Unrated severityNVD Advisory· Published May 27, 2026· Updated May 27, 2026
CVE-2026-45904
CVE-2026-45904
Description
In the Linux kernel, the following vulnerability has been resolved:
powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
The recent commit 1010b4c012b0 ("powerpc/eeh: Make EEH driver device hotplug safe") restructured the EEH driver to improve synchronization with the PCI hotplug layer.
However, it inadvertently moved pci_lock_rescan_remove() outside its intended scope in eeh_handle_normal_event(), leading to broken PCI error reporting and improper EEH event triggering. Specifically, eeh_handle_normal_event() acquired pci_lock_rescan_remove() before calling eeh_pe_bus_get(), but eeh_pe_bus_get() itself attempts to acquire the same lock internally, causing nested locking and disrupting normal EEH event handling paths.
This patch adds a boolean parameter do_lock to _eeh_pe_bus_get(), with two public wrappers: eeh_pe_bus_get() with locking enabled. eeh_pe_bus_get_nolock() that skips locking.
Callers that already hold pci_lock_rescan_remove() now use eeh_pe_bus_get_nolock() to avoid recursive lock acquisition.
Additionally, pci_lock_rescan_remove() calls are restored to the correct position—after eeh_pe_bus_get() and immediately before iterating affected PEs and devices. This ensures EEH-triggered PCI removes occur under proper bus rescan locking without recursive lock contention.
The eeh_pe_loc_get() function has been split into two functions: eeh_pe_loc_get(struct eeh_pe *pe) which retrieves the loc for given PE. eeh_pe_loc_get_bus(struct pci_bus *bus) which retrieves the location code for given bus.
This resolves lockdep warnings such as:
[ 84.964298] [ T928] ============================================ [ 84.964304] [ T928] WARNING: possible recursive locking detected [ 84.964311] [ T928] 6.18.0-rc3 #51 Not tainted [ 84.964315] [ T928] -------------------------------------------- [ 84.964320] [ T928] eehd/928 is trying to acquire lock: [ 84.964324] [ T928] c000000003b29d58 (pci_rescan_remove_lock){+.+.}-{3:3}, at: pci_lock_rescan_remove+0x28/0x40 [ 84.964342] [ T928] but task is already holding lock: [ 84.964347] [ T928] c000000003b29d58 (pci_rescan_remove_lock){+.+.}-{3:3}, at: pci_lock_rescan_remove+0x28/0x40 [ 84.964357] [ T928] other info that might help us debug this: [ 84.964363] [ T928] Possible unsafe locking scenario:
[ 84.964367] [ T928] CPU0 [ 84.964370] [ T928] ---- [ 84.964373] [ T928] lock(pci_rescan_remove_lock); [ 84.964378] [ T928] lock(pci_rescan_remove_lock); [ 84.964383] [ T928] * DEADLOCK *
[ 84.964388] [ T928] May be due to missing lock nesting notation
[ 84.964393] [ T928] 1 lock held by eehd/928: [ 84.964397] [ T928] #0: c000000003b29d58 (pci_rescan_remove_lock){+.+.}-{3:3}, at: pci_lock_rescan_remove+0x28/0x40 [ 84.964408] [ T928] stack backtrace: [ 84.964414] [ T928] CPU: 2 UID: 0 PID: 928 Comm: eehd Not tainted 6.18.0-rc3 #51 VOLUNTARY [ 84.964417] [ T928] Hardware name: IBM,9080-HEX POWER10 (architected) 0x800200 0xf000006 of:IBM,FW1060.00 (NH1060_022) hv:phyp pSeries [ 84.964419] [ T928] Call Trace: [ 84.964420] [ T928] [c0000011a7157990] [c000000001705de4] dump_stack_lvl+0xc8/0x130 (unreliable) [ 84.964424] [ T928] [c0000011a71579d0] [c0000000002f66e0] print_deadlock_bug+0x430/0x440 [ 84.964428] [ T928] [c0000011a7157a70] [c0000000002fd0c0] __lock_acquire+0x1530/0x2d80 [ 84.964431] [ T928] [c0000011a7157ba0] [c0000000002fea54] lock_acquire+0x144/0x410 [ 84.964433] [ T928] [c0000011a7157cb0] [c0000011a7157cb0] __mutex_lock+0xf4/0x1050 [ 84.964436] [ T928] [c0000011a7157e00] [c000000000de21d8] pci_lock_rescan_remove+0x28/0x40 [ 84.964439] [ T928] [c0000011a7157e20] [c00000000004ed98] eeh_pe_bus_get+0x48/0xc0 [ 84.964442] [ T928] [c0000011a7157e50] [c00000 ---truncated---
AI Insight
LLM-synthesized narrative grounded in this CVE's description and references.
