stable.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	stable@vger.kernel.org, Roland Dreier <roland@purestorage.com>,
	"Rafael J. Wysocki" <rafael.j.wysocki@intel.com>
Subject: [PATCH 4.1 25/65] ACPI / PNP: Avoid conflicting resource reservations
Date: Sun, 19 Jul 2015 12:07:44 -0700	[thread overview]
Message-ID: <20150719190810.308623668@linuxfoundation.org> (raw)
In-Reply-To: <20150719190809.469715936@linuxfoundation.org>

4.1-stable review patch.  If anyone has any objections, please let me know.

------------------

From: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>

commit 0f1b414d190724617eb1cdd615592fa8cd9d0b50 upstream.

Commit b9a5e5e18fbf "ACPI / init: Fix the ordering of
acpi_reserve_resources()" overlooked the fact that the memory
and/or I/O regions reserved by acpi_reserve_resources() may
conflict with those reserved by the PNP "system" driver.

If that conflict actually takes place, it causes the reservations
made by the "system" driver to fail while before commit b9a5e5e18fbf
all reservations made by it and by acpi_reserve_resources() would be
successful.  In turn, that allows the resources that haven't been
reserved by the "system" driver to be used by others (e.g. PCI) which
sometimes leads to functional problems (up to and including boot
failures).

To fix that issue, introduce a common resource reservation routine,
acpi_reserve_region(), to be used by both acpi_reserve_resources()
and the "system" driver, that will track all resources reserved by
it and avoid making conflicting requests.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=99831
Link: http://marc.info/?t=143389402600001&r=1&w=2
Fixes: b9a5e5e18fbf "ACPI / init: Fix the ordering of acpi_reserve_resources()"
Reported-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 drivers/acpi/osl.c      |    6 -
 drivers/acpi/resource.c |  160 ++++++++++++++++++++++++++++++++++++++++++++++++
 drivers/pnp/system.c    |   35 +++++++---
 include/linux/acpi.h    |   10 +++
 4 files changed, 197 insertions(+), 14 deletions(-)

--- a/drivers/acpi/osl.c
+++ b/drivers/acpi/osl.c
@@ -175,11 +175,7 @@ static void __init acpi_request_region (
 	if (!addr || !length)
 		return;
 
-	/* Resources are never freed */
-	if (gas->space_id == ACPI_ADR_SPACE_SYSTEM_IO)
-		request_region(addr, length, desc);
-	else if (gas->space_id == ACPI_ADR_SPACE_SYSTEM_MEMORY)
-		request_mem_region(addr, length, desc);
+	acpi_reserve_region(addr, length, gas->space_id, 0, desc);
 }
 
 static void __init acpi_reserve_resources(void)
--- a/drivers/acpi/resource.c
+++ b/drivers/acpi/resource.c
@@ -26,6 +26,7 @@
 #include <linux/device.h>
 #include <linux/export.h>
 #include <linux/ioport.h>
+#include <linux/list.h>
 #include <linux/slab.h>
 
