linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Andrea Parri (Microsoft)" <parri.andrea@gmail.com>
To: linux-kernel@vger.kernel.org
Cc: kys@microsoft.com, haiyangz@microsoft.com,
	sthemmin@microsoft.com, wei.liu@kernel.org,
	linux-hyperv@vger.kernel.org, mikelley@microsoft.com,
	"Andrea Parri (Microsoft)" <parri.andrea@gmail.com>
Subject: [PATCH v3 3/3] Drivers: hv: vmbus: Check for pending channel interrupts before taking a CPU offline
Date: Fri, 16 Apr 2021 16:34:49 +0200	[thread overview]
Message-ID: <20210416143449.16185-4-parri.andrea@gmail.com> (raw)
In-Reply-To: <20210416143449.16185-1-parri.andrea@gmail.com>

Check that enough time has passed such that the modify channel message
has been processed before taking a CPU offline.

Signed-off-by: Andrea Parri (Microsoft) <parri.andrea@gmail.com>
---
 drivers/hv/hv.c | 56 ++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 53 insertions(+), 3 deletions(-)

diff --git a/drivers/hv/hv.c b/drivers/hv/hv.c
index 3e6ff83adff42..e0c522d143a37 100644
--- a/drivers/hv/hv.c
+++ b/drivers/hv/hv.c
@@ -15,6 +15,7 @@
 #include <linux/hyperv.h>
 #include <linux/random.h>
 #include <linux/clockchips.h>
+#include <linux/delay.h>
 #include <linux/interrupt.h>
 #include <clocksource/hyperv_timer.h>
 #include <asm/mshyperv.h>
@@ -292,12 +293,50 @@ void hv_synic_disable_regs(unsigned int cpu)
 		disable_percpu_irq(vmbus_irq);
 }
 
+#define HV_MAX_TRIES 3
+/*
+ * Scan the event flags page of 'this' CPU looking for any bit that is set.  If we find one
+ * bit set, then wait for a few milliseconds.  Repeat these steps for a maximum of 3 times.
+ * Return 'true', if there is still any set bit after this operation; 'false', otherwise.
+ *
+ * If a bit is set, that means there is a pending channel interrupt.  The expectation is
+ * that the normal interrupt handling mechanism will find and process the channel interrupt
+ * "very soon", and in the process clear the bit.
+ */
+static bool hv_synic_event_pending(void)
+{
+	struct hv_per_cpu_context *hv_cpu = this_cpu_ptr(hv_context.cpu_context);
+	union hv_synic_event_flags *event =
+		(union hv_synic_event_flags *)hv_cpu->synic_event_page + VMBUS_MESSAGE_SINT;
+	unsigned long *recv_int_page = event->flags; /* assumes VMBus version >= VERSION_WIN8 */
+	bool pending;
+	u32 relid;
+	int tries = 0;
+
+retry:
+	pending = false;
+	for_each_set_bit(relid, recv_int_page, HV_EVENT_FLAGS_COUNT) {
+		/* Special case - VMBus channel protocol messages */
+		if (relid == 0)
+			continue;
+		pending = true;
+		break;
+	}
+	if (pending && tries++ < HV_MAX_TRIES) {
+		usleep_range(10000, 20000);
+		goto retry;
+	}
+	return pending;
+}
 
 int hv_synic_cleanup(unsigned int cpu)
 {
 	struct vmbus_channel *channel, *sc;
 	bool channel_found = false;
 
+	if (vmbus_connection.conn_state != CONNECTED)
+		goto always_cleanup;
+
 	/*
 	 * Hyper-V does not provide a way to change the connect CPU once
 	 * it is set; we must prevent the connect CPU from going offline
@@ -305,8 +344,7 @@ int hv_synic_cleanup(unsigned int cpu)
 	 * path where the vmbus is already disconnected, the CPU must be
 	 * allowed to shut down.
 	 */
-	if (cpu == VMBUS_CONNECT_CPU &&
-	    vmbus_connection.conn_state == CONNECTED)
+	if (cpu == VMBUS_CONNECT_CPU)
 		return -EBUSY;
 
 	/*
@@ -333,9 +371,21 @@ int hv_synic_cleanup(unsigned int cpu)
 	}
 	mutex_unlock(&vmbus_connection.channel_mutex);
 
-	if (channel_found && vmbus_connection.conn_state == CONNECTED)
+	if (channel_found)
+		return -EBUSY;
+
+	/*
+	 * channel_found == false means that any channels that were previously
+	 * assigned to the CPU have been reassigned elsewhere with a call of
+	 * vmbus_send_modifychannel().  Scan the event flags page looking for
+	 * bits that are set and waiting with a timeout for vmbus_chan_sched()
+	 * to process such bits.  If bits are still set after this operation
+	 * and VMBus is connected, fail the CPU offlining operation.
+	 */
+	if (vmbus_proto_version >= VERSION_WIN10_V4_1 && hv_synic_event_pending())
 		return -EBUSY;
 
+always_cleanup:
 	hv_stimer_legacy_cleanup(cpu);
 
 	hv_synic_disable_regs(cpu);
-- 
2.25.1


  parent reply	other threads:[~2021-04-16 14:35 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-04-16 14:34 [PATCH v3 0/3] Drivers: hv: vmbus: Introduce CHANNELMSG_MODIFYCHANNEL_RESPONSE Andrea Parri (Microsoft)
2021-04-16 14:34 ` [PATCH v3 1/3] Drivers: hv: vmbus: Introduce and negotiate VMBus protocol version 5.3 Andrea Parri (Microsoft)
2021-04-16 15:07   ` Michael Kelley
2021-04-16 14:34 ` [PATCH v3 2/3] Drivers: hv: vmbus: Drivers: hv: vmbus: Introduce CHANNELMSG_MODIFYCHANNEL_RESPONSE Andrea Parri (Microsoft)
2021-04-16 14:34 ` Andrea Parri (Microsoft) [this message]
2021-04-16 15:11   ` [PATCH v3 3/3] Drivers: hv: vmbus: Check for pending channel interrupts before taking a CPU offline Michael Kelley
2021-04-18 13:04 ` [PATCH v3 0/3] Drivers: hv: vmbus: Introduce CHANNELMSG_MODIFYCHANNEL_RESPONSE Wei Liu

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=20210416143449.16185-4-parri.andrea@gmail.com \
    --to=parri.andrea@gmail.com \
    --cc=haiyangz@microsoft.com \
    --cc=kys@microsoft.com \
    --cc=linux-hyperv@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mikelley@microsoft.com \
    --cc=sthemmin@microsoft.com \
    --cc=wei.liu@kernel.org \
    /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).