A recursive locking issue in the Linux kernel's EEH driver for PowerPC causes lockdep warnings and potential deadlock during PCI error handling, fixed by adjusting lock acquisition scope.
Vulnerability
In the Linux kernel, commit 1010b4c012b0 ("powerpc/eeh: Make EEH driver device hotplug safe") restructured the EEH (Enhanced Error Handling) driver for PowerPC but inadvertently moved pci_lock_rescan_remove() outside its intended scope in eeh_handle_normal_event(). This caused eeh_pe_bus_get() to attempt acquiring the same lock internally, leading to recursive locking. The bug affects kernels containing that commit (e.g., 6.18-rc3 and later). [1][2]
Exploitation
An attacker who can trigger a PCI error (e.g., via physical access or by inducing a hardware fault) can cause an EEH event that hits the recursive lock path. This results in lockdep warnings and may lead to a deadlock, preventing proper PCI error recovery and potentially hanging the system. No special privileges beyond the ability to trigger PCI errors are required.
Impact
The recursive lock acquisition disrupts normal EEH event handling, breaking PCI error reporting and recovery. The primary impact is denial of service: the system may become unresponsive or fail to recover from PCI errors, leading to data loss or system instability. No privilege escalation or information disclosure is described.
Mitigation
The fix introduces a boolean parameter do_lock to _eeh_pe_bus_get() and provides two wrappers: eeh_pe_bus_get() (with locking) and eeh_pe_bus_get_nolock() (without). Callers that already hold pci_lock_rescan_remove() now use the nolock variant. Additionally, pci_lock_rescan_remove() calls are restored to the correct position after eeh_pe_bus_get(). The fix is included in kernel commits [1] and [2]; users should update to a kernel version containing these patches.
AI Insight generated on May 27, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.
Affected products
1Patches
16815a8d2feb56powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.gitNarayana Murty NDec 10, 2025Fixed in 7.0via kernel-cna
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
788dd28fd496powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.gitNarayana Murty NDec 10, 2025Fixed in 5.15.202via kernel-cna
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index b1a5bba2e0b949..2c9dc733accfdf 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index b1a5bba2e0b949..2c9dc733accfdf 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 2f13d906e1fcb3..20106e490b9c68 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -847,7 +847,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -878,14 +878,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1093,7 +1094,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1217,7 +1218,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 2f13d906e1fcb3..20106e490b9c68 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -847,7 +847,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -878,14 +878,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1093,7 +1094,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1217,7 +1218,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index fea58e9546f981..6f69242142e0dd 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -811,6 +811,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -837,8 +855,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -846,7 +865,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -861,11 +880,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index fea58e9546f981..6f69242142e0dd 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -811,6 +811,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -837,8 +855,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -846,7 +865,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -861,11 +880,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
f49faa4a64f8powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.gitNarayana Murty NDec 10, 2025Fixed in 6.1.165via kernel-cna
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 514dd056c2c84b..b5709b9aed238f 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 514dd056c2c84b..b5709b9aed238f 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 9a761f46c7b23c..70251c8c365ef7 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -877,14 +877,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1089,7 +1090,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1213,7 +1214,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 9a761f46c7b23c..70251c8c365ef7 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -877,14 +877,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1089,7 +1090,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1213,7 +1214,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index 08095aeba5c983..b10fcca5204003 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index 08095aeba5c983..b10fcca5204003 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
87a1f93986aapowerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.gitNarayana Murty NDec 10, 2025Fixed in 6.6.128via kernel-cna
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 514dd056c2c84b..b5709b9aed238f 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 514dd056c2c84b..b5709b9aed238f 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index cc8bedf410ea70..632bf157636ec4 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -877,14 +877,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1089,7 +1090,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1213,7 +1214,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index cc8bedf410ea70..632bf157636ec4 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -877,14 +877,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1089,7 +1090,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1213,7 +1214,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index 08095aeba5c983..b10fcca5204003 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index 08095aeba5c983..b10fcca5204003 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
6e6561231c6cpowerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.gitNarayana Murty NDec 10, 2025Fixed in 6.18.14via kernel-cna
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
89810e2d8028powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.gitNarayana Murty NDec 10, 2025Fixed in 5.10.252via kernel-cna
