2019-08-20 01:55:16

by Dexuan Cui

[permalink] [raw]
Subject: [PATCH v3 08/12] Drivers: hv: vmbus: Ignore the offers when resuming from hibernation

When the VM resumes, the host re-sends the offers. We should not add the
offers to the global vmbus_connection.chn_list again.

This patch assumes the RELIDs of the channels don't change across
hibernation. Actually this is not always true, especially in the case of
NIC SR-IOV the VF vmbus device's RELID sometimes can change. A later patch
will address this issue by mapping the new offers to the old channels and
fixing up the old channels, if necessary.

Signed-off-by: Dexuan Cui <[email protected]>
---
drivers/hv/channel_mgmt.c | 29 ++++++++++++++++++++++++++++-
drivers/hv/connection.c | 27 +++++++++++++++++++++++++++
drivers/hv/hyperv_vmbus.h | 3 +++
3 files changed, 58 insertions(+), 1 deletion(-)

diff --git a/drivers/hv/channel_mgmt.c b/drivers/hv/channel_mgmt.c
index addcef5..f7a1184 100644
--- a/drivers/hv/channel_mgmt.c
+++ b/drivers/hv/channel_mgmt.c
@@ -854,12 +854,39 @@ void vmbus_initiate_unload(bool crash)
static void vmbus_onoffer(struct vmbus_channel_message_header *hdr)
{
struct vmbus_channel_offer_channel *offer;
- struct vmbus_channel *newchannel;
+ struct vmbus_channel *oldchannel, *newchannel;
+ size_t offer_sz;

offer = (struct vmbus_channel_offer_channel *)hdr;

trace_vmbus_onoffer(offer);

+ mutex_lock(&vmbus_connection.channel_mutex);
+ oldchannel = find_primary_channel_by_offer(offer);
+ mutex_unlock(&vmbus_connection.channel_mutex);
+
+ if (oldchannel != NULL) {
+ atomic_dec(&vmbus_connection.offer_in_progress);
+
+ /*
+ * We're resuming from hibernation: we expect the host to send
+ * exactly the same offers that we had before the hibernation.
+ */
+ offer_sz = sizeof(*offer);
+ if (memcmp(offer, &oldchannel->offermsg, offer_sz) == 0)
+ return;
+
+ pr_debug("Mismatched offer from the host (relid=%d)\n",
+ offer->child_relid);
+
+ print_hex_dump_debug("Old vmbus offer: ", DUMP_PREFIX_OFFSET,
+ 16, 4, &oldchannel->offermsg, offer_sz,
+ false);
+ print_hex_dump_debug("New vmbus offer: ", DUMP_PREFIX_OFFSET,
+ 16, 4, offer, offer_sz, false);
+ return;
+ }
+
/* Allocate the channel object and save this offer. */
newchannel = alloc_channel();
if (!newchannel) {
diff --git a/drivers/hv/connection.c b/drivers/hv/connection.c
index 09829e1..6c7a983 100644
--- a/drivers/hv/connection.c
+++ b/drivers/hv/connection.c
@@ -337,6 +337,33 @@ struct vmbus_channel *relid2channel(u32 relid)
}

/*
+ * find_primary_channel_by_offer - Get the channel object given the new offer.
+ * This is only used in the resume path of hibernation.
+ */
+struct vmbus_channel *
+find_primary_channel_by_offer(const struct vmbus_channel_offer_channel *offer)
+{
+ struct vmbus_channel *channel;
+ const guid_t *inst1, *inst2;
+
+ WARN_ON(!mutex_is_locked(&vmbus_connection.channel_mutex));
+
+ /* Ignore sub-channel offers. */
+ if (offer->offer.sub_channel_index != 0)
+ return NULL;
+
+ list_for_each_entry(channel, &vmbus_connection.chn_list, listentry) {
+ inst1 = &channel->offermsg.offer.if_instance;
+ inst2 = &offer->offer.if_instance;
+
+ if (guid_equal(inst1, inst2))
+ return channel;
+ }
+
+ return NULL;
+}
+
+/*
* vmbus_on_event - Process a channel event notification
*
* For batched channels (default) optimize host to guest signaling
diff --git a/drivers/hv/hyperv_vmbus.h b/drivers/hv/hyperv_vmbus.h
index 9f7fb6d..c42b46d 100644
--- a/drivers/hv/hyperv_vmbus.h
+++ b/drivers/hv/hyperv_vmbus.h
@@ -310,6 +310,9 @@ int vmbus_add_channel_kobj(struct hv_device *device_obj,

