PCI: Add d3cold/soft reset methods for devices with limited reset capability
v4 cover PCI: Add d3cold/soft reset methods for devices with limited reset capability
Some PCIe devices lack working reset methods for VFIO passthrough scenarios. These devices typically have no FLR, advertise NoSoftRst+ (blocking PM reset), and have broken or unavailable bus reset. When a VM crashes, VFIO cannot reset the device for reuse without a working reset method.
This series addresses the problem by adding two new reset methods and a device-specific quirk:
**Patch 1/3: d3cold reset method**
Adds D3cold as a general reset method with strict _PR3 requirement. Only attempts true power cycling via platform control (ACPI _PR3), never falls back to D3hot. This provides genuine power-off reset on platforms that support it.
**Patch 2/3: soft reset method**
Adds "soft" reset as absolute last resort when all other methods fail. Performs D3hot→D0 transition without checking NoSoftRst flag. Only becomes available when both pci_pm_reset() and pci_d3cold_reset() return -ENOTTY. Provides "better than nothing" reset for devices that would otherwise have no reset capability. Users retain full policy control via reset_methods sysfs.
**Patch 3/3: Qualcomm quirk_no_bus_reset**
Disables broken bus reset for Qualcomm ath11k/ath12k WiFi and SDX62/SDX65 modems. Testing proves this is device-specific: MediaTek MT7925e works correctly with bus reset using the same passive M.2-to-PCIe adapters where Qualcomm devices fail, confirming PERST# is properly wired and the issue is not deployment-specific.
mt7925e needs quirk_no_flr to work
(https://lore.kernel.org/all/[email protected]/)
other cards using this adpaters are using flr method with no problem.
**Changes in v4:**
- Split v3's d3cold method (which fell back to D3hot) into two separate methods based on maintainer feedback (Alex Williamson)
- d3cold now strictly requires _PR3, returns -ENOTTY otherwise
- New "soft" method handles D3hot transition as last resort
- Extracted pci_do_d3hot_transition() helper to avoid code duplication between pci_pm_reset() and pci_soft_reset()
- Updated Qualcomm quirk commit message to reference new architecture
**Testing:**
All testing performed on 2023 desktops with M.2 WiFi cards via passive M.2-to-PCIe adapters in standard PCIe expansion slots (00:1c.x ports).
Evidence for device-specific reset issues:
- Realtek RTL8922AE: 00:1c.1, reset_method="flr bus", FLR works correctly (Proves FLR signaling works through passive adapters)
- MediaTek MT7925e: 00:1c.1, reset_method="bus", bus reset works correctly (Has separate quirk_no_flr patch due to broken FLR, proves PERST# works)
- Qualcomm WCN785x: 00:1c.4, reset_method="bus", bus reset FAILS (Same adapters, same slot type - proves device-specific SBR issue)
All three devices use identical passive M.2-to-PCIe adapters in standard PCIe expansion slots. The fact that FLR works for Realtek and bus reset works for MediaTek but both fail for their respective broken devices confirms these are device-specific hardware issues, not deployment or adapter issues.
D3hot transition testing (documented in detail):
- Command register cleared: 0x0002 → 0x0000 → 0x0506 after driver init
- BARs preserved: 0x84200004 maintained throughout
- Device successfully reusable for VFIO passthrough after VM crash
Reset hierarchy after this series:
1. device_specific
2. acpi
3. flr
4. af_flr
5. pm (unchanged - still checks NoSoftRst)
6. bus
7. cxl_bus
8. d3cold (NEW - strict _PR3 requirement)
9. soft (NEW - D3hot last resort)
Jose Ignacio Tornos Martinez (3):
PCI: Add d3cold as general reset method
PCI: Add soft reset method as last resort
PCI: Disable broken bus reset on Qualcomm devices
--
2.53.0
PATCH v4 1/3 PCI: Add d3cold as general reset method
Add D3cold power cycle as a general PCI reset method for single-function devices on platforms with ACPI _PR3 power resources. This provides true power cycle reset capability when the platform can physically cut power to the device.
