linux-pci.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jiang Liu <liuj97@gmail.com>
To: Yinghai Lu <yinghai@kernel.org>,
	Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>,
	Bjorn Helgaas <bhelgaas@google.com>,
	Dan Zink <dan.zink@compaq.com>,
	Greg Kroah-Hartman <greg@kroah.com>,
	Dely Sy <dely.l.sy@intel.com>
Cc: Jiang Liu <jiang.liu@huawei.com>, Jiang Liu <liuj97@gmail.com>,
	Keping Chen <chenkeping@huawei.com>,
	linux-pci@vger.kernel.org
Subject: [PATCH RFC 08/17] PCI: fix two race windows when probing/removing SHPC controller
Date: Tue, 17 Apr 2012 00:29:02 +0800	[thread overview]
Message-ID: <1334593751-5916-9-git-send-email-jiang.liu@huawei.com> (raw)
In-Reply-To: <1334593751-5916-1-git-send-email-jiang.liu@huawei.com>

With current shpchp implementation, interrupt is enabled before corresponding
slot data structures are created. If interrupt happens immediately after
enabling the hotplug interrupt, shpchp_find_slot() may return NULL, thus
causes NULL pointer dereference. So create slot data structures before
enabling interrupt.

And cleanup_slots() is called to destroy slot data structures before disabling
interrupt, which may cause invalid memory access in shpc_isr().  So call
cleanup_slots() after disabling interrupt/removing the poll timer.

Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
---
 drivers/pci/hotplug/shpchp.h      |    1 +
 drivers/pci/hotplug/shpchp_core.c |    6 +++---
 drivers/pci/hotplug/shpchp_hpc.c  |   36 +++++++++++++++++++++---------------
 3 files changed, 25 insertions(+), 18 deletions(-)

diff --git a/drivers/pci/hotplug/shpchp.h b/drivers/pci/hotplug/shpchp.h
index ca64932..6691dc4 100644
--- a/drivers/pci/hotplug/shpchp.h
+++ b/drivers/pci/hotplug/shpchp.h
@@ -182,6 +182,7 @@ extern int shpchp_unconfigure_device(struct slot *p_slot);
 extern void cleanup_slots(struct controller *ctrl);
 extern void shpchp_queue_pushbutton_work(struct work_struct *work);
 extern int shpc_init( struct controller *ctrl, struct pci_dev *pdev);
+extern void shpc_enable_intr(struct controller *ctrl);
 
 static inline const char *slot_name(struct slot *slot)
 {
diff --git a/drivers/pci/hotplug/shpchp_core.c b/drivers/pci/hotplug/shpchp_core.c
index 7414fd9..aaa66be 100644
--- a/drivers/pci/hotplug/shpchp_core.c
+++ b/drivers/pci/hotplug/shpchp_core.c
@@ -318,14 +318,14 @@ static int shpc_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		goto err_out_release_ctlr;
 	}
 
+	shpc_enable_intr(ctrl);
+
 	rc = shpchp_create_ctrl_files(ctrl);
 	if (rc)
-		goto err_cleanup_slots;
+		goto err_out_release_ctlr;
 
 	return 0;
 
-err_cleanup_slots:
-	cleanup_slots(ctrl);
 err_out_release_ctlr:
 	ctrl->hpc_ops->release_ctlr(ctrl);
 err_out_free_ctrl:
diff --git a/drivers/pci/hotplug/shpchp_hpc.c b/drivers/pci/hotplug/shpchp_hpc.c
index 75ba231..2bc6182 100644
--- a/drivers/pci/hotplug/shpchp_hpc.c
+++ b/drivers/pci/hotplug/shpchp_hpc.c
@@ -592,6 +592,13 @@ static void hpc_release_ctlr(struct controller *ctrl)
 		shpc_writel(ctrl, SLOT_REG(i), slot_reg);
 	}
 
+	if (shpchp_poll_mode)
+		del_timer(&ctrl->poll_timer);
+	else {
+		free_irq(ctrl->pci_dev->irq, ctrl);
+		pci_disable_msi(ctrl->pci_dev);
+	}
+
 	cleanup_slots(ctrl);
 
 	/*
@@ -603,13 +610,6 @@ static void hpc_release_ctlr(struct controller *ctrl)
 	serr_int &= ~SERR_INTR_RSVDZ_MASK;
 	shpc_writel(ctrl, SERR_INTR_ENABLE, serr_int);
 
