[PATCH 0/5] Drivers: hv: Miscellaneous fixes

[PATCH 0/5] Drivers: hv: Miscellaneous fixes

[kys]

From: K. Y. Srinivasan <kys@microsoft.com>

Miscellaneous fixes.

Alex Ng (5):
  Tools: hv: vss: Skip freezing filesystems backed by loop
  Drivers: hv: balloon: Correctly update onlined page count
  Drivers: hv: balloon: Show the max dynamic memory assigned
  Drivers: hv: balloon: Initialize last_post_time on startup
  Drivers: hv: kvp: Use MAX_ADAPTER_ID_SIZE for translating adapter id

 drivers/hv/hv_balloon.c  |   12 ++++++------
 drivers/hv/hv_kvp.c      |    2 +-
 tools/hv/hv_vss_daemon.c |    7 +++++++
 3 files changed, 14 insertions(+), 7 deletions(-)

[PATCH 1/5] Tools: hv: vss: Skip freezing filesystems backed by loop

[kys]

From: Alex Ng <alexng@messages.microsoft.com>

Since a loop device is backed by a file, a backup will already result in
its parent filesystem being frozen. It's sufficient to just freeze the
parent filesystem, so we can skip the loop device.

This avoids a situation where a loop device and its parent filesystem are
both frozen and then thawed out of order. For example, if the loop device
is enumerated first, we would thaw it while its parent filesystem is still
frozen. The thaw operation fails and the loop device remains frozen.

Signed-off-by: Alex Ng <alexng@messages.microsoft.com>
Signed-off-by: Vyronas Tsingaras <vyronas@vtsingaras.me>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
 tools/hv/hv_vss_daemon.c |    7 +++++++
 1 files changed, 7 insertions(+), 0 deletions(-)

diff --git a/tools/hv/hv_vss_daemon.c b/tools/hv/hv_vss_daemon.c
index 7ba5419..b2b4ebf 100644
--- a/tools/hv/hv_vss_daemon.c
+++ b/tools/hv/hv_vss_daemon.c
@@ -21,6 +21,7 @@
 #include <sys/types.h>
 #include <sys/poll.h>
 #include <sys/ioctl.h>
+#include <sys/stat.h>
 #include <fcntl.h>
 #include <stdio.h>
 #include <mntent.h>
@@ -30,6 +31,7 @@
 #include <ctype.h>
 #include <errno.h>
 #include <linux/fs.h>
+#include <linux/major.h>
 #include <linux/hyperv.h>
 #include <syslog.h>
 #include <getopt.h>
@@ -70,6 +72,7 @@ static int vss_operate(int operation)
 	char match[] = "/dev/";
 	FILE *mounts;
 	struct mntent *ent;
+	struct stat sb;
 	char errdir[1024] = {0};
 	unsigned int cmd;
 	int error = 0, root_seen = 0, save_errno = 0;
@@ -92,6 +95,10 @@ static int vss_operate(int operation)
 	while ((ent = getmntent(mounts))) {
 		if (strncmp(ent->mnt_fsname, match, strlen(match)))
 			continue;
+		if (stat(ent->mnt_fsname, &sb) == -1)
+			continue;
+		if (S_ISBLK(sb.st_mode) && major(sb.st_rdev) == LOOP_MAJOR)
+			continue;
 		if (hasmntopt(ent, MNTOPT_RO) != NULL)
 			continue;
 		if (strcmp(ent->mnt_type, "vfat") == 0)
-- 
1.7.1

[PATCH 2/5] Drivers: hv: balloon: Correctly update onlined page count

[kys]

From: Alex Ng <alexng@messages.microsoft.com>

Previously, num_pages_onlined was updated using value from memory online
notifier. This is incorrect because they assume that all hot-added pages
are online, even though we only online the amount that's backed by the
host. We should update num_pages_onlined only when the balloon driver
marks a page as online.

