When hypervisor decides to decrease target balloon size while the
balloon driver tries to lock pages hypervisor may respond with
VMW_BALLOON_PPN_NOTNEEDED. Use this data and immediately stop reserving
pages and wait for the next update cycle to fetch new target instead
of continuing trying to lock pages until size of refused list grows
above VMW_BALLOON_MAX_REFUSED (16) pages.
Signed-off-by: Dmitry Torokhov <[email protected]>
---
drivers/misc/vmw_balloon.c | 13 ++++++++-----
1 files changed, 8 insertions(+), 5 deletions(-)
diff --git a/drivers/misc/vmw_balloon.c b/drivers/misc/vmw_balloon.c
index 2a1e804..4263d9d 100644
--- a/drivers/misc/vmw_balloon.c
+++ b/drivers/misc/vmw_balloon.c
@@ -45,7 +45,7 @@
MODULE_AUTHOR("VMware, Inc.");
MODULE_DESCRIPTION("VMware Memory Control (Balloon) Driver");
-MODULE_VERSION("1.2.1.1-k");
+MODULE_VERSION("1.2.1.2-k");
MODULE_ALIAS("dmi:*:svnVMware*:*");
MODULE_ALIAS("vmware_vmmemctl");
MODULE_LICENSE("GPL");
@@ -315,7 +315,8 @@ static bool vmballoon_send_get_target(struct vmballoon *b, u32 *new_target)
* fear that guest will need it. Host may reject some pages, we need to
* check the return value and maybe submit a different page.
*/
-static bool vmballoon_send_lock_page(struct vmballoon *b, unsigned long pfn)
+static bool vmballoon_send_lock_page(struct vmballoon *b, unsigned long pfn,
+ unsigned int *hv_status)
{
unsigned long status, dummy;
u32 pfn32;
@@ -326,7 +327,7 @@ static bool vmballoon_send_lock_page(struct vmballoon *b, unsigned long pfn)
STATS_INC(b->stats.lock);
- status = VMWARE_BALLOON_CMD(LOCK, pfn, dummy);
+ *hv_status = status = VMWARE_BALLOON_CMD(LOCK, pfn, dummy);
if (vmballoon_check_status(b, status))
return true;
@@ -410,6 +411,7 @@ static int vmballoon_reserve_page(struct vmballoon *b, bool can_sleep)
{
struct page *page;
gfp_t flags;
+ unsigned int hv_status;
bool locked = false;
do {
@@ -429,11 +431,12 @@ static int vmballoon_reserve_page(struct vmballoon *b, bool can_sleep)
}
/* inform monitor */
- locked = vmballoon_send_lock_page(b, page_to_pfn(page));
+ locked = vmballoon_send_lock_page(b, page_to_pfn(page), &hv_status);
if (!locked) {
STATS_INC(b->stats.refused_alloc);
- if (b->reset_required) {
+ if (hv_status == VMW_BALLOON_ERROR_RESET ||
+ hv_status == VMW_BALLOON_ERROR_PPN_NOTNEEDED) {
__free_page(page);
return -EIO;
}
--
1.7.3.2
On Fri, 3 Dec 2010 08:52:26 -0800
Dmitry Torokhov <[email protected]> wrote:
> When hypervisor decides to decrease target balloon size while the
> balloon driver tries to lock pages hypervisor may respond with
> VMW_BALLOON_PPN_NOTNEEDED. Use this data and immediately stop reserving
> pages and wait for the next update cycle to fetch new target instead
> of continuing trying to lock pages until size of refused list grows
> above VMW_BALLOON_MAX_REFUSED (16) pages.
OK, but ... why? What is the externally visible effect of this change?
On Tuesday, December 07, 2010 05:10:23 pm Andrew Morton wrote:
> On Fri, 3 Dec 2010 08:52:26 -0800
>
> Dmitry Torokhov <[email protected]> wrote:
> > When hypervisor decides to decrease target balloon size while the
> > balloon driver tries to lock pages hypervisor may respond with
> > VMW_BALLOON_PPN_NOTNEEDED. Use this data and immediately stop reserving
> > pages and wait for the next update cycle to fetch new target instead
> > of continuing trying to lock pages until size of refused list grows
> > above VMW_BALLOON_MAX_REFUSED (16) pages.
>
> OK, but ... why? What is the externally visible effect of this change?
As a result the driver stops bothering hypervisor with its attempts to
lock more pages that are not needed anymore. Most likely next order from
hypervisor will be to reduce ballon size anyway.
It is a small optimization.
Thanks,
Dmitry