From: Dexuan Cui <decui@microsoft.com>
To: gregkh@linuxfoundation.org, linux-kernel@vger.kernel.org,
driverdev-devel@linuxdriverproject.org, olaf@aepfle.de,
apw@canonical.com, jasowang@redhat.com, kys@microsoft.com,
vkuznets@redhat.com
Cc: haiyangz@microsoft.com, rolf.neugebauer@docker.com,
dave.scott@docker.com, ian.campbell@docker.com
Subject: [PATCH v3] Drivers: hv: vmbus: fix the race when querying & updating the percpu list
Date: Sat, 21 May 2016 23:02:46 -0700 [thread overview]
Message-ID: <1463896966-4745-1-git-send-email-decui@microsoft.com> (raw)
There is a rare race when we remove an entry from the global list
hv_context.percpu_list[cpu] in hv_process_channel_removal() ->
percpu_channel_deq() -> list_del(): at this time, if vmbus_on_event() ->
process_chn_event() -> pcpu_relid2channel() is trying to query the list,
we can get the kernel fault.
Similarly, we also have the issue in the code path: vmbus_process_offer() ->
percpu_channel_enq().
We can resolve the issue by disabling the tasklet when updating the list.
The patch also moves vmbus_release_relid() to a later place where
the channel has been removed from the per-cpu and the global lists.
Reported-by: Rolf Neugebauer <rolf.neugebauer@docker.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Dexuan Cui <decui@microsoft.com>
---
v2: added tasklet_schedule() after tasklet_enable(). Thanks, Vitaly!
v3: I shouldn't have moved percpu_channel_deq()
from hv_process_channel_removal() to vmbus_close_internal(): the
channel couldn't be removed from the per-cpu list, if the channel state
was not CHANNEL_OPENED_STATE. v3 fixed this.
v3 also added 2 wrapper functions for the disable/enable stuff.
v3 also moved vmbus_release_relid() to a later safe place.
You can also get v3 by
https://github.com/dcui/linux/commit/2f314fb9352020f70b094bf31539afd3a18a5545
drivers/hv/channel.c | 6 ++----
drivers/hv/channel_mgmt.c | 32 ++++++++++++++++++++++++++++----
include/linux/hyperv.h | 3 +++
3 files changed, 33 insertions(+), 8 deletions(-)
diff --git a/drivers/hv/channel.c b/drivers/hv/channel.c
index 56dd261..ef8e9b5 100644
--- a/drivers/hv/channel.c
+++ b/drivers/hv/channel.c
@@ -512,7 +512,6 @@ static void reset_channel_cb(void *arg)
static int vmbus_close_internal(struct vmbus_channel *channel)
{
struct vmbus_channel_close_channel *msg;
- struct tasklet_struct *tasklet;
int ret;
/*
@@ -524,8 +523,7 @@ static int vmbus_close_internal(struct vmbus_channel *channel)
* To resolve the race, we can serialize them by disabling the
* tasklet when the latter is running here.
*/
- tasklet = hv_context.event_dpc[channel->target_cpu];
- tasklet_disable(tasklet);
+ hv_event_tasklet_disable(channel);
/*
* In case a device driver's probe() fails (e.g.,
@@ -591,7 +589,7 @@ static int vmbus_close_internal(struct vmbus_channel *channel)
get_order(channel->ringbuffer_pagecount * PAGE_SIZE));
out:
- tasklet_enable(tasklet);
+ hv_event_tasklet_enable(channel);
return ret;
}
diff --git a/drivers/hv/channel_mgmt.c b/drivers/hv/channel_mgmt.c
index 38b682ba..3bcf141 100644
--- a/drivers/hv/channel_mgmt.c
+++ b/drivers/hv/channel_mgmt.c
@@ -21,6 +21,7 @@
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/kernel.h>
+#include <linux/interrupt.h>
#include <linux/sched.h>
#include <linux/wait.h>
#include <linux/mm.h>
@@ -303,16 +304,32 @@ static void vmbus_release_relid(u32 relid)
vmbus_post_msg(&msg, sizeof(struct vmbus_channel_relid_released));
}
+void hv_event_tasklet_disable(struct vmbus_channel *channel)
+{
+ struct tasklet_struct *tasklet;
+ tasklet = hv_context.event_dpc[channel->target_cpu];
+ tasklet_disable(tasklet);
+}
+
