The new GNTTABOP_unmap_and_duplicate operation doesn't zero the
mapping passed in new_addr, allowing us to perform batch unmaps in p2m
code without requiring the use of a multicall.
Signed-off-by: Roger Pau Monné <[email protected]>
Cc: Stefano Stabellini <[email protected]>
Cc: Konrad Rzeszutek Wilk <[email protected]>
---
I don't currently have a NFS server (the one I had is currently not
working due to SD card corruption) and I cannot set up one right now,
so I've only tested this with a raw image stored in a local disk.
---
arch/x86/include/asm/xen/page.h | 4 +-
arch/x86/xen/p2m.c | 131 ++++++++++++++++++++++++++++++++++-
drivers/xen/grant-table.c | 24 ++----
include/xen/interface/grant_table.h | 22 ++++++
4 files changed, 162 insertions(+), 19 deletions(-)
diff --git a/arch/x86/include/asm/xen/page.h b/arch/x86/include/asm/xen/page.h
index 6aef9fb..ea1dce6 100644
--- a/arch/x86/include/asm/xen/page.h
+++ b/arch/x86/include/asm/xen/page.h
@@ -51,8 +51,8 @@ extern unsigned long set_phys_range_identity(unsigned long pfn_s,
extern int m2p_add_override(unsigned long mfn, struct page *page,
struct gnttab_map_grant_ref *kmap_op);
-extern int m2p_remove_override(struct page *page,
- struct gnttab_map_grant_ref *kmap_op);
+extern int m2p_remove_override(struct page **pages,
+ struct gnttab_map_grant_ref *kmap_ops, int count);
extern struct page *m2p_find_override(unsigned long mfn);
extern unsigned long m2p_find_override_pfn(unsigned long mfn, unsigned long pfn);
diff --git a/arch/x86/xen/p2m.c b/arch/x86/xen/p2m.c
index 0d4ec35..e3d2668 100644
--- a/arch/x86/xen/p2m.c
+++ b/arch/x86/xen/p2m.c
@@ -154,6 +154,8 @@
#include <linux/hash.h>
#include <linux/sched.h>
#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/hardirq.h>
#include <asm/cache.h>
#include <asm/setup.h>
@@ -853,6 +855,7 @@ bool set_phys_to_machine(unsigned long pfn, unsigned long mfn)
static RESERVE_BRK_ARRAY(struct list_head, m2p_overrides, M2P_OVERRIDE_HASH);
static DEFINE_SPINLOCK(m2p_override_lock);
+extern bool gnttab_supports_duplicate;
static void __init m2p_override_init(void)
{
@@ -933,7 +936,7 @@ int m2p_add_override(unsigned long mfn, struct page *page,
return 0;
}
EXPORT_SYMBOL_GPL(m2p_add_override);
-int m2p_remove_override(struct page *page,
+int m2p_remove_override_single(struct page *page,
struct gnttab_map_grant_ref *kmap_op)
{
unsigned long flags;
@@ -1034,6 +1037,132 @@ int m2p_remove_override(struct page *page,
return 0;
}
+
+int m2p_remove_override_batch(struct page **pages,
+ struct gnttab_map_grant_ref *kmap_ops, int count)
+{
+ struct gnttab_unmap_and_duplicate *unmap_ops = NULL;
+ unsigned long *mfn = kcalloc(count, sizeof(mfn[0]), GFP_KERNEL);
+ unsigned long flags;
+ unsigned long pfn;
+ unsigned long uninitialized_var(address);
+ unsigned level;
+ struct page *page;
+ pte_t *ptep = NULL;
+ int ret = 0, i;
+
+
+ for (i = 0; i < count; i++) {
+ page = pages[i];
+ pfn = page_to_pfn(page);
+ mfn[i] = get_phys_to_machine(pfn);
+ if (mfn[i] == INVALID_P2M_ENTRY || !(mfn[i] & FOREIGN_FRAME_BIT)) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ if (!PageHighMem(page)) {
+ address = (unsigned long)__va(pfn << PAGE_SHIFT);
+ ptep = lookup_address(address, &level);
+
+ if (WARN(ptep == NULL || level != PG_LEVEL_4K,
+ "m2p_remove_override: pfn %lx not mapped", pfn)) {
+ ret = -EINVAL;
+ goto out;
+ }
+ }
+
+ spin_lock_irqsave(&m2p_override_lock, flags);
+ list_del(&page->lru);
+ spin_unlock_irqrestore(&m2p_override_lock, flags);
+ WARN_ON(!PagePrivate(page));
+ ClearPagePrivate(page);
+
+ set_phys_to_machine(pfn, page->index);
+ }
+
+ if (kmap_ops != NULL) {
+ struct page *scratch_page = get_balloon_scratch_page();
+ unsigned long scratch_page_address = (unsigned long)
+ __va(page_to_pfn(scratch_page) << PAGE_SHIFT);
+ int invcount = 0;
+
+ unmap_ops = kcalloc(count, sizeof(unmap_ops[0]), GFP_KERNEL);
+ if (!unmap_ops) {
+ put_balloon_scratch_page();
+ ret = -1;
+ goto out;
+ }
+ for (i = 0; i < count; i++) {
+ page = pages[i];
+ if (!PageHighMem(page)) {
+ unmap_ops[invcount].host_addr = kmap_ops[i].host_addr;
+ unmap_ops[invcount].new_addr = scratch_page_address;
+ unmap_ops[invcount].handle = kmap_ops[i].handle;
+ invcount++;
+ }
+ }
+ ret = HYPERVISOR_grant_table_op(GNTTABOP_unmap_and_duplicate, unmap_ops, invcount);
+ if (ret) {
+ printk(KERN_WARNING "m2p_remove_override_batch: GNTTABOP_unmap_and_duplicate failed");
+ put_balloon_scratch_page();
+ ret = -1;
+ goto out;
+ }
+ put_balloon_scratch_page();
+ }
+
+ for (i = 0; i < count; i++) {
+ /* p2m(m2p(mfn)) == FOREIGN_FRAME(mfn): the mfn is already present
+ * somewhere in this domain, even before being added to the
+ * m2p_override (see comment above in m2p_add_override).