The implementation strictly requires _PR3 to be present - the platform must be able to control device power. This ensures d3cold only attempts true power cycling, not falling back to D3hot transitions.
D3cold reset is placed at the end of the reset hierarchy since it requires specific platform support and should be tried after standard methods.
Reset hierarchy with this change:
1. device_specific
2. acpi
3. flr
4. af_flr
5. pm (D3hot via config space, checks NoSoftRst)
6. bus (SBR)
7. cxl_bus
8. d3cold (NEW - true power cycle, requires _PR3)
This benefits:
- Platforms with _PR3 support
- Single-function devices needing true power cycle
- VFIO passthrough scenarios where FLR/PM unavailable
---
v4: Strict _PR3 requirement - no D3hot fallback (Alex Williamson and Lukas
Wunner feedback)
v3: https://lore.kernel.org/all/[email protected]/
drivers/pci/pci.c | 40 ++++++++++++++++++++++++++++++++++++++++ include/linux/pci.h | 2 +- 2 files changed, 41 insertions(+), 1 deletion(-)
drivers/pci/pci.c +40 -0
@@ -4491,6 +4491,45 @@ static int pci_pm_reset(struct pci_dev *dev, bool probe)return ret; } +/** +* pci_d3cold_reset - Put device into D3cold and back to D0 for reset +* @dev: PCI device to reset +* @probe: if true, check if D3cold reset is supported; if false, perform reset +* +* Reset the device by transitioning through D3cold (actual power removal via +* platform power control) and back to D0. This requires ACPI _PR3 power +* resources to be present - the platform must be able to physically cut power +* to the device. +* +* Only available for single-function devices to avoid affecting other +* functions in multi-function devices. +* +* Returns 0 if device can be/was reset this way, -ENOTTY if not supported, +* or other negative error code on failure. +*/ +static int pci_d3cold_reset (struct pci_dev *dev,bool probe) +{ +int ret; + +if (dev->multifunction) +return -ENOTTY ; + +if (!pci_pr3_present (dev)) +return -ENOTTY ; + +if (probe) +return 0 ; + +if (dev->current_state !=PCI_D0 ) +return -EINVAL ; + + ret =pci_set_power_state (dev, PCI_D3cold); +if (ret) +return ret; + +return pci_set_power_state (dev,PCI_D0 ); +} +/** * pcie_wait_for_link_status - Wait for link status change * @pdev: Device whose link to wait for.
@@ -5065,6 +5104,7 @@ const struct pci_reset_fn_method pci_reset_fn_methods[] = { { pci_pm_reset, .name = "pm" }, { pci_reset_bus_function, .name = "bus" }, { cxl_reset_bus_function, .name = "cxl_bus" }, + { pci_d3cold_reset, .name = "d3cold" }, };/**
include/linux/pci.h +1 -1
@@ -51,7 +51,7 @@ PCI_STATUS_PARITY)/* Number of reset methods used in pci_reset_fn_methods array in pci.c */ -#define PCI_NUM_RESET_METHODS 8 +#define PCI_NUM_RESET_METHODS 9#define PCI_RESET_PROBE true#define PCI_RESET_DO_RESET false
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v4 1/3] PCI: Add d3cold as general reset method
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider: - [Critical] Missing IOMMU preparation and completion calls in the `pci_d3cold_reset` flow. - [High]…
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
- [Critical] Missing IOMMU preparation and completion calls in the pci_d3cold_reset flow.
- [High] Missing wait for device readiness (CRS polling) after transitioning from D3cold back to D0.
- [High] Bypassing device quirks and user constraints (dev->no_d3cold and dev->d3cold_allowed) for D3cold transitions. --
commit 6ccb7429bcb1b8c8a16f329c1b8caf009c316653
Author: Jose Ignacio Tornos Martinez <[email protected]>
Subject: PCI: Add d3cold as general reset method
This commit adds a new PCI reset method using D3cold power cycling for single-function devices. It utilizes ACPI _PR3 power resources to physically cut power to the device.