+void hv_event_tasklet_enable(struct vmbus_channel *channel)
+{
+ struct tasklet_struct *tasklet;
+ tasklet = hv_context.event_dpc[channel->target_cpu];
+ tasklet_enable(tasklet);
+
+ /* In case there is any pending event */
+ tasklet_schedule(tasklet);
+}
+
void hv_process_channel_removal(struct vmbus_channel *channel, u32 relid)
{
unsigned long flags;
struct vmbus_channel *primary_channel;
- vmbus_release_relid(relid);
-
BUG_ON(!channel->rescind);
BUG_ON(!mutex_is_locked(&vmbus_connection.channel_mutex));
+ hv_event_tasklet_disable(channel);
if (channel->target_cpu != get_cpu()) {
put_cpu();
smp_call_function_single(channel->target_cpu,
@@ -321,6 +338,7 @@ void hv_process_channel_removal(struct vmbus_channel *channel, u32 relid)
percpu_channel_deq(channel);
put_cpu();
}
+ hv_event_tasklet_enable(channel);
if (channel->primary_channel == NULL) {
list_del(&channel->listentry);
@@ -341,6 +359,8 @@ void hv_process_channel_removal(struct vmbus_channel *channel, u32 relid)
cpumask_clear_cpu(channel->target_cpu,
&primary_channel->alloced_cpus_in_node);
+ vmbus_release_relid(relid);
+
free_channel(channel);
}
@@ -409,6 +429,7 @@ static void vmbus_process_offer(struct vmbus_channel *newchannel)
init_vp_index(newchannel, dev_type);
+ hv_event_tasklet_disable(channel);
if (newchannel->target_cpu != get_cpu()) {
put_cpu();
smp_call_function_single(newchannel->target_cpu,
@@ -418,6 +439,7 @@ static void vmbus_process_offer(struct vmbus_channel *newchannel)
percpu_channel_enq(newchannel);
put_cpu();
}
+ hv_event_tasklet_enable(channel);
/*
* This state is used to indicate a successful open
@@ -463,12 +485,11 @@ static void vmbus_process_offer(struct vmbus_channel *newchannel)
return;
err_deq_chan:
- vmbus_release_relid(newchannel->offermsg.child_relid);
-
mutex_lock(&vmbus_connection.channel_mutex);
list_del(&newchannel->listentry);
mutex_unlock(&vmbus_connection.channel_mutex);
+ hv_event_tasklet_disable(channel);
if (newchannel->target_cpu != get_cpu()) {
put_cpu();
smp_call_function_single(newchannel->target_cpu,
@@ -477,6 +498,9 @@ err_deq_chan:
percpu_channel_deq(newchannel);
put_cpu();
}
+ hv_event_tasklet_enable(channel);
+
+ vmbus_release_relid(newchannel->offermsg.child_relid);
err_free_chan:
free_channel(newchannel);
diff --git a/include/linux/hyperv.h b/include/linux/hyperv.h
index 7be7237..8070cda 100644
--- a/include/linux/hyperv.h
+++ b/include/linux/hyperv.h
@@ -1328,6 +1328,9 @@ extern bool vmbus_prep_negotiate_resp(struct icmsg_hdr *,
struct icmsg_negotiate *, u8 *, int,
int);
+void hv_event_tasklet_disable(struct vmbus_channel *channel);
+void hv_event_tasklet_enable(struct vmbus_channel *channel);
+
void hv_process_channel_removal(struct vmbus_channel *channel, u32 relid);
/*
--
2.1.0
next reply other threads:[~2016-05-22 4:22 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-05-22 6:02 Dexuan Cui [this message]
2016-05-31 16:26 ` [PATCH v3] Drivers: hv: vmbus: fix the race when querying & updating the percpu list Vitaly Kuznetsov
2016-06-01 6:39 ` Dexuan Cui
2016-06-01 8:58 ` Vitaly Kuznetsov
2016-06-01 10:01 ` Dexuan Cui
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1463896966-4745-1-git-send-email-decui@microsoft.com \
--to=decui@microsoft.com \
--cc=apw@canonical.com \
--cc=dave.scott@docker.com \
--cc=driverdev-devel@linuxdriverproject.org \
--cc=gregkh@linuxfoundation.org \
--cc=haiyangz@microsoft.com \
--cc=ian.campbell@docker.com \
--cc=jasowang@redhat.com \
--cc=kys@microsoft.com \
--cc=linux-kernel@vger.kernel.org \
--cc=olaf@aepfle.de \
--cc=rolf.neugebauer@docker.com \
--cc=vkuznets@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).