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index b1a5bba2e0b949..2c9dc733accfdf 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index b1a5bba2e0b949..2c9dc733accfdf 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 2f13d906e1fcb3..20106e490b9c68 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -847,7 +847,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -878,14 +878,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1093,7 +1094,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1217,7 +1218,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 2f13d906e1fcb3..20106e490b9c68 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -847,7 +847,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -878,14 +878,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1093,7 +1094,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1217,7 +1218,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index fea58e9546f981..6f69242142e0dd 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -811,6 +811,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -837,8 +855,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -846,7 +865,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -861,11 +880,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index fea58e9546f981..6f69242142e0dd 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -811,6 +811,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -837,8 +855,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -846,7 +865,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -861,11 +880,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
f8b16d5764eepowerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.gitNarayana Murty NDec 10, 2025Fixed in 6.12.75via kernel-cna
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index c73e4225e84a56..51a6d881c22929 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index c73e4225e84a56..51a6d881c22929 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
b85ee287bfe5powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.gitNarayana Murty NDec 10, 2025Fixed in 6.19.4via kernel-cna
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
87a1f93986aapowerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.gitNarayana Murty NDec 10, 2025via nvd-ref
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 514dd056c2c84b..b5709b9aed238f 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 514dd056c2c84b..b5709b9aed238f 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index cc8bedf410ea70..632bf157636ec4 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -877,14 +877,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1089,7 +1090,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1213,7 +1214,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index cc8bedf410ea70..632bf157636ec4 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -877,14 +877,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1089,7 +1090,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1213,7 +1214,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index 08095aeba5c983..b10fcca5204003 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index 08095aeba5c983..b10fcca5204003 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
f49faa4a64f8powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.gitNarayana Murty NDec 10, 2025via nvd-ref
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 514dd056c2c84b..b5709b9aed238f 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 514dd056c2c84b..b5709b9aed238f 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 9a761f46c7b23c..70251c8c365ef7 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -877,14 +877,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1089,7 +1090,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1213,7 +1214,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 9a761f46c7b23c..70251c8c365ef7 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -877,14 +877,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1089,7 +1090,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1213,7 +1214,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index 08095aeba5c983..b10fcca5204003 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index 08095aeba5c983..b10fcca5204003 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
b85ee287bfe5powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.gitNarayana Murty NDec 10, 2025via nvd-ref
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
f8b16d5764eepowerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.gitNarayana Murty NDec 10, 2025via nvd-ref
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index c73e4225e84a56..51a6d881c22929 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index c73e4225e84a56..51a6d881c22929 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
788dd28fd496powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.gitNarayana Murty NDec 10, 2025via nvd-ref
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index b1a5bba2e0b949..2c9dc733accfdf 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index b1a5bba2e0b949..2c9dc733accfdf 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 2f13d906e1fcb3..20106e490b9c68 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -847,7 +847,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -878,14 +878,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1093,7 +1094,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1217,7 +1218,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 2f13d906e1fcb3..20106e490b9c68 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -847,7 +847,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -878,14 +878,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1093,7 +1094,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1217,7 +1218,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index fea58e9546f981..6f69242142e0dd 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -811,6 +811,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -837,8 +855,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -846,7 +865,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -861,11 +880,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index fea58e9546f981..6f69242142e0dd 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -811,6 +811,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -837,8 +855,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -846,7 +865,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -861,11 +880,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
6e6561231c6cpowerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.gitNarayana Murty NDec 10, 2025via nvd-ref
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
89810e2d8028powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.gitNarayana Murty NDec 10, 2025via nvd-ref
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index b1a5bba2e0b949..2c9dc733accfdf 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index b1a5bba2e0b949..2c9dc733accfdf 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 2f13d906e1fcb3..20106e490b9c68 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -847,7 +847,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -878,14 +878,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1093,7 +1094,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1217,7 +1218,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 2f13d906e1fcb3..20106e490b9c68 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -847,7 +847,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -878,14 +878,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1093,7 +1094,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1217,7 +1218,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index fea58e9546f981..6f69242142e0dd 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -811,6 +811,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -837,8 +855,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -846,7 +865,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -861,11 +880,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index fea58e9546f981..6f69242142e0dd 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -811,6 +811,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -837,8 +855,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -846,7 +865,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -861,11 +880,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