Signed-off-by: Alex Ng <alexng@messages.microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
 drivers/hv/hv_balloon.c |    7 +++----
 1 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/drivers/hv/hv_balloon.c b/drivers/hv/hv_balloon.c
index f5728de..0a5c318 100644
--- a/drivers/hv/hv_balloon.c
+++ b/drivers/hv/hv_balloon.c
@@ -584,10 +584,6 @@ static int hv_memory_notifier(struct notifier_block *nb, unsigned long val,
 
 	switch (val) {
 	case MEM_ONLINE:
-		spin_lock_irqsave(&dm_device.ha_lock, flags);
-		dm_device.num_pages_onlined += mem->nr_pages;
-		spin_unlock_irqrestore(&dm_device.ha_lock, flags);
-		/* Fall through */
 	case MEM_CANCEL_ONLINE:
 		if (dm_device.ha_waiting) {
 			dm_device.ha_waiting = false;
@@ -644,6 +640,9 @@ static void hv_page_online_one(struct hv_hotadd_state *has, struct page *pg)
 	__online_page_set_limits(pg);
 	__online_page_increment_counters(pg);
 	__online_page_free(pg);
+
+	WARN_ON_ONCE(!spin_is_locked(&dm_device.ha_lock));
+	dm_device.num_pages_onlined++;
 }
 
 static void hv_bring_pgs_online(struct hv_hotadd_state *has,
-- 
1.7.1

[PATCH 3/5] Drivers: hv: balloon: Show the max dynamic memory assigned

[kys]

From: Alex Ng <alexng@messages.microsoft.com>

Previously we were only showing max number of pages. We should make it
more clear that this value is the max amount of dynamic memory that the
Hyper-V host is willing to assign to this guest.

Signed-off-by: Alex Ng <alexng@messages.microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
 drivers/hv/hv_balloon.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/hv/hv_balloon.c b/drivers/hv/hv_balloon.c
index 0a5c318..7cec482 100644
--- a/drivers/hv/hv_balloon.c
+++ b/drivers/hv/hv_balloon.c
@@ -1035,8 +1035,8 @@ static void process_info(struct hv_dynmem_device *dm, struct dm_info_msg *msg)
 		if (info_hdr->data_size == sizeof(__u64)) {
 			__u64 *max_page_count = (__u64 *)&info_hdr[1];
 
-			pr_info("INFO_TYPE_MAX_PAGE_CNT = %llu\n",
-				*max_page_count);
+			pr_info("Max. dynamic memory size: %llu MB\n",
+				(*max_page_count) >> (20 - PAGE_SHIFT));
 		}
 
 		break;
-- 
1.7.1

[PATCH 4/5] Drivers: hv: balloon: Initialize last_post_time on startup

[kys]

From: Alex Ng <alexng@messages.microsoft.com>

When left uninitialized, this sometimes fails the following check in
post_status():

	if (!time_after(now, (last_post_time + HZ))) {
		return;
        }

This causes unnecessary delays in reporting memory pressure to host after
booting up.

Signed-off-by: Alex Ng <alexng@messages.microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
 drivers/hv/hv_balloon.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/hv/hv_balloon.c b/drivers/hv/hv_balloon.c
index 7cec482..db0e665 100644
--- a/drivers/hv/hv_balloon.c
+++ b/drivers/hv/hv_balloon.c
@@ -1655,6 +1655,7 @@ static int balloon_probe(struct hv_device *dev,
 	}
 
 	dm_device.state = DM_INITIALIZED;
+	last_post_time = jiffies;
 
 	return 0;
 
-- 
1.7.1

[PATCH 5/5] Drivers: hv: kvp: Use MAX_ADAPTER_ID_SIZE for translating adapter id

[kys]

From: Alex Ng <alexng@messages.microsoft.com>

There's a bug which passes the output buffer size as MAX_IP_ADDR_SIZE,
when converting the adapter_id field to UTF16. This is much larger than
the actual size (MAX_ADAPTER_ID_SIZE). Fix this by passing the proper
size.

Fortunately, the translation is limited by the length of the input. This
explains why we haven't seen output buffer overflow conditions.

Signed-off-by: Alex Ng <alexng@messages.microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
 drivers/hv/hv_kvp.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/hv/hv_kvp.c b/drivers/hv/hv_kvp.c
index 9a90b91..5eed1e7 100644
--- a/drivers/hv/hv_kvp.c
+++ b/drivers/hv/hv_kvp.c
@@ -304,7 +304,7 @@ static int process_ob_ipinfo(void *in_msg, void *out_msg, int op)
 				strlen((char *)in->body.kvp_ip_val.adapter_id),
 				UTF16_HOST_ENDIAN,
 				(wchar_t *)out->kvp_ip_val.adapter_id,
-				MAX_IP_ADDR_SIZE);
+				MAX_ADAPTER_ID_SIZE);
 		if (len < 0)
 			return len;
 
-- 
1.7.1