-	if (shpchp_poll_mode)
-		del_timer(&ctrl->poll_timer);
-	else {
-		free_irq(ctrl->pci_dev->irq, ctrl);
-		pci_disable_msi(ctrl->pci_dev);
-	}
-
 	iounmap(ctrl->creg);
 	release_mem_region(ctrl->mmio_base, ctrl->mmio_size);
 }
@@ -1081,6 +1081,20 @@ int shpc_init(struct controller *ctrl, struct pci_dev *pdev)
 	shpc_get_max_bus_speed(ctrl);
 	shpc_get_cur_bus_speed(ctrl);
 
+	return 0;
+
+	/* We end up here for the many possible ways to fail this API.  */
+abort_iounmap:
+	iounmap(ctrl->creg);
+abort:
+	return rc;
+}
+
+void shpc_enable_intr(struct controller *ctrl)
+{
+	u8 hp_slot;
+	u32 tempdword, slot_reg;
+
 	/*
 	 * Unmask all event interrupts of all slots
 	 */
@@ -1102,12 +1116,4 @@ int shpc_init(struct controller *ctrl, struct pci_dev *pdev)
 		tempdword = shpc_readl(ctrl, SERR_INTR_ENABLE);
 		ctrl_dbg(ctrl, "SERR_INTR_ENABLE = %x\n", tempdword);
 	}
-
-	return 0;
-
-	/* We end up here for the many possible ways to fail this API.  */
-abort_iounmap:
-	iounmap(ctrl->creg);
-abort:
-	return rc;
 }
-- 
1.7.5.4


  parent reply	other threads:[~2012-04-16 16:33 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-04-16 16:28 [PATCH RFC 00/17] Introduce a global lock to serialize all PCI hotplug Jiang Liu
2012-04-16 16:28 ` [PATCH RFC 01/17] PCI: introduce pci_bus_get()/pci_bus_put() to hide PCI implementation details Jiang Liu
2012-04-16 16:28 ` [PATCH RFC 02/17] PCI: introduce recursive rwsem to serialize PCI hotplug operations Jiang Liu
2012-04-16 16:28 ` [PATCH RFC 03/17] PCI: replace pci_remove_rescan_mutex with the PCI hotplug lock Jiang Liu
2012-04-16 16:28 ` [PATCH RFC 04/17] PCI: serialize hotplug operations triggered by PCI hotplug sysfs interfaces Jiang Liu
2012-04-16 16:28 ` [PATCH RFC 05/17] PCI: correctly flush workqueue when destroy pcie hotplug controller Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 06/17] PCI: prepare for serializing hotplug operations triggered by pciehp driver Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 07/17] PCI: serialize hotplug operaitons triggered by the " Jiang Liu
2012-04-16 16:29 ` Jiang Liu [this message]
2012-04-16 16:29 ` [PATCH RFC 09/17] PCI: correctly flush workqueues and timer when destroy SHPC controller Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 10/17] PCI: serialize hotplug operaitons triggered by the shpchp driver Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 11/17] PCI: release IO resource in error handling path in cpcihp_generic_init() Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 12/17] PCI: clean up all resources in error handling path in zt5550_hc_init_one() Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 13/17] PCI: trivial code clean up in cpci_hotplug_core.c Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 14/17] PCI: fix race windows when shutting down cpcihp controller Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 15/17] PCI: hold a reference count to the PCI bus used by cpcihp drivers Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 16/17] PCI: serialize PCI hotplug operations triggered " Jiang Liu
2012-04-16 16:29 ` [PATCH RFC 17/17] PCI: serialize PCI hotplug operations triggered by fakephp drivers Jiang Liu
2012-04-16 21:33 ` [PATCH RFC 00/17] Introduce a global lock to serialize all PCI hotplug Greg KH
2012-04-17 11:57   ` Jiang Liu
2012-04-17 14:53   ` 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=1334593751-5916-9-git-send-email-jiang.liu@huawei.com \
    --to=liuj97@gmail.com \
    --cc=bhelgaas@google.com \
    --cc=chenkeping@huawei.com \
    --cc=dan.zink@compaq.com \
    --cc=dely.l.sy@intel.com \
    --cc=greg@kroah.com \
    --cc=jiang.liu@huawei.com \
    --cc=kaneshige.kenji@jp.fujitsu.com \
    --cc=linux-pci@vger.kernel.org \
    --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).