+ * If there are no other entries in the m2p_override corresponding
+ * to this mfn, then remove the FOREIGN_FRAME_BIT from the p2m for
+ * the original pfn (the one shared by the frontend): the backend
+ * cannot do any IO on this page anymore because it has been
+ * unshared. Removing the FOREIGN_FRAME_BIT from the p2m entry of
+ * the original pfn causes mfn_to_pfn(mfn) to return the frontend
+ * pfn again. */
+ pfn = page_to_pfn(pages[i]);
+ mfn[i] &= ~FOREIGN_FRAME_BIT;
+ ret = __get_user(pfn, &machine_to_phys_mapping[mfn[i]]);
+ if (ret == 0 && get_phys_to_machine(pfn) == FOREIGN_FRAME(mfn[i]) &&
+ m2p_find_override(mfn[i]) == NULL)
+ set_phys_to_machine(pfn, mfn[i]);
+ }
+
+out:
+ kfree(unmap_ops);
+ kfree(mfn);
+ return ret;
+}
+
+int m2p_remove_override(struct page **pages,
+ struct gnttab_map_grant_ref *kmap_ops, int count)
+{
+ int i, ret;
+ bool lazy = false;
+
+ if (gnttab_supports_duplicate) {
+ ret = m2p_remove_override_batch(pages, kmap_ops, count);
+ if (ret)
+ return ret;
+ } else {
+ if (!in_interrupt() && paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE) {
+ arch_enter_lazy_mmu_mode();
+ lazy = true;
+ }
+ for (i = 0; i < count; i++) {
+ ret = m2p_remove_override_single(pages[i], kmap_ops ?
+ &kmap_ops[i] : NULL);
+ if (ret)
+ return ret;
+ }
+ if (lazy)
+ arch_leave_lazy_mmu_mode();
+ }
+ return ret;
+}
EXPORT_SYMBOL_GPL(m2p_remove_override);
struct page *m2p_find_override(unsigned long mfn)
diff --git a/drivers/xen/grant-table.c b/drivers/xen/grant-table.c
index d5418c1..835f600 100644
--- a/drivers/xen/grant-table.c
+++ b/drivers/xen/grant-table.c
@@ -64,6 +64,7 @@ static int gnttab_free_count;
static grant_ref_t gnttab_free_head;
static DEFINE_SPINLOCK(gnttab_list_lock);
unsigned long xen_hvm_resume_frames;
+bool gnttab_supports_duplicate;
EXPORT_SYMBOL_GPL(xen_hvm_resume_frames);
static union {
@@ -934,8 +935,7 @@ int gnttab_unmap_refs(struct gnttab_unmap_grant_ref *unmap_ops,
struct gnttab_map_grant_ref *kmap_ops,
struct page **pages, unsigned int count)
{
- int i, ret;
- bool lazy = false;
+ int ret;
ret = HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, unmap_ops, count);
if (ret)
@@ -944,20 +944,7 @@ int gnttab_unmap_refs(struct gnttab_unmap_grant_ref *unmap_ops,
if (xen_feature(XENFEAT_auto_translated_physmap))
return ret;
- if (!in_interrupt() && paravirt_get_lazy_mode() == PARAVIRT_LAZY_NONE) {
- arch_enter_lazy_mmu_mode();
- lazy = true;
- }
-
- for (i = 0; i < count; i++) {
- ret = m2p_remove_override(pages[i], kmap_ops ?