 #ifdef CONFIG_X86
@@ -621,3 +622,162 @@ int acpi_dev_filter_resource_type(struct
 	return (type & types) ? 0 : 1;
 }
 EXPORT_SYMBOL_GPL(acpi_dev_filter_resource_type);
+
+struct reserved_region {
+	struct list_head node;
+	u64 start;
+	u64 end;
+};
+
+static LIST_HEAD(reserved_io_regions);
+static LIST_HEAD(reserved_mem_regions);
+
+static int request_range(u64 start, u64 end, u8 space_id, unsigned long flags,
+			 char *desc)
+{
+	unsigned int length = end - start + 1;
+	struct resource *res;
+
+	res = space_id == ACPI_ADR_SPACE_SYSTEM_IO ?
+		request_region(start, length, desc) :
+		request_mem_region(start, length, desc);
+	if (!res)
+		return -EIO;
+
+	res->flags &= ~flags;
+	return 0;
+}
+
+static int add_region_before(u64 start, u64 end, u8 space_id,
+			     unsigned long flags, char *desc,
+			     struct list_head *head)
+{
+	struct reserved_region *reg;
+	int error;
+
+	reg = kmalloc(sizeof(*reg), GFP_KERNEL);
+	if (!reg)
+		return -ENOMEM;
+
+	error = request_range(start, end, space_id, flags, desc);
+	if (error)
+		return error;
+
+	reg->start = start;
+	reg->end = end;
+	list_add_tail(&reg->node, head);
+	return 0;
+}
+
+/**
+ * acpi_reserve_region - Reserve an I/O or memory region as a system resource.
+ * @start: Starting address of the region.
+ * @length: Length of the region.
+ * @space_id: Identifier of address space to reserve the region from.
+ * @flags: Resource flags to clear for the region after requesting it.
+ * @desc: Region description (for messages).
+ *
+ * Reserve an I/O or memory region as a system resource to prevent others from
+ * using it.  If the new region overlaps with one of the regions (in the given
+ * address space) already reserved by this routine, only the non-overlapping
+ * parts of it will be reserved.
+ *
+ * Returned is either 0 (success) or a negative error code indicating a resource
+ * reservation problem.  It is the code of the first encountered error, but the
+ * routine doesn't abort until it has attempted to request all of the parts of
+ * the new region that don't overlap with other regions reserved previously.
+ *
+ * The resources requested by this routine are never released.
+ */
+int acpi_reserve_region(u64 start, unsigned int length, u8 space_id,
+			unsigned long flags, char *desc)
+{
+	struct list_head *regions;
+	struct reserved_region *reg;
+	u64 end = start + length - 1;
+	int ret = 0, error = 0;
+
+	if (space_id == ACPI_ADR_SPACE_SYSTEM_IO)
+		regions = &reserved_io_regions;
+	else if (space_id == ACPI_ADR_SPACE_SYSTEM_MEMORY)
+		regions = &reserved_mem_regions;
+	else
+		return -EINVAL;
+
+	if (list_empty(regions))
+		return add_region_before(start, end, space_id, flags, desc, regions);
+
+	list_for_each_entry(reg, regions, node)
+		if (reg->start == end + 1) {
+			/* The new region can be prepended to this one. */
+			ret = request_range(start, end, space_id, flags, desc);
+			if (!ret)
+				reg->start = start;
+
+			return ret;
+		} else if (reg->start > end) {
+			/* No overlap.  Add the new region here and get out. */
+			return add_region_before(start, end, space_id, flags,
+						 desc, &reg->node);
+		} else if (reg->end == start - 1) {
+			goto combine;
+		} else if (reg->end >= start) {
+			goto overlap;
+		}
+
+	/* The new region goes after the last existing one. */
+	return add_region_before(start, end, space_id, flags, desc, regions);
+
+ overlap:
+	/*
+	 * The new region overlaps an existing one.
+	 *
+	 * The head part of the new region immediately preceding the existing
+	 * overlapping one can be combined with it right away.
+	 */
+	if (reg->start > start) {
+		error = request_range(start, reg->start - 1, space_id, flags, desc);
+		if (error)
+			ret = error;
+		else
+			reg->start = start;
+	}
+
+ combine:
+	/*
+	 * The new region is adjacent to an existing one.  If it extends beyond
+	 * that region all the way to the next one, it is possible to combine
+	 * all three of them.
+	 */
+	while (reg->end < end) {
+		struct reserved_region *next = NULL;
+		u64 a = reg->end + 1, b = end;
+
+		if (!list_is_last(&reg->node, regions)) {
+			next = list_next_entry(reg, node);
+			if (next->start <= end)
+				b = next->start - 1;
+		}
+		error = request_range(a, b, space_id, flags, desc);
+		if (!error) {
+			if (next && next->start == b + 1) {
+				reg->end = next->end;
+				list_del(&next->node);
+				kfree(next);
+			} else {
+				reg->end = end;
+				break;
+			}
+		} else if (next) {
+			if (!ret)
+				ret = error;
+
+			reg = next;
+		} else {
+			break;
+		}
+	}
+
+	return ret ? ret : error;
+}
+EXPORT_SYMBOL_GPL(acpi_reserve_region);
--- a/drivers/pnp/system.c
+++ b/drivers/pnp/system.c
@@ -7,6 +7,7 @@
  *	Bjorn Helgaas <bjorn.helgaas@hp.com>
  */
 
+#include <linux/acpi.h>
 #include <linux/pnp.h>
 #include <linux/device.h>
 #include <linux/init.h>
@@ -22,25 +23,41 @@ static const struct pnp_device_id pnp_de
 	{"", 0}
 };
 
