_SHIFT;
+ }
+
+ rc = 0;
+
+ mmap_out:
+ up_read(&mm->mmap_sem);
+ ret = rc;
+ }
+ break;
+
+ case IOCTL_PRIVCMD_MMAPBATCH: {
+ privcmd_mmapbatch_t m;
+ struct mm_struct *mm = current->mm;
+ struct vm_area_struct *vma;
+ xen_pfn_t __user *p;
+ unsigned long addr, mfn;
+ int i;
+
+ if (!is_initial_xendomain())
+ return -EPERM;
+
+ if (copy_from_user(&m, udata, sizeof(m)))
+ return -EFAULT;
+
+ if ((m.num <= 0) || (m.num > (LONG_MAX >> PAGE_SHIFT)))
+ return -EINVAL;
+
+ down_read(&mm->mmap_sem);
+
+ vma = find_vma(mm, m.addr);
+ if (!vma ||
+ (m.addr != vma->vm_start) ||
+ ((m.addr + ((unsigned long)m.num<<PAGE_SHIFT)) !=
+ vma->vm_end) ||
+ !privcmd_enforce_singleshot_mapping(vma)) {
+ up_read(&mm->mmap_sem);
+ return -EINVAL;
+ }
+
+ p = m.arr;
+ addr = m.addr;
+ for (i = 0; i < m.num; i++, addr += PAGE_SIZE, p++) {
+ if (get_user(mfn, p)) {
+ up_read(&mm->mmap_sem);
+ return -EFAULT;
+ }
+
+ ret = direct_remap_pfn_range(vma, addr & PAGE_MASK,
+ mfn, PAGE_SIZE,
+ vma->vm_page_prot, m.dom);
+ if (ret < 0)
+ put_user(0xF0000000 | mfn, p);
+ }
+
+ up_read(&mm->mmap_sem);
+ ret = 0;
+ }
+ break;
+
+ default:
+ ret = -EINVAL;
+ break;
+ }
+
+ return ret;
+}
- case IOCTL_PRIVCMD_GET_MACH2PHYS_START_MFN:
- {
- unsigned long m2pv = (unsigned long)machine_to_phys_mapping;
- pgd_t *pgd = pgd_offset_k(m2pv);
- pud_t *pud = pud_offset(pgd, m2pv);
- pmd_t *pmd = pmd_offset(pud, m2pv);
- unsigned long m2p_start_mfn = (*(unsigned long *)pmd) >> PAGE_SHIFT;
- ret = put_user(m2p_start_mfn, (unsigned long *)data) ? -EFAULT: 0;
- }
- break;
-
- default:
- ret = -EINVAL;
- break;
- }
- return ret;
+#ifndef HAVE_ARCH_PRIVCMD_MMAP
+static struct page *privcmd_nopage(struct vm_area_struct *vma,
+ unsigned long address,
+ int *type)
+{
+ return NOPAGE_SIGBUS;
}
+static struct vm_operations_struct privcmd_vm_ops = {
+ .nopage = privcmd_nopage
+};
+
static int privcmd_mmap(struct file * file, struct vm_area_struct * vma)
{
- /* DONTCOPY is essential for Xen as copy_page_range is broken. */
- vma->vm_flags |= VM_RESERVED | VM_IO | VM_DONTCOPY;
+ /* Unsupported for auto-translate guests. */
+ if (xen_feature(XENFEAT_auto_translated_physmap))
+ return -ENOSYS;
+
+ /* DONTCOPY is essential for Xen as copy_page_range is broken. */
+ vma->vm_flags |= VM_RESERVED | VM_IO | VM_DONTCOPY | VM_PFNMAP;
+ vma->vm_ops = &privcmd_vm_ops;
+ vma->vm_private_data = NULL;
- return 0;
+ return 0;
}
+static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
+{
+ return (xchg(&vma->vm_private_data, (void *)1) == NULL);
+}
+#endif
+
static struct file_operations privcmd_file_ops = {
- ioctl : privcmd_ioctl,
- mmap: privcmd_mmap
+ .ioctl = privcmd_ioctl,
+ .mmap = privcmd_mmap,
};
+static int capabilities_read(char *page, char **start, off_t off,
+ int count, int *eof, void *data)
+{
+ int len = 0;
+ *page = 0;
+
+ if (is_initial_xendomain())
+ len = sprintf( page, "control_d\n" );
+
+ *eof = 1;
+ return len;
+}
static int __init privcmd_init(void)
{
- if ( !(xen_start_info.flags & SIF_PRIVILEGED) )
- return 0;
+ if (!is_running_on_xen())
+ return -ENODEV;
+
+ privcmd_intf = create_xen_proc_entry("privcmd", 0400);
+ if (privcmd_intf != NULL)
+ privcmd_intf->proc_fops = &privcmd_file_ops;
- privcmd_intf = create_xen_proc_entry("privcmd", 0400);
- if ( privcmd_intf != NULL )
- privcmd_intf->proc_fops = &privcmd_file_ops;
+ capabilities_intf = create_xen_proc_entry("capabilities", 0400 );
+ if (capabilities_intf != NULL)
+ capabilities_intf->read_proc = capabilities_read;
- return 0;
+ return 0;
}
__initcall(privcmd_init);
git://git.onelab.eu / linux-2.6.git / blobdiff