From: Jiang Liu <liuj97@gmail.com>
To: "Rafael J . Wysocki" <rjw@sisk.pl>,
Yinghai Lu <yinghai@kernel.org>, Tony Luck <tony.luck@intel.com>,
Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>,
Wen Congyang <wency@cn.fujitsu.com>,
Tang Chen <tangchen@cn.fujitsu.com>,
Taku Izumi <izumi.taku@jp.fujitsu.com>,
Bjorn Helgaas <bhelgaas@google.com>
Cc: Jiang Liu <jiang.liu@huawei.com>,
Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>,
Huang Ying <ying.huang@intel.com>,
Bob Moore <robert.moore@intel.com>, Len Brown <lenb@kernel.org>,
"Srivatsa S . Bhat" <srivatsa.bhat@linux.vnet.ibm.com>,
Yijing Wang <wangyijing@huawei.com>,
Hanjun Guo <guohanjun@huawei.com>, Jiang Liu <liuj97@gmail.com>,
linux-kernel@vger.kernel.org, linux-acpi@vger.kernel.org,
linux-pci@vger.kernel.org, linux-mm@kvack.org
Subject: [ACPIHP PATCH part2 08/13] ACPIHP: provide interface to cancel inprogress hotplug operations
Date: Sun, 4 Nov 2012 20:50:10 +0800 [thread overview]
Message-ID: <1352033415-5606-9-git-send-email-jiang.liu@huawei.com> (raw)
In-Reply-To: <1352033415-5606-1-git-send-email-jiang.liu@huawei.com>
Some hotplug operations, such as hot-removal of memory device, may take
very long or even infinite time. One possible solution is to time out
and retry, but it's sub-optimal.
This patch implements interfaces to cancel inprogress ACPI system device
hotplug operations, so user could cancel a long-standing hotplug request
on demand.
Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
Signed-off-by: Hanjun Guo <guohanjun@huawei.com>
---
drivers/acpi/hotplug/Makefile | 1 +
drivers/acpi/hotplug/acpihp_drv.h | 19 ++++
drivers/acpi/hotplug/cancel.c | 174 +++++++++++++++++++++++++++++++++++++
3 files changed, 194 insertions(+)
create mode 100644 drivers/acpi/hotplug/cancel.c
diff --git a/drivers/acpi/hotplug/Makefile b/drivers/acpi/hotplug/Makefile
index bfb677f..f72f2c3 100644
--- a/drivers/acpi/hotplug/Makefile
+++ b/drivers/acpi/hotplug/Makefile
@@ -13,3 +13,4 @@ acpihp_slot-$(CONFIG_ACPI_HOTPLUG_SLOT_FAKE) += slot_fake.o
obj-$(CONFIG_ACPI_HOTPLUG_DRIVER) += acpihp_drv.o
acpihp_drv-y = drv_main.o
acpihp_drv-y += dependency.o
+acpihp_drv-y += cancel.o
diff --git a/drivers/acpi/hotplug/acpihp_drv.h b/drivers/acpi/hotplug/acpihp_drv.h
index 32ea054..dd8ea92 100644
--- a/drivers/acpi/hotplug/acpihp_drv.h
+++ b/drivers/acpi/hotplug/acpihp_drv.h
@@ -38,8 +38,20 @@ enum acpihp_drv_cmd {
ACPIHP_DRV_CMD_MAX
};
+enum acpihp_drv_cancel_state {
+ ACPIHP_DRV_CANCEL_INIT = 0,
+ ACPIHP_DRV_CANCEL_STARTED,
+ ACPIHP_DRV_CANCEL_OK,
+ ACPIHP_DRV_CANCEL_FAILED,
+ ACPIHP_DRV_CANCEL_MISSED,
+ ACPIHP_DRV_CANCEL_FINISHED
+};
+
struct acpihp_slot_drv {
struct mutex op_mutex;
+ atomic_t cancel_state;
+ atomic_t cancel_users;
+ struct acpihp_cancel_context cancel_ctx;
};
struct acpihp_slot_dependency {
@@ -62,4 +74,11 @@ int acpihp_drv_filter_dependency_list(struct list_head *old_head,
int acpihp_drv_generate_dependency_list(struct acpihp_slot *slot,
struct list_head *slot_list, enum acpihp_drv_cmd cmd);
+void acpihp_drv_cancel_init(struct list_head *list);
+void acpihp_drv_cancel_notify(struct acpihp_slot *slot,
+ enum acpihp_drv_cancel_state state);
+void acpihp_drv_cancel_fini(struct list_head *list);
+int acpihp_drv_cancel_start(struct list_head *list);
+int acpihp_drv_cancel_wait(struct list_head *list);
+
#endif /* __ACPIHP_DRV_H__ */
diff --git a/drivers/acpi/hotplug/cancel.c b/drivers/acpi/hotplug/cancel.c
new file mode 100644
index 0000000..c515c28
--- /dev/null
+++ b/drivers/acpi/hotplug/cancel.c
@@ -0,0 +1,174 @@
+/*
+ * Copyright (C) 2012 Huawei Tech. Co., Ltd.