- &kmap_ops[i] : NULL);
- if (ret)
- return ret;
- }
-
- if (lazy)
- arch_leave_lazy_mmu_mode();
+ ret = m2p_remove_override(pages, kmap_ops, count);
return ret;
}
@@ -1247,6 +1234,11 @@ int gnttab_init(void)
gnttab_free_count = nr_init_grefs - NR_RESERVED_ENTRIES;
gnttab_free_head = NR_RESERVED_ENTRIES;
+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_and_duplicate, NULL, 0))
+ gnttab_supports_duplicate = false;
+ else
+ gnttab_supports_duplicate = true;
+
printk("Grant table initialized\n");
return 0;
diff --git a/include/xen/interface/grant_table.h b/include/xen/interface/grant_table.h
index e40fae9..1f49dc1 100644
--- a/include/xen/interface/grant_table.h
+++ b/include/xen/interface/grant_table.h
@@ -428,6 +428,27 @@ struct gnttab_unmap_and_replace {
DEFINE_GUEST_HANDLE_STRUCT(gnttab_unmap_and_replace);
/*
+ * GNTTABOP_unmap_and_duplicate: Destroy one or more grant-reference mappings
+ * tracked by <handle> but atomically replace the page table entry with one
+ * pointing to the machine address under <new_addr>.
+ * NOTES:
+ * 1. The call may fail in an undefined manner if either mapping is not
+ * tracked by <handle>.
+ * 2. After executing a batch of unmaps, it is guaranteed that no stale
+ * mappings will remain in the device or host TLBs.
+ */
+#define GNTTABOP_unmap_and_duplicate 12
+struct gnttab_unmap_and_duplicate {
+ /* IN parameters. */
+ uint64_t host_addr;
+ uint64_t new_addr;
+ grant_handle_t handle;
+ /* OUT parameters. */
+ int16_t status; /* => enum grant_status */
+};
+DEFINE_GUEST_HANDLE_STRUCT(gnttab_unmap_and_duplicate_t);
+
+/*
* GNTTABOP_set_version: Request a particular version of the grant
* table shared table structure. This operation can only be performed
* once in any given domain. It must be performed before any grants
@@ -522,6 +543,7 @@ DEFINE_GUEST_HANDLE_STRUCT(gnttab_get_version);
#define GNTST_bad_copy_arg (-10) /* copy arguments cross page boundary. */
#define GNTST_address_too_big (-11) /* transfer page address too large. */
#define GNTST_eagain (-12) /* Operation not done; try again. */
+#define GNTST_enosys (-13) /* Operation not implemented. */
#define GNTTABOP_error_msgs { \
"okay", \
--
1.7.7.5 (Apple Git-26)
On 31/07/13 17:07, Roger Pau Monne wrote:
> The new GNTTABOP_unmap_and_duplicate operation doesn't zero the
> mapping passed in new_addr, allowing us to perform batch unmaps in p2m
> code without requiring the use of a multicall.
Thanks. This looks like it should be a good performance improvement for
userspace-based backends. Do you have any performance measurements for
(e.g.) qemu's disk backend?
> --- a/arch/x86/xen/p2m.c
> +++ b/arch/x86/xen/p2m.c
[...]
> +int m2p_remove_override_batch(struct page **pages,
> + struct gnttab_map_grant_ref *kmap_ops, int count)
> +{
Is it possible to refactor this function and
m2p_remove_override_single() to share some common helper functions?
> + struct gnttab_unmap_and_duplicate *unmap_ops = NULL;
> + unsigned long *mfn = kcalloc(count, sizeof(mfn[0]), GFP_KERNEL);
> + unsigned long flags;
> + unsigned long pfn;
> + unsigned long uninitialized_var(address);
> + unsigned level;
> + struct page *page;
> + pte_t *ptep = NULL;
> + int ret = 0, i;
> +
> +
> + for (i = 0; i < count; i++) {
> + page = pages[i];
> + pfn = page_to_pfn(page);
The bits in this loop...
[...]
> + for (i = 0; i < count; i++) {
> + /* p2m(m2p(mfn)) == FOREIGN_FRAME(mfn): the mfn is already present
> + * somewhere in this domain, even before being added to the
... and this one look they should be common between the two.
David
On 01/08/13 11:53, David Vrabel wrote:
> On 31/07/13 17:07, Roger Pau Monne wrote:
>> The new GNTTABOP_unmap_and_duplicate operation doesn't zero the
>> mapping passed in new_addr, allowing us to perform batch unmaps in p2m
>> code without requiring the use of a multicall.
>
> Thanks. This looks like it should be a good performance improvement for
> userspace-based backends. Do you have any performance measurements for
> (e.g.) qemu's disk backend?
Not yet, I might be able to get some initial performance measurements
tomorrow.
>
>> --- a/arch/x86/xen/p2m.c
>> +++ b/arch/x86/xen/p2m.c
> [...]
>> +int m2p_remove_override_batch(struct page **pages,
>> + struct gnttab_map_grant_ref *kmap_ops, int count)
>> +{
>
> Is it possible to refactor this function and
> m2p_remove_override_single() to share some common helper functions?
Yes, I will split those common parts into inline functions and share
them between both implementations.
Roger.