815a8d2feb56powerpc/eeh: fix recursive pci_lock_rescan_remove locking in EEH event handling
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.gitNarayana Murty NDec 10, 2025via nvd-ref
6 files changed · +156 −20
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/include/asm/eeh.h+2 −0 modifieddiff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h index 5e34611de9ef40..b7ebb4ac2c7103 100644 --- a/arch/powerpc/include/asm/eeh.h +++ b/arch/powerpc/include/asm/eeh.h @@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root, void eeh_pe_restore_bars(struct eeh_pe *pe); const char *eeh_pe_loc_get(struct eeh_pe *pe); struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe); +const char *eeh_pe_loc_get_bus(struct pci_bus *bus); +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe); void eeh_show_enabled(void); int __init eeh_init(struct eeh_ops *ops);
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_driver.c+6 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index ef78ff77cf8f21..028f6915853234 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe) pci_lock_rescan_remove(); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (!bus) { pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n", __func__, pe->phb->global_number, pe->addr); @@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe) /* Log the event */ if (pe->type & EEH_PE_PHB) { pr_err("EEH: Recovering PHB#%x, location: %s\n", - pe->phb->global_number, eeh_pe_loc_get(pe)); + pe->phb->global_number, eeh_pe_loc_get_bus(bus)); } else { struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb); pr_err("EEH: Recovering PHB#%x-PE#%x\n", pe->phb->global_number, pe->addr); pr_err("EEH: PE location: %s, PHB location: %s\n", - eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe)); + eeh_pe_loc_get_bus(bus), + eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe))); } #ifdef CONFIG_STACKTRACE @@ -1098,7 +1099,7 @@ recover_failed: eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true); eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED); - bus = eeh_pe_bus_get(pe); + bus = eeh_pe_bus_get_nolock(pe); if (bus) pci_hp_remove_devices(bus); else @@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void) (phb_pe->state & EEH_PE_RECOVERING)) continue; - bus = eeh_pe_bus_get(phb_pe); + bus = eeh_pe_bus_get_nolock(phb_pe); if (!bus) { pr_err("%s: Cannot find PCI bus for " "PHB#%x-PE#%x\n",
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
arch/powerpc/kernel/eeh_pe.c+70 −5 modifieddiff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c index e740101fadf3b1..040e8f69a4aa86 100644 --- a/arch/powerpc/kernel/eeh_pe.c +++ b/arch/powerpc/kernel/eeh_pe.c @@ -812,6 +812,24 @@ void eeh_pe_restore_bars(struct eeh_pe *pe) const char *eeh_pe_loc_get(struct eeh_pe *pe) { struct pci_bus *bus = eeh_pe_bus_get(pe); + return eeh_pe_loc_get_bus(bus); +} + +/** + * eeh_pe_loc_get_bus - Retrieve location code binding to the given PCI bus + * @bus: PCI bus + * + * Retrieve the location code associated with the given PCI bus. If the bus + * is a root bus, the location code is fetched from the PHB device tree node + * or root port. Otherwise, the location code is obtained from the device + * tree node of the upstream bridge of the bus. The function walks up the + * bus hierarchy if necessary, checking each node for the appropriate + * location code property ("ibm,io-base-loc-code" for root buses, + * "ibm,slot-location-code" for others). If no location code is found, + * returns "N/A". + */ +const char *eeh_pe_loc_get_bus(struct pci_bus *bus) +{ struct device_node *dn; const char *loc = NULL; @@ -838,8 +856,9 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) } /** - * eeh_pe_bus_get - Retrieve PCI bus according to the given PE + * _eeh_pe_bus_get - Retrieve PCI bus according to the given PE * @pe: EEH PE + * @do_lock: Is the caller already held the pci_lock_rescan_remove? * * Retrieve the PCI bus according to the given PE. Basically, * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the @@ -847,7 +866,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe) * returned for BUS PE. However, we don't have associated PCI * bus for DEVICE PE. */ -struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +static struct pci_bus *_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock) { struct eeh_dev *edev; struct pci_dev *pdev; @@ -862,11 +881,58 @@ struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) /* Retrieve the parent PCI bus of first (top) PCI device */ edev = list_first_entry_or_null(&pe->edevs, struct eeh_dev, entry); - pci_lock_rescan_remove(); + if (do_lock) + pci_lock_rescan_remove(); pdev = eeh_dev_to_pci_dev(edev); if (pdev) bus = pdev->bus; - pci_unlock_rescan_remove(); + if (do_lock) + pci_unlock_rescan_remove(); return bus; } + +/** + * eeh_pe_bus_get - Retrieve PCI bus associated with the given EEH PE, locking + * if needed + * @pe: Pointer to the EEH PE + * + * This function is a wrapper around _eeh_pe_bus_get(), which retrieves the PCI + * bus associated with the provided EEH PE structure. It acquires the PCI + * rescans lock to ensure safe access to shared data during the retrieval + * process. This function should be used when the caller requires the PCI bus + * while holding the rescan/remove lock, typically during operations that modify + * or inspect PCIe device state in a safe manner. + * + * RETURNS: + * A pointer to the PCI bus associated with the EEH PE, or NULL if none found. + */ + +struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, true); +} + +/** + * eeh_pe_bus_get_nolock - Retrieve PCI bus associated with the given EEH PE + * without locking + * @pe: Pointer to the EEH PE + * + * This function is a variant of _eeh_pe_bus_get() that retrieves the PCI bus + * associated with the specified EEH PE without acquiring the + * pci_lock_rescan_remove lock. It should only be used when the caller can + * guarantee safe access to PE structures without the need for that lock, + * typically in contexts where the lock is already held locking is otherwise + * managed. + * + * RETURNS: + * pointer to the PCI bus associated with the EEH PE, or NULL if none is found. + * + * NOTE: + * Use this function carefully to avoid race conditions and data corruption. + */ + +struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe) +{ + return _eeh_pe_bus_get(pe, false); +} -- cgit 1.3-korg
Vulnerability mechanics
Root cause
"Recursive locking of pci_lock_rescan_remove in EEH event handling due to eeh_pe_bus_get() acquiring the same lock already held by its caller eeh_handle_normal_event()."
