Skip to contentSkip to conversation

PCI/proc: Fix race between pci_proc_init() and pci_bus_add_device()

PATCH Applied Krzysztof Wilczyński 1 participant 3 messages Jun 6 – Jun 11 PCI

v3 of 4 Applied Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 PCI/proc: Fix race between pci_proc_init() and pci_bus_add_device()

pci_proc_attach_device() creates procfs entries for PCI devices and is called from pci_bus_add_device(). It lazily creates the per-bus procfs directory (bus->procdir) via proc_mkdir() on first use, and returns early if proc_initialized is not yet set.

On x86 with ACPI, PCI enumeration occurs at subsys_initcall, before pci_proc_init() sets proc_initialized at device_initcall. The for_each_pci_dev() loop in pci_proc_init() then creates procfs entries for these already-enumerated devices, but runs without holding pci_rescan_remove_lock.

On ARM64 with devicetree, PCI host bridges probe at device_initcall. With async probing enabled, pci_bus_add_device() can run concurrently with pci_proc_init(), and both may call pci_proc_attach_device() for the same device or for different devices on the same bus. As pci_host_probe() holds pci_rescan_remove_lock while pci_proc_init() does not, there is no serialisation between the two paths.

When two threads concurrently call pci_proc_attach_device() for devices on the same bus, both observe bus->procdir as NULL and both call proc_mkdir(). The proc filesystem serialises directory creation internally, so only one caller succeeds. The other receives NULL (duplicate entry) and unconditionally stores it to bus->procdir, corrupting the valid pointer set by the first caller.

Thus, extract the bus procfs directory creation from pci_proc_attach_device() into a new pci_proc_attach_bus() function, and call it from the two bus creation paths: pci_register_host_bridge() for root buses and pci_alloc_child_bus() for child buses. These are the only two callers of pci_alloc_bus(), so for buses created after proc_initialized is set, bus->procdir is in place before any device can be added to the bus.

Therefore, by the time pci_proc_attach_device() runs on these buses, bus->procdir is already set and the racy proc_mkdir() call is never reached.

For buses created before pci_proc_init() sets proc_initialized (the common x86 ACPI case), the bus creation hooks return early. As such, add a fallback call to pci_proc_attach_bus() from pci_proc_attach_device() to handle these pre-init buses.

Additionally, wrap the for_each_pci_dev() loop in pci_proc_init() with pci_lock_rescan_remove() to serialise against concurrent PCI bus operations, add an early return in pci_proc_attach_device() when dev->procent is already set to make the function idempotent, and clear bus->procdir in pci_proc_detach_bus() to prevent use of a dangling pointer after proc_remove().

---
Changes in v3:
https://lore.kernel.org/linux-pci/[email protected]/

- Used local variables in pci_proc_attach_bus() and pci_proc_attach_device() to ensure that bus->procdir and dev->procent are only assigned on success. If either the proc_mkdir() or proc_create_data() returns NULL, the struct field will not see a new assignment, preventing a valid pointer from being overwritten. This was reported by Sashiko, see:
https://sashiko.dev/#/patchset/20260606170146.673140-1-kwilczynski%40kernel.org?part=1

Changes in v2:
https://lore.kernel.org/linux-pci/[email protected]/

- Extracted bus->procdir creation into new pci_proc_attach_bus() function, addressing the suggestion to move proc_mkdir() to a bus creation path, as per Bjorn Helgaas' feedback.
- Added bus->procdir = NULL in pci_proc_detach_bus() for symmetry and to prevent dangling pointer after proc_remove().

 drivers/pci/pci.h   |  6 ++--
 drivers/pci/probe.c |  6 ++++
 drivers/pci/proc.c  | 68 ++++++++++++++++++++++++++++++++-------------
 3 files changed, 58 insertions(+), 22 deletions(-)
drivers/pci/pci.h +4 -2
index 13d998fbacce..5e8520bb229c 100644
@@ -343,13 +343,15 @@ void pci_allocate_vc_save_buffers(struct pci_dev *dev);
 
 /* PCI /proc functions */
 #ifdef CONFIG_PROC_FS
+int pci_proc_attach_bus(struct pci_bus *bus);
+int pci_proc_detach_bus(struct pci_bus *bus);
 int pci_proc_attach_device(struct pci_dev *dev);
 int pci_proc_detach_device(struct pci_dev *dev);
-int pci_proc_detach_bus(struct pci_bus *bus);
 #else
+static inline int pci_proc_attach_bus(struct pci_bus *bus) { return 0; }
+static inline int pci_proc_detach_bus(struct pci_bus *bus) { return 0; }
 static inline int pci_proc_attach_device(struct pci_dev *dev) { return 0; }
 static inline int pci_proc_detach_device(struct pci_dev *dev) { return 0; }
-static inline int pci_proc_detach_bus(struct pci_bus *bus) { return 0; }
 #endif
 
 /* Functions for PCI Hotplug drivers to use */
drivers/pci/probe.c +6 -0
index bccc7a4bdd79..90248813d2cf 100644
@@ -1071,6 +1071,9 @@ static int pci_register_host_bridge(struct pci_host_bridge *bridge)
 	/* Create legacy_io and legacy_mem files for this bus */
 	pci_create_legacy_files(bus);
 