struct vmbus_channel *relid2channel(u32 relid);

+struct vmbus_channel *
+find_primary_channel_by_offer(const struct vmbus_channel_offer_channel *offer);
+
void vmbus_free_channels(void);

/* Connection interface */
--
1.8.3.1


2019-08-23 23:37:51

by Michael Kelley (LINUX)

[permalink] [raw]
Subject: RE: [PATCH v3 08/12] Drivers: hv: vmbus: Ignore the offers when resuming from hibernation

From: Dexuan Cui <[email protected]> Sent: Monday, August 19, 2019 6:52 PM
>
> When the VM resumes, the host re-sends the offers. We should not add the
> offers to the global vmbus_connection.chn_list again.
>
> This patch assumes the RELIDs of the channels don't change across
> hibernation. Actually this is not always true, especially in the case of
> NIC SR-IOV the VF vmbus device's RELID sometimes can change. A later patch
> will address this issue by mapping the new offers to the old channels and
> fixing up the old channels, if necessary.
>
> Signed-off-by: Dexuan Cui <[email protected]>
> ---
> drivers/hv/channel_mgmt.c | 29 ++++++++++++++++++++++++++++-
> drivers/hv/connection.c | 27 +++++++++++++++++++++++++++
> drivers/hv/hyperv_vmbus.h | 3 +++
> 3 files changed, 58 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/hv/channel_mgmt.c b/drivers/hv/channel_mgmt.c
> index addcef5..f7a1184 100644
> --- a/drivers/hv/channel_mgmt.c
> +++ b/drivers/hv/channel_mgmt.c
> @@ -854,12 +854,39 @@ void vmbus_initiate_unload(bool crash)
> static void vmbus_onoffer(struct vmbus_channel_message_header *hdr)
> {
> struct vmbus_channel_offer_channel *offer;
> - struct vmbus_channel *newchannel;
> + struct vmbus_channel *oldchannel, *newchannel;
> + size_t offer_sz;
>
> offer = (struct vmbus_channel_offer_channel *)hdr;
>
> trace_vmbus_onoffer(offer);
>
> + mutex_lock(&vmbus_connection.channel_mutex);
> + oldchannel = find_primary_channel_by_offer(offer);
> + mutex_unlock(&vmbus_connection.channel_mutex);
> +
> + if (oldchannel != NULL) {
> + atomic_dec(&vmbus_connection.offer_in_progress);
> +
> + /*
> + * We're resuming from hibernation: we expect the host to send
> + * exactly the same offers that we had before the hibernation.
> + */
> + offer_sz = sizeof(*offer);
> + if (memcmp(offer, &oldchannel->offermsg, offer_sz) == 0)
> + return;
> +
> + pr_debug("Mismatched offer from the host (relid=%d)\n",
> + offer->child_relid);
> +
> + print_hex_dump_debug("Old vmbus offer: ", DUMP_PREFIX_OFFSET,
> + 16, 4, &oldchannel->offermsg, offer_sz,
> + false);
> + print_hex_dump_debug("New vmbus offer: ", DUMP_PREFIX_OFFSET,
> + 16, 4, offer, offer_sz, false);
> + return;
> + }
> +
> /* Allocate the channel object and save this offer. */
> newchannel = alloc_channel();
> if (!newchannel) {
> diff --git a/drivers/hv/connection.c b/drivers/hv/connection.c
> index 09829e1..6c7a983 100644
> --- a/drivers/hv/connection.c
> +++ b/drivers/hv/connection.c
> @@ -337,6 +337,33 @@ struct vmbus_channel *relid2channel(u32 relid)
> }
>
> /*
> + * find_primary_channel_by_offer - Get the channel object given the new offer.
> + * This is only used in the resume path of hibernation.
> + */
> +struct vmbus_channel *
> +find_primary_channel_by_offer(const struct vmbus_channel_offer_channel *offer)
> +{
> + struct vmbus_channel *channel;
> + const guid_t *inst1, *inst2;
> +
> + WARN_ON(!mutex_is_locked(&vmbus_connection.channel_mutex));
> +
> + /* Ignore sub-channel offers. */
> + if (offer->offer.sub_channel_index != 0)
> + return NULL;
> +
> + list_for_each_entry(channel, &vmbus_connection.chn_list, listentry) {
> + inst1 = &channel->offermsg.offer.if_instance;
> + inst2 = &offer->offer.if_instance;
> +
> + if (guid_equal(inst1, inst2))
> + return channel;
> + }
> +
> + return NULL;
> +}