+#ifdef CONFIG_ACPI
+static bool __reserve_range(u64 start, unsigned int length, bool io, char *desc)
+{
+	u8 space_id = io ? ACPI_ADR_SPACE_SYSTEM_IO : ACPI_ADR_SPACE_SYSTEM_MEMORY;
+	return !acpi_reserve_region(start, length, space_id, IORESOURCE_BUSY, desc);
+}
+#else
+static bool __reserve_range(u64 start, unsigned int length, bool io, char *desc)
+{
+	struct resource *res;
+
+	res = io ? request_region(start, length, desc) :
+		request_mem_region(start, length, desc);
+	if (res) {
+		res->flags &= ~IORESOURCE_BUSY;
+		return true;
+	}
+	return false;
+}
+#endif
+
 static void reserve_range(struct pnp_dev *dev, struct resource *r, int port)
 {
 	char *regionid;
 	const char *pnpid = dev_name(&dev->dev);
 	resource_size_t start = r->start, end = r->end;
-	struct resource *res;
+	bool reserved;
 
 	regionid = kmalloc(16, GFP_KERNEL);
 	if (!regionid)
 		return;
 
 	snprintf(regionid, 16, "pnp %s", pnpid);
-	if (port)
-		res = request_region(start, end - start + 1, regionid);
-	else
-		res = request_mem_region(start, end - start + 1, regionid);
-	if (res)
-		res->flags &= ~IORESOURCE_BUSY;
-	else
+	reserved = __reserve_range(start, end - start + 1, !!port, regionid);
+	if (!reserved)
 		kfree(regionid);
 
 	/*
@@ -49,7 +66,7 @@ static void reserve_range(struct pnp_dev
 	 * have double reservations.
 	 */
 	dev_info(&dev->dev, "%pR %s reserved\n", r,
-		 res ? "has been" : "could not be");
+		 reserved ? "has been" : "could not be");
 }
 
 static void reserve_resources_of_dev(struct pnp_dev *dev)
--- a/include/linux/acpi.h
+++ b/include/linux/acpi.h
@@ -332,6 +332,9 @@ int acpi_check_region(resource_size_t st
 
 int acpi_resources_are_enforced(void);
 
+int acpi_reserve_region(u64 start, unsigned int length, u8 space_id,
+			unsigned long flags, char *desc);
+
 #ifdef CONFIG_HIBERNATION
 void __init acpi_no_s4_hw_signature(void);
 #endif
@@ -527,6 +530,13 @@ static inline int acpi_check_region(reso
 	return 0;
 }
 
+static inline int acpi_reserve_region(u64 start, unsigned int length,
+				      u8 space_id, unsigned long flags,
+				      char *desc)
+{
+	return -ENXIO;
+}
+
 struct acpi_table_header;
 static inline int acpi_table_parse(char *id,
 				int (*handler)(struct acpi_table_header *))



  parent reply	other threads:[~2015-07-19 19:08 UTC|newest]