Attack vector
An attacker does not directly trigger this bug; it is a kernel self-deadlock that occurs when the EEH (Enhanced Error Handling) daemon processes a PCI error event. When `eeh_handle_normal_event()` acquires `pci_lock_rescan_remove()` and then calls `eeh_pe_bus_get()`, which internally tries to acquire the same lock, a recursive locking attempt occurs. This results in a lockdep warning and disrupts PCI error recovery, potentially causing improper EEH event handling and broken PCI error reporting. The precondition is that a PCI error event must be processed on a system running a kernel with the faulty commit 1010b4c012b0.
Affected code
The bug is in `arch/powerpc/kernel/eeh_driver.c` in `eeh_handle_normal_event()` and `eeh_handle_special_event()`, and in `arch/powerpc/kernel/eeh_pe.c` in `eeh_pe_bus_get()`. The patch modifies `eeh_pe.c` to introduce `_eeh_pe_bus_get()` with a `do_lock` parameter and two wrappers (`eeh_pe_bus_get` and `eeh_pe_bus_get_nolock`), and updates `eeh_driver.c` callers to use the nolock variant. The header `arch/powerpc/include/asm/eeh.h` gains declarations for `eeh_pe_bus_get_nolock` and `eeh_pe_loc_get_bus`.
What the fix does
The patch refactors `eeh_pe_bus_get()` into a private `_eeh_pe_bus_get(struct eeh_pe *pe, bool do_lock)` that conditionally acquires `pci_lock_rescan_remove()` only when `do_lock` is true. Two public wrappers are provided: `eeh_pe_bus_get()` (locking enabled) and `eeh_pe_bus_get_nolock()` (no locking). All callers in `eeh_driver.c` that already hold `pci_lock_rescan_remove()` are switched to `eeh_pe_bus_get_nolock()`, eliminating the recursive lock. Additionally, `eeh_pe_loc_get()` is split into `eeh_pe_loc_get()` and `eeh_pe_loc_get_bus()` so that location-code retrieval can use the already-obtained bus pointer without re-entering the locking path.
Preconditions
- inputA PCI EEH error event must occur on a system running a kernel that includes commit 1010b4c012b0.
- configThe EEH daemon (eehd) must process the event, entering eeh_handle_normal_event().
Generated on May 27, 2026. Inputs: CWE entries + fix-commit diffs from this CVE's patches. Citations validated against bundle.
References
8- git.kernel.org/stable/c/6e6561231c6cfc32c5631aeecc0928ff2b14265cnvd
- git.kernel.org/stable/c/788dd28fd49610d6047cbb15dbf1186afffdfbafnvd
- git.kernel.org/stable/c/815a8d2feb5615ae7f0b5befd206af0b0160614cnvd
- git.kernel.org/stable/c/87a1f93986aa1500b85aeff16b0b71c29ea116eanvd
- git.kernel.org/stable/c/89810e2d80281d42f855fac813786758ee16e323nvd
- git.kernel.org/stable/c/b85ee287bfe52c6b2d9b41758b5e0d08679d5b39nvd
- git.kernel.org/stable/c/f49faa4a64f8ac0e38983e606075b25dfcfc9ad4nvd
- git.kernel.org/stable/c/f8b16d5764ee1e78c1ef333017ad383ffe76fcdcnvd
News mentions
0No linked articles in our index yet.