+ * Copyright (C) 2012 Jiang Liu <jiang.liu@huawei.com>
+ * Copyright (C) 2012 Hanjun Guo <guohanjun@huawei.com>
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ */
+
+#include <linux/wait.h>
+#include <acpi/acpi_hotplug.h>
+#include "acpihp_drv.h"
+
+/*
+ * Implement interfaces to cancel inprogress hotplug operations.
+ * Currently only CONFIGURE and RELEASE operation stages support cancellation.
+ * Caller must serialize calls to following functions by holding the
+ * state_machine_mutex lock:
+ * acpihp_drv_cancel_init()
+ * acpihp_drv_cancel_start()
+ * acpihp_drv_cancel_fini()
+ */
+static DECLARE_WAIT_QUEUE_HEAD(acpihp_drv_cancel_queue);
+
+static int acpihp_drv_check_cancel(struct acpihp_cancel_context *ctx)
+{
+ struct acpihp_slot_drv *drv_data;
+
+ BUG_ON(ctx == NULL);
+ drv_data = container_of(ctx, struct acpihp_slot_drv, cancel_ctx);
+
+ return atomic_read(&drv_data->cancel_state) != ACPIHP_DRV_CANCEL_INIT;
+}
+
+void acpihp_drv_cancel_init(struct list_head *list)
+{
+ struct acpihp_slot_drv *drv_data;
+ struct acpihp_slot_dependency *dep;
+
+ list_for_each_entry(dep, list, node) {
+ acpihp_drv_get_data(dep->slot, &drv_data);
+ drv_data->cancel_ctx.check_cancel = acpihp_drv_check_cancel;
+ atomic_set(&drv_data->cancel_state, ACPIHP_DRV_CANCEL_INIT);
+ atomic_set(&drv_data->cancel_users, 0);
+ }
+}
+
+/*
+ * Start cancellation on a list of hotplug slots.
+ *
+ * Caller must provide mechanism to avoid currently running
+ * acpihp_drv_cancel_start() and acpihp_drv_cancel_fini()
+ * on the same list.
+ */
+int acpihp_drv_cancel_start(struct list_head *list)
+{
+ struct acpihp_slot_drv *drv_data;
+ struct acpihp_slot_dependency *dep;
+
+ if (list_empty(list)) {
+ ACPIHP_DEBUG("dependency list is empty.\n");
+ return 0;
+ }
+
+ /* Start cancellation on all slots. */
+ list_for_each_entry(dep, list, node) {
+ acpihp_drv_get_data(dep->slot, &drv_data);
+ atomic_inc(&drv_data->cancel_users);
+ atomic_cmpxchg(&drv_data->cancel_state,
+ ACPIHP_DRV_CANCEL_INIT,
+ ACPIHP_DRV_CANCEL_STARTED);
+ }
+
+ return 0;
+}
+
+/* Notify that the slot reaches a stable state */
+void acpihp_drv_cancel_notify(struct acpihp_slot *slot,
+ enum acpihp_drv_cancel_state state)
+{
+ int old;
+ struct acpihp_slot_drv *drv_data;
+
+ acpihp_drv_get_data(slot, &drv_data);
+ old = atomic_cmpxchg(&drv_data->cancel_state, ACPIHP_DRV_CANCEL_INIT,
+ ACPIHP_DRV_CANCEL_FINISHED);
+ if (old != ACPIHP_DRV_CANCEL_INIT) {
+ atomic_set(&drv_data->cancel_state, state);
+ wake_up_all(&acpihp_drv_cancel_queue);
+ }
+}
+
+/*
+ * Wait for all slots on the list to reach a stable state and then check
+ * cancellation result.