Thread overview: 70+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-07-19 19:07 [PATCH 4.1 00/65] 4.1.3-stable review Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 01/65] fs: Add helper functions for permanently empty directories Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 02/65] sysctl: Allow creating permanently empty directories that serve as mountpoints Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 03/65] proc: Allow creating permanently empty directories that serve as mount points Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 04/65] kernfs: Add support for always empty directories Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 05/65] sysfs: Add support for permanently empty directories to serve as mount points Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 06/65] sysfs: Create mountpoints with sysfs_create_mount_point Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 07/65] mnt: Update fs_fully_visible to test for permanently empty directories Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 08/65] mnt: Refactor the logic for mounting sysfs and proc in a user namespace Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 09/65] mnt: Modify fs_fully_visible to deal with locked ro nodev and atime Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 10/65] gpio: crystalcove: set IRQCHIP_SKIP_SET_WAKE for the irqchip Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 11/65] gpio: rcar: Check for irq_set_irq_wake() failures Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 12/65] rcu: Correctly handle non-empty Tiny RCU callback list with none ready Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 13/65] ipr: Increase default adapter init stage change timeout Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 14/65] Disable write buffering on Toshiba ToPIC95 Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 15/65] ALSA: pcm: Fix pcm_class sysfs output Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 16/65] ALSA: hda - Fix Dock Headphone on Thinkpad X250 seen as a Line Out Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 17/65] ALSA: hda - set proper caps for newer AMD hda audio in KB/KV Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 18/65] ALSA: hda - Disable widget power-save for VIA codecs Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 19/65] ALSA: hda - restore the MIC FIXUP for some Dell machines Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 20/65] ALSA: hda - Add headset support to Acer Aspire V5 Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 21/65] ALSA: hda - Fix the dock headphone output on Fujitsu Lifebook E780 Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 22/65] ALSA: hda - Add a fixup for Dell E7450 Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 23/65] ACPI / init: Switch over platform to the ACPI mode later Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 24/65] ACPI / PM: Add missing pm_generic_complete() invocation Greg Kroah-Hartman
2015-07-19 19:07 ` Greg Kroah-Hartman [this message]
2015-07-19 19:07 ` [PATCH 4.1 26/65] iio: accel: kxcjk-1013: add the "KXCJ9000" ACPI id Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 27/65] tools selftests: Fix clean target with make 3.81 Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 28/65] ARC: add smp barriers around atomics per Documentation/atomic_ops.txt Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 29/65] ARC: add compiler barrier to LLSC based cmpxchg Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 30/65] arc: fix use of uninitialized arc_pmu Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 31/65] power_supply: Fix NULL pointer dereference during bq27x00_battery probe Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 32/65] power_supply: Fix possible NULL pointer dereference on early uevent Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 33/65] mei: me: wait for power gating exit confirmation Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 34/65] mei: txe: reduce suspend/resume time Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 35/65] arm64: Do not attempt to use init_mm in reset_context() Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 36/65] arm64: entry: fix context tracking for el0_sp_pc Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 37/65] arm64: mm: Fix freeing of the wrong memmap entries with !SPARSEMEM_VMEMMAP Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 38/65] arm64: vdso: work-around broken ELF toolchains in Makefile Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 39/65] mm: kmemleak: allow safe memory scanning during kmemleak disabling Greg Kroah-Hartman
2015-07-19 19:07 ` [PATCH 4.1 40/65] mm: kmemleak_alloc_percpu() should follow the gfp from per_alloc() Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 41/65] mm, thp: respect MPOL_PREFERRED policy with non-local node Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 42/65] regmap: Fix regmap_bulk_read in BE mode Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 43/65] regmap: Fix possible shift overflow in regmap_field_init() Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 44/65] regulator: max77686: fix gpio_enabled shift wrapping bug Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 45/65] regulator: core: fix constraints output buffer Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 46/65] livepatch: add module locking around kallsyms calls Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 48/65] spi: orion: Fix maximum baud rates for Armada 370/XP Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 49/65] spi: pl022: Specify num-cs property as required in devicetree binding Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 50/65] scsi_transport_srp: Introduce srp_wait_for_queuecommand() Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 51/65] scsi_transport_srp: Fix a race condition Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 52/65] IB/srp: Remove an extraneous scsi_host_put() from an error path Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 53/65] IB/srp: Fix a connection setup race Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 54/65] IB/srp: Fix connection state tracking Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 55/65] IB/srp: Fix reconnection failure handling Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 56/65] genirq: devres: Fix testing return value of request_any_context_irq() Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 57/65] video: mxsfb: Make sure axi clock is enabled when accessing registers Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 58/65] leds / PM: fix hibernation on arm when gpio-led used with CPU led trigger Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 59/65] mtd: fix: avoid race condition when accessing mtd->usecount Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 61/65] PCI: Propagate the "ignore hotplug" setting to parent Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 62/65] PCI: Add pci_bus_addr_t Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 63/65] PCI: pciehp: Wait for hotplug command completion where necessary Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 64/65] of/pci: Fix pci_address_to_pio() conversion of CPU address to I/O port Greg Kroah-Hartman
2015-07-19 19:08 ` [PATCH 4.1 65/65] Input: pixcir_i2c_ts - fix receive error Greg Kroah-Hartman
2015-07-20  3:19 ` [PATCH 4.1 00/65] 4.1.3-stable review Guenter Roeck
2015-07-20 19:26   ` Greg Kroah-Hartman
2015-07-20  6:33 ` Sudip Mukherjee
2015-07-20 19:27   ` Greg Kroah-Hartman
2015-07-20 17:17 ` Shuah Khan
2015-07-20 19:27   ` Greg Kroah-Hartman

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=20150719190810.308623668@linuxfoundation.org \
    --to=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rafael.j.wysocki@intel.com \
    --cc=roland@purestorage.com \
    --cc=stable@vger.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).