+	/* Create procfs directory for this bus */
+	pci_proc_attach_bus(bus);
+
 	if (parent)
 		dev_info(parent, "PCI host bridge to bus %s\n", name);
 	else
@@ -1279,6 +1282,9 @@ static struct pci_bus *pci_alloc_child_bus(struct pci_bus *parent,
 	/* Create legacy_io and legacy_mem files for this bus */
 	pci_create_legacy_files(child);
 
+	/* Create procfs directory for this bus */
+	pci_proc_attach_bus(child);
+
 	return child;
 }
 
drivers/pci/proc.c +48 -20

100 lines not on this page · raw message

Signature · 1 line
2.54.0
author Applied
#1 lore raw

Hello,

Quoted 42 lines

pci_proc_attach_device() creates procfs entries for PCI devices and is called from pci_bus_add_device(). It lazily creates the per-bus procfs directory (bus->procdir) via proc_mkdir() on first use, and returns early if proc_initialized is not yet set.

On x86 with ACPI, PCI enumeration occurs at subsys_initcall, before pci_proc_init() sets proc_initialized at device_initcall. The for_each_pci_dev() loop in pci_proc_init() then creates procfs entries for these already-enumerated devices, but runs without holding pci_rescan_remove_lock.

On ARM64 with devicetree, PCI host bridges probe at device_initcall. With async probing enabled, pci_bus_add_device() can run concurrently with pci_proc_init(), and both may call pci_proc_attach_device() for the same device or for different devices on the same bus. As pci_host_probe() holds pci_rescan_remove_lock while pci_proc_init() does not, there is no serialisation between the two paths.

When two threads concurrently call pci_proc_attach_device() for devices on the same bus, both observe bus->procdir as NULL and both call proc_mkdir(). The proc filesystem serialises directory creation internally, so only one caller succeeds. The other receives NULL (duplicate entry) and unconditionally stores it to bus->procdir, corrupting the valid pointer set by the first caller.

Thus, extract the bus procfs directory creation from pci_proc_attach_device() into a new pci_proc_attach_bus() function, and call it from the two bus creation paths: pci_register_host_bridge() for root buses and pci_alloc_child_bus() for child buses. These are the only two callers of pci_alloc_bus(), so for buses created after proc_initialized is set, bus->procdir is in place before any device can be added to the bus.

Therefore, by the time pci_proc_attach_device() runs on these buses, bus->procdir is already set and the racy proc_mkdir() call is never reached.

For buses created before pci_proc_init() sets proc_initialized (the common x86 ACPI case), the bus creation hooks return early. As such, add a fallback call to pci_proc_attach_bus() from pci_proc_attach_device() to handle these pre-init buses.

Quoted 8 lines

Additionally, wrap the for_each_pci_dev() loop in pci_proc_init() with pci_lock_rescan_remove() to serialise against concurrent PCI bus operations, add an early return in pci_proc_attach_device() when dev->procent is already set to make the function idempotent, and clear bus->procdir in pci_proc_detach_bus() to prevent use of a dangling pointer after proc_remove().

Applied to procfs, as I want this to have some soak time via linux-next, plus 0-day bot will test it for me, too.

Lorenzo, do you think you could re-test this on your hardware platform?

Shuan, any chance for you to test this again on your RISC-V based platform?

Thank you!

Krzysztof

#2 lore raw

Hello,

[...]

Quoted 24 lines
Quoted 24 lines

Thus, extract the bus procfs directory creation from pci_proc_attach_device() into a new pci_proc_attach_bus() function, and call it from the two bus creation paths: pci_register_host_bridge() for root buses and pci_alloc_child_bus() for child buses. These are the only two callers of pci_alloc_bus(), so for buses created after proc_initialized is set, bus->procdir is in place before any device can be added to the bus.

Therefore, by the time pci_proc_attach_device() runs on these buses, bus->procdir is already set and the racy proc_mkdir() call is never reached.

For buses created before pci_proc_init() sets proc_initialized (the common x86 ACPI case), the bus creation hooks return early. As such, add a fallback call to pci_proc_attach_bus() from pci_proc_attach_device() to handle these pre-init buses.

Additionally, wrap the for_each_pci_dev() loop in pci_proc_init() with pci_lock_rescan_remove() to serialise against concurrent PCI bus operations, add an early return in pci_proc_attach_device() when dev->procent is already set to make the function idempotent, and clear bus->procdir in pci_proc_detach_bus() to prevent use of a dangling pointer after proc_remove().

Quoted 4 lines
Quoted 1 line

Applied to procfs, as I want this to have some soak time via linux-next, plus 0-day bot will test it for me, too.

I will be sending a v4 shortly, which aims to solve this problem with a different approach as moving procfs directories to the bus creation time has an unwelcome side effect of leaving empty directories for post-init creation time, such as remove/rescan or when devices are added via a hotplug, etc.

Thank you!

Krzysztof