+ */
+int acpihp_drv_cancel_wait(struct list_head *list)
+{
+ int state, result = 0;
+ struct acpihp_slot_drv *drv_data;
+ struct acpihp_slot_dependency *dep;
+
+ list_for_each_entry(dep, list, node) {
+ acpihp_drv_get_data(dep->slot, &drv_data);
+ wait_event(acpihp_drv_cancel_queue,
+ atomic_read(&drv_data->cancel_state)
+ != ACPIHP_DRV_CANCEL_STARTED);
+
+ state = atomic_read(&drv_data->cancel_state);
+ if (state == ACPIHP_DRV_CANCEL_FAILED) {
+ ACPIHP_SLOT_DEBUG(dep->slot,
+ "fails to cancel operation.\n");
+ result = result ? : -EBUSY;
+ } else if (state == ACPIHP_DRV_CANCEL_MISSED) {
+ ACPIHP_SLOT_DEBUG(dep->slot,
+ "misses to cancel operation.\n");
+ result = result ? : -EBUSY;
+ }
+
+ atomic_set(&drv_data->cancel_state,
+ ACPIHP_DRV_CANCEL_FINISHED);
+ atomic_dec(&drv_data->cancel_users);
+ wake_up_all(&acpihp_drv_cancel_queue);
+ }
+
+ return result;
+}
+
+/*
+ * Wait for all cancellation threads to give up their reference count.
+ *
+ * Caller must provide mechanism to avoid currently running
+ * acpihp_drv_cancel_start() and acpihp_drv_cancel_fini()
+ * on the same list.
+ */
+void acpihp_drv_cancel_fini(struct list_head *list)
+{
+ int state;
+ struct acpihp_slot_drv *drv_data;
+ struct acpihp_slot_dependency *dep;
+
+ list_for_each_entry(dep, list, node) {
+ acpihp_drv_get_data(dep->slot, &drv_data);
+
+ /*
+ * Wake up all cancellation threads if they are still
+ * STARTED state.
+ */
+ state = atomic_cmpxchg(&drv_data->cancel_state,
+ ACPIHP_DRV_CANCEL_STARTED,
+ ACPIHP_DRV_CANCEL_MISSED);
+ if (state == ACPIHP_DRV_CANCEL_STARTED)
+ wake_up_all(&acpihp_drv_cancel_queue);
+
+ /* Wait for all cancellation threads to exit */
+ wait_event(acpihp_drv_cancel_queue,
+ !atomic_read(&drv_data->cancel_users));
+ }
+}
--
1.7.9.5
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2012-11-04 12:51 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-11-04 12:50 [ACPIHP PATCH part2 00/13] introduce ACPI based system device hotplug driver Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 01/13] ACPIHP: introduce interfaces to scan and walk ACPI devices attached to a slot Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 02/13] ACPIHP: use klist to manage " Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 03/13] ACPIHP: add callbacks into acpi_device_ops to support new hotplug framework Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 04/13] ACPIHP: provide interfaces to manage driver data associated with hotplug slots Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 05/13] ACPIHP: implement utility interfaces to support system device hotplug Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 06/13] ACPIHP: implement ACPI system device hotplug driver skeleton Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 07/13] ACPIHP: analyse dependencies among ACPI hotplug slots Jiang Liu
2012-11-04 12:50 ` Jiang Liu [this message]
2012-11-04 12:50 ` [ACPIHP PATCH part2 09/13] ACPIHP: configure/unconfigure system devices attached to a hotplug slot Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 10/13] ACPIHP: implement the core state machine to manage hotplug slots Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 11/13] ACPIHP: block ACPI device driver from unloading when doing hotplug Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 12/13] ACPIHP: implement sysfs interfaces for system device hotplug Jiang Liu
2012-11-04 12:50 ` [ACPIHP PATCH part2 13/13] ACPIHP: handle ACPI device hotplug events Jiang 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=1352033415-5606-9-git-send-email-jiang.liu@huawei.com \
--to=liuj97@gmail.com \
--cc=bhelgaas@google.com \
--cc=guohanjun@huawei.com \
--cc=isimatu.yasuaki@jp.fujitsu.com \
--cc=izumi.taku@jp.fujitsu.com \
--cc=jiang.liu@huawei.com \
--cc=kaneshige.kenji@jp.fujitsu.com \
--cc=lenb@kernel.org \
--cc=linux-acpi@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-pci@vger.kernel.org \
--cc=rjw@sisk.pl \
--cc=robert.moore@intel.com \
--cc=srivatsa.bhat@linux.vnet.ibm.com \
--cc=tangchen@cn.fujitsu.com \
--cc=tony.luck@intel.com \
--cc=wangyijing@huawei.com \
--cc=wency@cn.fujitsu.com \
--cc=ying.huang@intel.com \
--cc=yinghai@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).