Any particular reason this new function is in connection.c instead of
putting it in channel_mgmt.c where it is called?

> +
> +/*
> * vmbus_on_event - Process a channel event notification
> *
> * For batched channels (default) optimize host to guest signaling
> diff --git a/drivers/hv/hyperv_vmbus.h b/drivers/hv/hyperv_vmbus.h
> index 9f7fb6d..c42b46d 100644
> --- a/drivers/hv/hyperv_vmbus.h
> +++ b/drivers/hv/hyperv_vmbus.h
> @@ -310,6 +310,9 @@ int vmbus_add_channel_kobj(struct hv_device *device_obj,
>
> struct vmbus_channel *relid2channel(u32 relid);
>
> +struct vmbus_channel *
> +find_primary_channel_by_offer(const struct vmbus_channel_offer_channel *offer);
> +
> void vmbus_free_channels(void);
>
> /* Connection interface */
> --
> 1.8.3.1

2019-08-31 00:25:43

by Dexuan Cui

[permalink] [raw]
Subject: RE: [PATCH v3 08/12] Drivers: hv: vmbus: Ignore the offers when resuming from hibernation

> From: Michael Kelley <[email protected]>
> Sent: Friday, August 23, 2019 12:57 PM
>
> From: Dexuan Cui <[email protected]> Sent: August 19, 2019 6:52 PM
> >
> > diff --git a/drivers/hv/connection.c b/drivers/hv/connection.c
> > @@ -337,6 +337,33 @@ struct vmbus_channel *relid2channel(u32 relid)
> > }
> >
> > /*
> > + * find_primary_channel_by_offer - Get the channel object given the new
> offer.
> > + * This is only used in the resume path of hibernation.
> > + */
> > +struct vmbus_channel *
> > +find_primary_channel_by_offer(const struct vmbus_channel_offer_channel
> *offer)
> > +{
> > + struct vmbus_channel *channel;
> > + const guid_t *inst1, *inst2;
> > +
> > + WARN_ON(!mutex_is_locked(&vmbus_connection.channel_mutex));
> > +
> > + /* Ignore sub-channel offers. */
> > + if (offer->offer.sub_channel_index != 0)
> > + return NULL;
> > +
> > + list_for_each_entry(channel, &vmbus_connection.chn_list, listentry) {
> > + inst1 = &channel->offermsg.offer.if_instance;
> > + inst2 = &offer->offer.if_instance;
> > +
> > + if (guid_equal(inst1, inst2))
> > + return channel;
> > + }
> > +
> > + return NULL;
> > +}
>
> Any particular reason this new function is in connection.c instead of
> putting it in channel_mgmt.c where it is called?

There is a similar function relid2channel(), which is in connection.c.

Since the new function is only used in channel_mgmt.c. I'll move it to
channel_mgmt.c in v4.

Thanks,
-- Dexuan