Commit a5deabe0 authored by Andres Lagar-Cavilla's avatar Andres Lagar-Cavilla Committed by Konrad Rzeszutek Wilk

Xen: Fix retry calls into PRIVCMD_MMAPBATCH*.

When a foreign mapper attempts to map guest frames that are paged out,
the mapper receives an ENOENT response and will have to try again
while a helper process pages the target frame back in.

Gating checks on PRIVCMD_MMAPBATCH* ioctl args were preventing retries
of mapping calls.

Permit subsequent calls to update a sub-range of the VMA, iff nothing
is yet mapped in that range.

Since it is now valid to call PRIVCMD_MMAPBATCH* multiple times, only
set vma->vm_private_data if the parameters are valid and (if
necessary) the pages for the auto_translated_physmap case have been
allocated.  This prevents subsequent calls from incorrectly entering
the 'retry' path when there are no pages allocated etc.
Signed-off-by: default avatarAndres Lagar-Cavilla <andres@lagarcavilla.org>
Signed-off-by: default avatarDavid Vrabel <david.vrabel@citrix.com>
Signed-off-by: default avatarKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
parent 669b0ae9
...@@ -43,9 +43,10 @@ MODULE_LICENSE("GPL"); ...@@ -43,9 +43,10 @@ MODULE_LICENSE("GPL");
#define PRIV_VMA_LOCKED ((void *)1) #define PRIV_VMA_LOCKED ((void *)1)
#ifndef HAVE_ARCH_PRIVCMD_MMAP static int privcmd_vma_range_is_mapped(
static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma); struct vm_area_struct *vma,
#endif unsigned long addr,
unsigned long nr_pages);
static long privcmd_ioctl_hypercall(void __user *udata) static long privcmd_ioctl_hypercall(void __user *udata)
{ {
...@@ -225,9 +226,9 @@ static long privcmd_ioctl_mmap(void __user *udata) ...@@ -225,9 +226,9 @@ static long privcmd_ioctl_mmap(void __user *udata)
vma = find_vma(mm, msg->va); vma = find_vma(mm, msg->va);
rc = -EINVAL; rc = -EINVAL;
if (!vma || (msg->va != vma->vm_start) || if (!vma || (msg->va != vma->vm_start) || vma->vm_private_data)
!privcmd_enforce_singleshot_mapping(vma))
goto out_up; goto out_up;
vma->vm_private_data = PRIV_VMA_LOCKED;
} }
state.va = vma->vm_start; state.va = vma->vm_start;
...@@ -358,7 +359,7 @@ static int alloc_empty_pages(struct vm_area_struct *vma, int numpgs) ...@@ -358,7 +359,7 @@ static int alloc_empty_pages(struct vm_area_struct *vma, int numpgs)
kfree(pages); kfree(pages);
return -ENOMEM; return -ENOMEM;
} }
BUG_ON(vma->vm_private_data != PRIV_VMA_LOCKED); BUG_ON(vma->vm_private_data != NULL);
vma->vm_private_data = pages; vma->vm_private_data = pages;
return 0; return 0;
...@@ -421,19 +422,43 @@ static long privcmd_ioctl_mmap_batch(void __user *udata, int version) ...@@ -421,19 +422,43 @@ static long privcmd_ioctl_mmap_batch(void __user *udata, int version)
vma = find_vma(mm, m.addr); vma = find_vma(mm, m.addr);
if (!vma || if (!vma ||
vma->vm_ops != &privcmd_vm_ops || vma->vm_ops != &privcmd_vm_ops) {
(m.addr != vma->vm_start) ||
((m.addr + (nr_pages << PAGE_SHIFT)) != vma->vm_end) ||
!privcmd_enforce_singleshot_mapping(vma)) {
up_write(&mm->mmap_sem);
ret = -EINVAL; ret = -EINVAL;
goto out; goto out_unlock;
}
/*
* Caller must either:
*
* Map the whole VMA range, which will also allocate all the
* pages required for the auto_translated_physmap case.
*
* Or
*
* Map unmapped holes left from a previous map attempt (e.g.,
* because those foreign frames were previously paged out).
*/
if (vma->vm_private_data == NULL) {
if (m.addr != vma->vm_start ||
m.addr + (nr_pages << PAGE_SHIFT) != vma->vm_end) {
ret = -EINVAL;
goto out_unlock;
} }
if (xen_feature(XENFEAT_auto_translated_physmap)) { if (xen_feature(XENFEAT_auto_translated_physmap)) {
ret = alloc_empty_pages(vma, m.num); ret = alloc_empty_pages(vma, m.num);
if (ret < 0) { if (ret < 0)
up_write(&mm->mmap_sem); goto out_unlock;
goto out; } else
vma->vm_private_data = PRIV_VMA_LOCKED;
} else {
if (m.addr < vma->vm_start ||
m.addr + (nr_pages << PAGE_SHIFT) > vma->vm_end) {
ret = -EINVAL;
goto out_unlock;
}
if (privcmd_vma_range_is_mapped(vma, m.addr, nr_pages)) {
ret = -EINVAL;
goto out_unlock;
} }
} }
...@@ -466,8 +491,11 @@ static long privcmd_ioctl_mmap_batch(void __user *udata, int version) ...@@ -466,8 +491,11 @@ static long privcmd_ioctl_mmap_batch(void __user *udata, int version)
out: out:
free_page_list(&pagelist); free_page_list(&pagelist);
return ret; return ret;
out_unlock:
up_write(&mm->mmap_sem);
goto out;
} }
static long privcmd_ioctl(struct file *file, static long privcmd_ioctl(struct file *file,
...@@ -540,9 +568,24 @@ static int privcmd_mmap(struct file *file, struct vm_area_struct *vma) ...@@ -540,9 +568,24 @@ static int privcmd_mmap(struct file *file, struct vm_area_struct *vma)
return 0; return 0;
} }
static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma) /*
* For MMAPBATCH*. This allows asserting the singleshot mapping
* on a per pfn/pte basis. Mapping calls that fail with ENOENT
* can be then retried until success.
*/
static int is_mapped_fn(pte_t *pte, struct page *pmd_page,
unsigned long addr, void *data)
{
return pte_none(*pte) ? 0 : -EBUSY;
}
static int privcmd_vma_range_is_mapped(
struct vm_area_struct *vma,
unsigned long addr,
unsigned long nr_pages)
{ {
return !cmpxchg(&vma->vm_private_data, NULL, PRIV_VMA_LOCKED); return apply_to_page_range(vma->vm_mm, addr, nr_pages << PAGE_SHIFT,
is_mapped_fn, NULL) != 0;
} }
const struct file_operations xen_privcmd_fops = { const struct file_operations xen_privcmd_fops = {
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment