[PATCH v4 0/9] driver core: Fix some race conditions

[PATCH v4 0/9] driver core: Fix some race conditions

[Douglas Anderson]

The main goal of this series is to fix the observed bug talked about
in the first patch ("driver core: Don't let a device probe until it's
ready"). That patch fixes a problem that has been observed in the real
world and could land even if the rest of the patches are found
unacceptable or need to be spun.

That said, during patch review Danilo correctly pointed out that many
of the bitfield accesses in "struct device" are unsafe. I added a
bunch of patches in the series to address each one.

Danilo said he's most worried about "can_match", so I put that one
first. After that, I tried to transition bitfields to flags in reverse
order to when the bitfield was added.

Even if transitioning from bitfields to flags isn't truly needed for
correctness, it seems silly (and wasteful of space in struct device)
to have some in bitfields and some as flags. Thus I didn't spend time
for each bitfield showing that it's truly needed for correctness.

Transition was done semi manually. Presumably someone skilled at
coccinelle could do a better job, but I just used sed in a heavy-
handed manner and then reviewed/fixed the results, undoing anything my
script got wrong. My terrible/ugly script was:

var=can_match
caps="${var^^}"
for f in $(git grep -l "[>\.]${var}[^1-9_a-zA-Z\[]"); do
  echo $f
  sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)->${var} = true/set_bit(DEV_FLAG_${caps}, \&\\1->flags)/" "$f"
  sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)\.${var} = true/dev_set_${caps}(\&\\1)/" "$f"
  sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)->${var} = false/clear_bit(DEV_FLAG_${caps}, \&\\1->flags)/" "$f"
  sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)\.${var} = false/dev_clear_${caps}(\&\\1)/" "$f"
  sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)->${var} = \([^;]*\)/assign_bit(DEV_FLAG_${caps}, \&\\1->flags, \\2)/" "$f"
  sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)\.${var} = \([^;]*\)/dev_assign_${caps}(\&\\1, \\2)/" "$f"
  sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)->${var}\([^1-9_a-zA-Z\[]\)/test_bit(DEV_FLAG_${caps}, \&\\1->flags)\\2/" "$f"
  sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)\.${var}\([^1-9_a-zA-Z\[]\)/dev_${caps}(\&\\1)\\2/" "$f"
done

From v3 to v4, I transitioned to accessor functions with another ugly
sed script. I had git format the old patches, then transformed them
with:

for f in *.patch; do
  echo $f
  sed -i~ -e "s/test_and_set_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags)/dev_test_and_set_\\L\\1(\\2)/" "$f"
  sed -i~ -e "s/test_and_set_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags)/dev_test_and_set_\\L\\1(\\2)/" "$f"
  sed -i~ -e "s/test_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags)/dev_\\L\\1(\\2)/" "$f"
  sed -i~ -e "s/test_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags)/dev_\\L\\1(\\2)/" "$f"
  sed -i~ -e "s/set_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags)/dev_set_\\L\\1(\\2)/" "$f"
  sed -i~ -e "s/set_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags)/dev_set_\\L\\1(\\2)/" "$f"
  sed -i~ -e "s/clear_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags)/dev_clear_\\L\\1(\\2)/" "$f"
  sed -i~ -e "s/clear_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags)/dev_clear_\\L\\1(\\2)/" "$f"
  sed -i~ -e "s/assign_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags, \(.*\))/dev_assign_\\L\\1(\\2, \\3)/" "$f"
  sed -i~ -e "s/assign_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags, \(.*\))/dev_assign_\\L\\1(\\2, \\3)/" "$f"
done

...and then did a few manual touchups for spacing.

NOTE: one potentially "controversial" choice I made in some patches
was to always reserve a flag ID even if a flag is only used under
certain CONFIG_ settings. This is a change from how things were
before. Keeping the numbering consistent and allowing easy
compile-testing of both CONFIG settings seemed worth it, especially
since it won't take up any extra space until we've added a lot more
flags.

I only marked the first patch as a "Fix" since it is the only one
fixing observed problems. Other patches could be considered fixes too
if folks want.

I tested the first patch in the series backported to kernel 6.6 on the
Pixel phone that was experiencing the race. I added extra printouts to
make sure that the problem was hitting / addressed. The rest of the
patches are tested with allmodconfig with arm32, arm64, ppc, and
x86. I boot tested on an arm64 Chromebook running mainline.

Changes in v4:
- Use accessor functions for flags

Changes in v3:
- Use a new "flags" bitfield
- Add missing \n in probe error message

Changes in v2:
- Instead of adjusting the ordering, use "ready_to_probe" flag

Douglas Anderson (9):
  driver core: Don't let a device probe until it's ready
  driver core: Replace dev->can_match with dev_can_match()
  driver core: Replace dev->dma_iommu with dev_dma_iommu()
  driver core: Replace dev->dma_skip_sync with dev_dma_skip_sync()
  driver core: Replace dev->dma_ops_bypass with dev_dma_ops_bypass()
  driver core: Replace dev->state_synced with dev_state_synced()
  driver core: Replace dev->dma_coherent with dev_dma_coherent()
  driver core: Replace dev->of_node_reused with dev_of_node_reused()
  driver core: Replace dev->offline + ->offline_disabled with accessors

 arch/arc/mm/dma.c                             |   4 +-
 arch/arm/mach-highbank/highbank.c             |   2 +-
 arch/arm/mach-mvebu/coherency.c               |   2 +-
 arch/arm/mm/dma-mapping-nommu.c               |   4 +-
 arch/arm/mm/dma-mapping.c                     |  28 ++--
 arch/arm64/kernel/cpufeature.c                |   2 +-
 arch/arm64/mm/dma-mapping.c                   |   2 +-
 arch/mips/mm/dma-noncoherent.c                |   2 +-
 arch/powerpc/kernel/dma-iommu.c               |   8 +-
 .../platforms/pseries/hotplug-memory.c        |   4 +-
 arch/riscv/mm/dma-noncoherent.c               |   2 +-
 drivers/acpi/scan.c                           |   2 +-
 drivers/base/core.c                           |  53 +++++---
 drivers/base/cpu.c                            |   4 +-
 drivers/base/dd.c                             |  28 ++--
 drivers/base/memory.c                         |   2 +-
 drivers/base/pinctrl.c                        |   2 +-
 drivers/base/platform.c                       |   2 +-
 drivers/dma/ti/k3-udma-glue.c                 |   6 +-
 drivers/dma/ti/k3-udma.c                      |   6 +-
 drivers/iommu/dma-iommu.c                     |   9 +-
 drivers/iommu/iommu.c                         |   5 +-
 drivers/net/pcs/pcs-xpcs-plat.c               |   2 +-
 drivers/of/device.c                           |   6 +-
 drivers/pci/of.c                              |   2 +-
 drivers/pci/pwrctrl/core.c                    |   2 +-
 drivers/regulator/bq257xx-regulator.c         |   2 +-
 drivers/regulator/rk808-regulator.c           |   2 +-
 drivers/tty/serial/serial_base_bus.c          |   2 +-
 drivers/usb/gadget/udc/aspeed-vhub/dev.c      |   2 +-
 include/linux/device.h                        | 120 ++++++++++++------
 include/linux/dma-map-ops.h                   |   6 +-
 include/linux/dma-mapping.h                   |   2 +-
 include/linux/iommu-dma.h                     |   3 +-
 kernel/cpu.c                                  |   4 +-
 kernel/dma/mapping.c                          |  12 +-
 mm/hmm.c                                      |   2 +-
 37 files changed, 206 insertions(+), 142 deletions(-)

-- 
2.53.0.1213.gd9a14994de-goog

[PATCH v4 4/9] driver core: Replace dev->dma_skip_sync with dev_dma_skip_sync()

[Douglas Anderson]

In C, bitfields are not necessarily safe to modify from multiple
threads without locking. Switch "dma_skip_sync" over to the "flags"
field so modifications are safe.

Cc: Alexander Lobakin <aleksander.lobakin@intel.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
---
Not fixing any known bugs; problem is theoretical and found by code
inspection. Change is done somewhat manually and only lightly tested
(mostly compile-time tested).

NOTE: even though previously we only took up a bit if
CONFIG_DMA_NEED_SYNC, in this change I reserve the bit
unconditionally. While we could get the "dynamic" behavior by changing
the flags definition to be an "enum", it doesn't seem worth it at this
point.

Changes in v4:
- Use accessor functions for flags

Changes in v3:
- New

 include/linux/device.h      | 8 ++++----
 include/linux/dma-map-ops.h | 4 ++--
 include/linux/dma-mapping.h | 2 +-
 kernel/dma/mapping.c        | 8 ++++----
 mm/hmm.c                    | 2 +-
 5 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/include/linux/device.h b/include/linux/device.h
index feb11ba1ba71..fde0c72c3159 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -471,11 +471,14 @@ struct device_physical_location {
  *		until other devices probe successfully.
  * @DEV_FLAG_DMA_IOMMU: Device is using default IOMMU implementation for DMA and
  *		doesn't rely on dma_ops structure.
+ * @DEV_FLAG_DMA_SKIP_SYNC: DMA sync operations can be skipped for coherent
+ *		buffers.
  */
 enum struct_device_flags {
 	DEV_FLAG_READY_TO_PROBE = 0,
 	DEV_FLAG_CAN_MATCH = 1,
 	DEV_FLAG_DMA_IOMMU = 2,
+	DEV_FLAG_DMA_SKIP_SYNC = 3,
 
 	DEV_FLAG_COUNT
 };
@@ -569,7 +572,6 @@ enum struct_device_flags {
  *		and optionall (if the coherent mask is large enough) also
  *		for dma allocations.  This flag is managed by the dma ops
  *		instance from ->dma_supported.
- * @dma_skip_sync: DMA sync operations can be skipped for coherent buffers.
  * @flags:	DEV_FLAG_XXX flags. Use atomic bitfield operations to modify.
  *
  * At the lowest level, every device in a Linux system is represented by an
@@ -686,9 +688,6 @@ struct device {
 #ifdef CONFIG_DMA_OPS_BYPASS
 	bool			dma_ops_bypass : 1;
 #endif
-#ifdef CONFIG_DMA_NEED_SYNC
-	bool			dma_skip_sync:1;
-#endif
 
 	DECLARE_BITMAP(flags, DEV_FLAG_COUNT);
 };
@@ -718,6 +717,7 @@ static inline bool dev_test_and_set_##accessor_name(struct device *dev) \
 __create_dev_flag_accessors(ready_to_probe, DEV_FLAG_READY_TO_PROBE);
 __create_dev_flag_accessors(can_match, DEV_FLAG_CAN_MATCH);
 __create_dev_flag_accessors(dma_iommu, DEV_FLAG_DMA_IOMMU);
+__create_dev_flag_accessors(dma_skip_sync, DEV_FLAG_DMA_SKIP_SYNC);
 
 /**
  * struct device_link - Device link representation.
diff --git a/include/linux/dma-map-ops.h b/include/linux/dma-map-ops.h
index 60b63756df82..edd7de60a957 100644
--- a/include/linux/dma-map-ops.h
+++ b/include/linux/dma-map-ops.h
@@ -245,8 +245,8 @@ static inline void dma_reset_need_sync(struct device *dev)
 {
 #ifdef CONFIG_DMA_NEED_SYNC
 	/* Reset it only once so that the function can be called on hotpath */
-	if (unlikely(dev->dma_skip_sync))
-		dev->dma_skip_sync = false;
+	if (unlikely(dev_dma_skip_sync(dev)))
+		dev_clear_dma_skip_sync(dev);
 #endif
 }
 
diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
index 99ef042ecdb4..ef4feb2b37d8 100644
--- a/include/linux/dma-mapping.h
+++ b/include/linux/dma-mapping.h
@@ -419,7 +419,7 @@ bool __dma_need_sync(struct device *dev, dma_addr_t dma_addr);
 static inline bool dma_dev_need_sync(const struct device *dev)
 {
 	/* Always call DMA sync operations when debugging is enabled */
-	return !dev->dma_skip_sync || IS_ENABLED(CONFIG_DMA_API_DEBUG);
+	return !dev_dma_skip_sync(dev) || IS_ENABLED(CONFIG_DMA_API_DEBUG);
 }
 
 static inline void dma_sync_single_for_cpu(struct device *dev, dma_addr_t addr,
diff --git a/kernel/dma/mapping.c b/kernel/dma/mapping.c
index 6d3dd0bd3a88..76fc65d1a8a4 100644
--- a/kernel/dma/mapping.c
+++ b/kernel/dma/mapping.c
@@ -467,7 +467,7 @@ bool dma_need_unmap(struct device *dev)
 {
 	if (!dma_map_direct(dev, get_dma_ops(dev)))
 		return true;
-	if (!dev->dma_skip_sync)
+	if (!dev_dma_skip_sync(dev))
 		return true;
 	return IS_ENABLED(CONFIG_DMA_API_DEBUG);
 }
@@ -483,16 +483,16 @@ static void dma_setup_need_sync(struct device *dev)
 		 * mapping, if any. During the device initialization, it's
 		 * enough to check only for the DMA coherence.
 		 */
-		dev->dma_skip_sync = dev_is_dma_coherent(dev);
+		dev_assign_dma_skip_sync(dev, dev_is_dma_coherent(dev));
 	else if (!ops->sync_single_for_device && !ops->sync_single_for_cpu &&
 		 !ops->sync_sg_for_device && !ops->sync_sg_for_cpu)
 		/*
 		 * Synchronization is not possible when none of DMA sync ops
 		 * is set.
 		 */
-		dev->dma_skip_sync = true;
+		dev_set_dma_skip_sync(dev);
 	else
-		dev->dma_skip_sync = false;
+		dev_clear_dma_skip_sync(dev);
 }
 #else /* !CONFIG_DMA_NEED_SYNC */
 static inline void dma_setup_need_sync(struct device *dev) { }
diff --git a/mm/hmm.c b/mm/hmm.c
index 5955f2f0c83d..c72c9ddfdb2f 100644
--- a/mm/hmm.c
+++ b/mm/hmm.c
@@ -709,7 +709,7 @@ int hmm_dma_map_alloc(struct device *dev, struct hmm_dma_map *map,
 	 * best approximation to ensure no swiotlb buffering happens.
 	 */
 #ifdef CONFIG_DMA_NEED_SYNC
-	dma_need_sync = !dev->dma_skip_sync;
+	dma_need_sync = !dev_dma_skip_sync(dev);
 #endif /* CONFIG_DMA_NEED_SYNC */
 	if (dma_need_sync || dma_addressing_limited(dev))
 		return -EOPNOTSUPP;
-- 
2.53.0.1213.gd9a14994de-goog

[PATCH v4 9/9] driver core: Replace dev->offline + ->offline_disabled with accessors

[Douglas Anderson]

In C, bitfields are not necessarily safe to modify from multiple
threads without locking. Switch "offline" and "offline_disabled" over
to the "flags" field so modifications are safe.

Cc: Rafael J. Wysocki <rafael@kernel.org>
Acked-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
---
Not fixing any known bugs; problem is theoretical and found by code
inspection. Change is done somewhat manually and only lightly tested
(mostly compile-time tested).

Changes in v4:
- Use accessor functions for flags

Changes in v3:
- New

 arch/arm64/kernel/cpufeature.c                 |  2 +-
 .../powerpc/platforms/pseries/hotplug-memory.c |  4 ++--
 drivers/acpi/scan.c                            |  2 +-
 drivers/base/core.c                            | 18 +++++++++---------
 drivers/base/cpu.c                             |  4 ++--
 drivers/base/memory.c                          |  2 +-
 include/linux/device.h                         | 12 ++++++------
 kernel/cpu.c                                   |  4 ++--
 8 files changed, 24 insertions(+), 24 deletions(-)

diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index 32c2dbcc0c64..c832aa565dc2 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -4042,7 +4042,7 @@ static int enable_mismatched_32bit_el0(unsigned int cpu)
 	 */
 	lucky_winner = cpu_32bit ? cpu : cpumask_any_and(cpu_32bit_el0_mask,
 							 cpu_active_mask);
-	get_cpu_device(lucky_winner)->offline_disabled = true;
+	dev_set_offline_disabled(get_cpu_device(lucky_winner));
 	setup_elf_hwcaps(compat_elf_hwcaps);
 	elf_hwcap_fixup();
 	pr_info("Asymmetric 32-bit EL0 support detected on CPU %u; CPU hot-unplug disabled on CPU %u\n",
diff --git a/arch/powerpc/platforms/pseries/hotplug-memory.c b/arch/powerpc/platforms/pseries/hotplug-memory.c
index b2f14db59034..75f85a5da981 100644
--- a/arch/powerpc/platforms/pseries/hotplug-memory.c
+++ b/arch/powerpc/platforms/pseries/hotplug-memory.c
@@ -213,9 +213,9 @@ static int dlpar_change_lmb_state(struct drmem_lmb *lmb, bool online)
 		return -EINVAL;
 	}
 
-	if (online && mem_block->dev.offline)
+	if (online && dev_offline(&mem_block->dev))
 		rc = device_online(&mem_block->dev);
-	else if (!online && !mem_block->dev.offline)
+	else if (!online && !dev_offline(&mem_block->dev))
 		rc = device_offline(&mem_block->dev);
 	else
 		rc = 0;
diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c
index e8cdbdb46fdb..1353adbcc234 100644
--- a/drivers/acpi/scan.c
+++ b/drivers/acpi/scan.c
@@ -122,7 +122,7 @@ bool acpi_scan_is_offline(struct acpi_device *adev, bool uevent)
 	mutex_lock_nested(&adev->physical_node_lock, SINGLE_DEPTH_NESTING);
 
 	list_for_each_entry(pn, &adev->physical_node_list, node)
-		if (device_supports_offline(pn->dev) && !pn->dev->offline) {
+		if (device_supports_offline(pn->dev) && !dev_offline(pn->dev)) {
 			if (uevent)
 				kobject_uevent_env(&pn->dev->kobj, KOBJ_CHANGE, envp);
 
diff --git a/drivers/base/core.c b/drivers/base/core.c
index f12f3b53b4d0..79c85dbcffad 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -2788,7 +2788,7 @@ static ssize_t online_show(struct device *dev, struct device_attribute *attr,
 	bool val;
 
 	device_lock(dev);
-	val = !dev->offline;
+	val = !dev_offline(dev);
 	device_unlock(dev);
 	return sysfs_emit(buf, "%u\n", val);
 }
@@ -2913,7 +2913,7 @@ static int device_add_attrs(struct device *dev)
 	if (error)
 		goto err_remove_type_groups;
 
-	if (device_supports_offline(dev) && !dev->offline_disabled) {
+	if (device_supports_offline(dev) && !dev_offline_disabled(dev)) {
 		error = device_create_file(dev, &dev_attr_online);
 		if (error)
 			goto err_remove_dev_groups;
@@ -4178,7 +4178,7 @@ static int device_check_offline(struct device *dev, void *not_used)
 	if (ret)
 		return ret;
 
-	return device_supports_offline(dev) && !dev->offline ? -EBUSY : 0;
+	return device_supports_offline(dev) && !dev_offline(dev) ? -EBUSY : 0;
 }
 
 /**
@@ -4196,7 +4196,7 @@ int device_offline(struct device *dev)
 {
 	int ret;
 
-	if (dev->offline_disabled)
+	if (dev_offline_disabled(dev))
 		return -EPERM;
 
 	ret = device_for_each_child(dev, NULL, device_check_offline);
@@ -4205,13 +4205,13 @@ int device_offline(struct device *dev)
 
 	device_lock(dev);
 	if (device_supports_offline(dev)) {
-		if (dev->offline) {
+		if (dev_offline(dev)) {
 			ret = 1;
 		} else {
 			ret = dev->bus->offline(dev);
 			if (!ret) {
 				kobject_uevent(&dev->kobj, KOBJ_OFFLINE);
-				dev->offline = true;
+				dev_set_offline(dev);
 			}
 		}
 	}
@@ -4236,11 +4236,11 @@ int device_online(struct device *dev)
 
 	device_lock(dev);
 	if (device_supports_offline(dev)) {
-		if (dev->offline) {
+		if (dev_offline(dev)) {
 			ret = dev->bus->online(dev);
 			if (!ret) {
 				kobject_uevent(&dev->kobj, KOBJ_ONLINE);
-				dev->offline = false;
+				dev_clear_offline(dev);
 			}
 		} else {
 			ret = 1;
@@ -4714,7 +4714,7 @@ static int device_attrs_change_owner(struct device *dev, kuid_t kuid,
 	if (error)
 		return error;
 
-	if (device_supports_offline(dev) && !dev->offline_disabled) {
+	if (device_supports_offline(dev) && !dev_offline_disabled(dev)) {
 		/* Change online device attributes of @dev to @kuid/@kgid. */
 		error = sysfs_file_change_owner(kobj, dev_attr_online.attr.name,
 						kuid, kgid);
diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
index 875abdc9942e..19d288a3c80c 100644
--- a/drivers/base/cpu.c
+++ b/drivers/base/cpu.c
@@ -422,8 +422,8 @@ int register_cpu(struct cpu *cpu, int num)
 	cpu->dev.id = num;
 	cpu->dev.bus = &cpu_subsys;
 	cpu->dev.release = cpu_device_release;
-	cpu->dev.offline_disabled = !cpu->hotpluggable;
-	cpu->dev.offline = !cpu_online(num);
+	dev_assign_offline_disabled(&cpu->dev, !cpu->hotpluggable);
+	dev_assign_offline(&cpu->dev, !cpu_online(num));
 	cpu->dev.of_node = of_get_cpu_node(num, NULL);
 	cpu->dev.groups = common_cpu_attr_groups;
 	if (cpu->hotpluggable)
diff --git a/drivers/base/memory.c b/drivers/base/memory.c
index a3091924918b..5005654f44fa 100644
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -697,7 +697,7 @@ static int __add_memory_block(struct memory_block *memory)
 	memory->dev.id = memory->start_section_nr / sections_per_block;
 	memory->dev.release = memory_block_release;
 	memory->dev.groups = memory_memblk_attr_groups;
-	memory->dev.offline = memory->state == MEM_OFFLINE;
+	dev_assign_offline(&memory->dev, memory->state == MEM_OFFLINE);
 
 	ret = device_register(&memory->dev);
 	if (ret) {
diff --git a/include/linux/device.h b/include/linux/device.h
index 8132aab17e04..65fc6c566cc6 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -485,6 +485,8 @@ struct device_physical_location {
  *		architecture supports non-coherent devices.
  * @DEV_FLAG_OF_NODE_REUSED: Set if the device-tree node is shared with an
  *		ancestor device.
+ * @DEV_FLAG_OFFLINE_DISABLED: If set, the device is permanently online.
+ * @DEV_FLAG_OFFLINE: Set after successful invocation of bus type's .offline().
  */
 enum struct_device_flags {
 	DEV_FLAG_READY_TO_PROBE = 0,
@@ -495,6 +497,8 @@ enum struct_device_flags {
 	DEV_FLAG_STATE_SYNCED = 5,
 	DEV_FLAG_DMA_COHERENT = 6,
 	DEV_FLAG_OF_NODE_REUSED = 7,
+	DEV_FLAG_OFFLINE_DISABLED = 8,
+	DEV_FLAG_OFFLINE = 9,
 
 	DEV_FLAG_COUNT
 };
@@ -573,9 +577,6 @@ enum struct_device_flags {
  * @removable:  Whether the device can be removed from the system. This
  *              should be set by the subsystem / bus driver that discovered
  *              the device.
- *
- * @offline_disabled: If set, the device is permanently online.
- * @offline:	Set after successful invocation of bus type's .offline().
  * @flags:	DEV_FLAG_XXX flags. Use atomic bitfield operations to modify.
  *
  * At the lowest level, every device in a Linux system is represented by an
@@ -680,9 +681,6 @@ struct device {
 
 	enum device_removable	removable;
 
-	bool			offline_disabled:1;
-	bool			offline:1;
-
 	DECLARE_BITMAP(flags, DEV_FLAG_COUNT);
 };
 
@@ -716,6 +714,8 @@ __create_dev_flag_accessors(dma_ops_bypass, DEV_FLAG_DMA_OPS_BYPASS);
 __create_dev_flag_accessors(state_synced, DEV_FLAG_STATE_SYNCED);
 __create_dev_flag_accessors(dma_coherent, DEV_FLAG_DMA_COHERENT);
 __create_dev_flag_accessors(of_node_reused, DEV_FLAG_OF_NODE_REUSED);
+__create_dev_flag_accessors(offline_disabled, DEV_FLAG_OFFLINE_DISABLED);
+__create_dev_flag_accessors(offline, DEV_FLAG_OFFLINE);
 
 /**
  * struct device_link - Device link representation.
diff --git a/kernel/cpu.c b/kernel/cpu.c
index bc4f7a9ba64e..f975bb34915b 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -2639,7 +2639,7 @@ static void cpuhp_offline_cpu_device(unsigned int cpu)
 {
 	struct device *dev = get_cpu_device(cpu);
 
-	dev->offline = true;
+	dev_set_offline(dev);
 	/* Tell user space about the state change */
 	kobject_uevent(&dev->kobj, KOBJ_OFFLINE);
 }
@@ -2648,7 +2648,7 @@ static void cpuhp_online_cpu_device(unsigned int cpu)
 {
 	struct device *dev = get_cpu_device(cpu);
 
-	dev->offline = false;
+	dev_clear_offline(dev);
 	/* Tell user space about the state change */
 	kobject_uevent(&dev->kobj, KOBJ_ONLINE);
 }
-- 
2.53.0.1213.gd9a14994de-goog

Re: [PATCH v4 0/9] driver core: Fix some race conditions

[Rafael J. Wysocki]

On Sat, Apr 4, 2026 at 2:07 AM Douglas Anderson <dianders@chromium.org> wrote:
>
> The main goal of this series is to fix the observed bug talked about
> in the first patch ("driver core: Don't let a device probe until it's
> ready"). That patch fixes a problem that has been observed in the real
> world and could land even if the rest of the patches are found
> unacceptable or need to be spun.
>
> That said, during patch review Danilo correctly pointed out that many
> of the bitfield accesses in "struct device" are unsafe. I added a
> bunch of patches in the series to address each one.
>
> Danilo said he's most worried about "can_match", so I put that one
> first. After that, I tried to transition bitfields to flags in reverse
> order to when the bitfield was added.
>
> Even if transitioning from bitfields to flags isn't truly needed for
> correctness, it seems silly (and wasteful of space in struct device)
> to have some in bitfields and some as flags. Thus I didn't spend time
> for each bitfield showing that it's truly needed for correctness.
>
> Transition was done semi manually. Presumably someone skilled at
> coccinelle could do a better job, but I just used sed in a heavy-
> handed manner and then reviewed/fixed the results, undoing anything my
> script got wrong. My terrible/ugly script was:
>
> var=can_match
> caps="${var^^}"
> for f in $(git grep -l "[>\.]${var}[^1-9_a-zA-Z\[]"); do
>   echo $f
>   sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)->${var} = true/set_bit(DEV_FLAG_${caps}, \&\\1->flags)/" "$f"
>   sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)\.${var} = true/dev_set_${caps}(\&\\1)/" "$f"
>   sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)->${var} = false/clear_bit(DEV_FLAG_${caps}, \&\\1->flags)/" "$f"
>   sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)\.${var} = false/dev_clear_${caps}(\&\\1)/" "$f"
>   sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)->${var} = \([^;]*\)/assign_bit(DEV_FLAG_${caps}, \&\\1->flags, \\2)/" "$f"
>   sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)\.${var} = \([^;]*\)/dev_assign_${caps}(\&\\1, \\2)/" "$f"
>   sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)->${var}\([^1-9_a-zA-Z\[]\)/test_bit(DEV_FLAG_${caps}, \&\\1->flags)\\2/" "$f"
>   sed -i~ -e "s/\([a-zA-Z_0-9\.>()-][a-zA-Z_0-9\.>()-]*\)\.${var}\([^1-9_a-zA-Z\[]\)/dev_${caps}(\&\\1)\\2/" "$f"
> done
>
> From v3 to v4, I transitioned to accessor functions with another ugly
> sed script. I had git format the old patches, then transformed them
> with:
>
> for f in *.patch; do
>   echo $f
>   sed -i~ -e "s/test_and_set_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags)/dev_test_and_set_\\L\\1(\\2)/" "$f"
>   sed -i~ -e "s/test_and_set_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags)/dev_test_and_set_\\L\\1(\\2)/" "$f"
>   sed -i~ -e "s/test_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags)/dev_\\L\\1(\\2)/" "$f"
>   sed -i~ -e "s/test_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags)/dev_\\L\\1(\\2)/" "$f"
>   sed -i~ -e "s/set_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags)/dev_set_\\L\\1(\\2)/" "$f"
>   sed -i~ -e "s/set_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags)/dev_set_\\L\\1(\\2)/" "$f"
>   sed -i~ -e "s/clear_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags)/dev_clear_\\L\\1(\\2)/" "$f"
>   sed -i~ -e "s/clear_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags)/dev_clear_\\L\\1(\\2)/" "$f"
>   sed -i~ -e "s/assign_bit(DEV_FLAG_\([^,]*\), \&\(.*\)->flags, \(.*\))/dev_assign_\\L\\1(\\2, \\3)/" "$f"
>   sed -i~ -e "s/assign_bit(DEV_FLAG_\([^,]*\), \(.*\)\.flags, \(.*\))/dev_assign_\\L\\1(\\2, \\3)/" "$f"
> done
>
> ...and then did a few manual touchups for spacing.
>
> NOTE: one potentially "controversial" choice I made in some patches
> was to always reserve a flag ID even if a flag is only used under
> certain CONFIG_ settings. This is a change from how things were
> before. Keeping the numbering consistent and allowing easy
> compile-testing of both CONFIG settings seemed worth it, especially
> since it won't take up any extra space until we've added a lot more
> flags.
>
> I only marked the first patch as a "Fix" since it is the only one
> fixing observed problems. Other patches could be considered fixes too
> if folks want.
>
> I tested the first patch in the series backported to kernel 6.6 on the
> Pixel phone that was experiencing the race. I added extra printouts to
> make sure that the problem was hitting / addressed. The rest of the
> patches are tested with allmodconfig with arm32, arm64, ppc, and
> x86. I boot tested on an arm64 Chromebook running mainline.
>
> Changes in v4:
> - Use accessor functions for flags
>
> Changes in v3:
> - Use a new "flags" bitfield
> - Add missing \n in probe error message
>
> Changes in v2:
> - Instead of adjusting the ordering, use "ready_to_probe" flag
>
> Douglas Anderson (9):
>   driver core: Don't let a device probe until it's ready
>   driver core: Replace dev->can_match with dev_can_match()
>   driver core: Replace dev->dma_iommu with dev_dma_iommu()
>   driver core: Replace dev->dma_skip_sync with dev_dma_skip_sync()
>   driver core: Replace dev->dma_ops_bypass with dev_dma_ops_bypass()
>   driver core: Replace dev->state_synced with dev_state_synced()
>   driver core: Replace dev->dma_coherent with dev_dma_coherent()
>   driver core: Replace dev->of_node_reused with dev_of_node_reused()
>   driver core: Replace dev->offline + ->offline_disabled with accessors
>
>  arch/arc/mm/dma.c                             |   4 +-
>  arch/arm/mach-highbank/highbank.c             |   2 +-
>  arch/arm/mach-mvebu/coherency.c               |   2 +-
>  arch/arm/mm/dma-mapping-nommu.c               |   4 +-
>  arch/arm/mm/dma-mapping.c                     |  28 ++--
>  arch/arm64/kernel/cpufeature.c                |   2 +-
>  arch/arm64/mm/dma-mapping.c                   |   2 +-
>  arch/mips/mm/dma-noncoherent.c                |   2 +-
>  arch/powerpc/kernel/dma-iommu.c               |   8 +-
>  .../platforms/pseries/hotplug-memory.c        |   4 +-
>  arch/riscv/mm/dma-noncoherent.c               |   2 +-
>  drivers/acpi/scan.c                           |   2 +-
>  drivers/base/core.c                           |  53 +++++---
>  drivers/base/cpu.c                            |   4 +-
>  drivers/base/dd.c                             |  28 ++--
>  drivers/base/memory.c                         |   2 +-
>  drivers/base/pinctrl.c                        |   2 +-
>  drivers/base/platform.c                       |   2 +-
>  drivers/dma/ti/k3-udma-glue.c                 |   6 +-
>  drivers/dma/ti/k3-udma.c                      |   6 +-
>  drivers/iommu/dma-iommu.c                     |   9 +-
>  drivers/iommu/iommu.c                         |   5 +-
>  drivers/net/pcs/pcs-xpcs-plat.c               |   2 +-
>  drivers/of/device.c                           |   6 +-
>  drivers/pci/of.c                              |   2 +-
>  drivers/pci/pwrctrl/core.c                    |   2 +-
>  drivers/regulator/bq257xx-regulator.c         |   2 +-
>  drivers/regulator/rk808-regulator.c           |   2 +-
>  drivers/tty/serial/serial_base_bus.c          |   2 +-
>  drivers/usb/gadget/udc/aspeed-vhub/dev.c      |   2 +-
>  include/linux/device.h                        | 120 ++++++++++++------
>  include/linux/dma-map-ops.h                   |   6 +-
>  include/linux/dma-mapping.h                   |   2 +-
>  include/linux/iommu-dma.h                     |   3 +-
>  kernel/cpu.c                                  |   4 +-
>  kernel/dma/mapping.c                          |  12 +-
>  mm/hmm.c                                      |   2 +-
>  37 files changed, 206 insertions(+), 142 deletions(-)
>
> --

For the whole set

Reviewed-by: Rafael J. Wysocki (Intel) <rafael@kernel.org>

Re: [PATCH v4 0/9] driver core: Fix some race conditions

[Greg Kroah-Hartman]

On Fri, Apr 03, 2026 at 05:04:54PM -0700, Douglas Anderson wrote:
> NOTE: one potentially "controversial" choice I made in some patches
> was to always reserve a flag ID even if a flag is only used under
> certain CONFIG_ settings. This is a change from how things were
> before. Keeping the numbering consistent and allowing easy
> compile-testing of both CONFIG settings seemed worth it, especially
> since it won't take up any extra space until we've added a lot more
> flags.

Nah, this is fine, I don't see any problems with this as the original
code kind of was doing the same thing with the "hole" in the structure
if those options were not enabled.

> I only marked the first patch as a "Fix" since it is the only one
> fixing observed problems. Other patches could be considered fixes too
> if folks want.
> 
> I tested the first patch in the series backported to kernel 6.6 on the
> Pixel phone that was experiencing the race. I added extra printouts to
> make sure that the problem was hitting / addressed. The rest of the
> patches are tested with allmodconfig with arm32, arm64, ppc, and
> x86. I boot tested on an arm64 Chromebook running mainline.

I'm guessing your tests passed?  :)

Anyway, this looks great, unless there are any objections, other than
the "needs to be undefined", which a follow-on patch can handle, I'll
queue them up next week for 7.1-rc1.

thanks,

greg k-h

Re: [PATCH v4 0/9] driver core: Fix some race conditions

[Danilo Krummrich]

On Sun Apr 5, 2026 at 7:27 AM CEST, Greg Kroah-Hartman wrote:
> Anyway, this looks great, unless there are any objections, other than
> the "needs to be undefined", which a follow-on patch can handle, I'll
> queue them up next week for 7.1-rc1.

Sounds good, for the series:

Reviewed-by: Danilo Krummrich <dakr@kernel.org>

Re: [PATCH v4 0/9] driver core: Fix some race conditions

[Doug Anderson]

Hi,

On Sat, Apr 4, 2026 at 10:28 PM Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
>
> On Fri, Apr 03, 2026 at 05:04:54PM -0700, Douglas Anderson wrote:
> > NOTE: one potentially "controversial" choice I made in some patches
> > was to always reserve a flag ID even if a flag is only used under
> > certain CONFIG_ settings. This is a change from how things were
> > before. Keeping the numbering consistent and allowing easy
> > compile-testing of both CONFIG settings seemed worth it, especially
> > since it won't take up any extra space until we've added a lot more
> > flags.
>
> Nah, this is fine, I don't see any problems with this as the original
> code kind of was doing the same thing with the "hole" in the structure
> if those options were not enabled.
>
> > I only marked the first patch as a "Fix" since it is the only one
> > fixing observed problems. Other patches could be considered fixes too
> > if folks want.
> >
> > I tested the first patch in the series backported to kernel 6.6 on the
> > Pixel phone that was experiencing the race. I added extra printouts to
> > make sure that the problem was hitting / addressed. The rest of the
> > patches are tested with allmodconfig with arm32, arm64, ppc, and
> > x86. I boot tested on an arm64 Chromebook running mainline.
>
> I'm guessing your tests passed?  :)

Yup, all the tests that I've run have passed. I also threw in an
"allnoconfig" compile test just for good measure.


> Anyway, this looks great, unless there are any objections, other than
> the "needs to be undefined", which a follow-on patch can handle, I'll
> queue them up next week for 7.1-rc1.

Thanks. As per the other thread, I'm happy if you or Danilo want to
apply it, and I'm happy if you want to make minor fixups when
applying.

When I see the patches applied, I'll send a followup patch to address
the "needs to be undefined" comment, unless Danilo makes that change
himself when applying.

-Doug