Quoted 4 lines
drivers/pci/pci.c +0
-0
[ ... ]
drivers/pci/pci.c static int pci_pm_reset(struct pci_dev *dev, bool probe) · line 4521
Quoted 12 lines
drivers/pci/pci.c +12
-0+static int pci_d3cold_reset (struct pci_dev *dev,bool probe) +{ +int ret; + +if (dev->multifunction) +return -ENOTTY ; + +if (!pci_pr3_present (dev)) +return -ENOTTY ; + +if (probe) +return 0 ;
Could this transition devices into D3cold even if they have the state explicitly disabled via quirks or user constraints?
Should this check dev->no_d3cold and dev->d3cold_allowed before proceeding with the transition?
drivers/pci/pci.c static int pci_pm_reset(struct pci_dev *dev, bool probe) · line 4526
Quoted 5 lines
drivers/pci/pci.c +5
-0+ +if (dev->current_state !=PCI_D0 ) +return -EINVAL ; + + ret =pci_set_power_state (dev, PCI_D3cold);
Does this reset method need to pause IOMMU fault handling and wait for pending DMA transactions before dropping physical power?
Other reset methods like pci_pm_reset() call pci_dev_reset_iommu_prepare() before performing the hardware reset. Is similar synchronization required here to prevent IOMMU faults?
drivers/pci/pci.c static int pci_pm_reset(struct pci_dev *dev, bool probe) · line 4531
Quoted 5 lines
drivers/pci/pci.c +5
-0+if (ret) +return ret; + +return pci_set_power_state (dev,PCI_D0 ); +}
Does the device require time to reinitialize its link and internal state after returning to PCI_D0?
The generic reset handler will immediately invoke pci_dev_restore() after this returns. Should this function call pci_dev_wait() to check for Configuration Request Retry Status (CRS) clearance, similar to how pci_pm_reset() waits?
Also, if pci_dev_reset_iommu_prepare() is needed earlier in this function, should pci_dev_reset_iommu_done() be called here before returning?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
PATCH v4 2/3 PCI: Add soft reset method as last resort
In reply to Jose Ignacio Tornos MartinezAdd a software-initiated "soft" reset method that attempts D3hot->D0 transition as an absolute last resort when all other reset methods have failed.
Some devices incorrectly advertise NoSoftRst+ (blocking PM reset) but the D3hot transition does provide sufficient reset for certain use cases, particularly VFIO passthrough scenarios. This method provides a "better than nothing" option when the device would otherwise have no reset capability.
The method only becomes available when:
- pci_pm_reset() is unavailable (typically blocked by NoSoftRst+)
- pci_d3cold_reset() is unavailable (no platform _PR3 support)
- Device has PM capability (required for D3hot transition)
Extract the D3hot transition logic into a shared helper function (pci_do_d3hot_transition) used by both pci_pm_reset and pci_soft_reset.
Reset hierarchy with this change:
1. device_specific
2. acpi
3. flr
4. af_flr
5. pm (proper method, checks NoSoftRst)
6. bus
7. cxl_bus
8. d3cold (requires _PR3)
9. soft (NEW - D3hot without NoSoftRst check, absolute last resort)
---
v4: Implements D3hot transition as last resort when pm/d3cold unavailable
v3: https://lore.kernel.org/all/[email protected]/
drivers/pci/pci.c | 98 ++++++++++++++++++++++++++++++++++----------- include/linux/pci.h | 2 +- 2 files changed, 76 insertions(+), 24 deletions(-)
drivers/pci/pci.c +75 -23
@@ -4437,6 +4437,43 @@ static int pci_af_flr(struct pci_dev *dev, bool probe)return ret; } +/** +* pci_do_d3hot_transition - Perform D3hot->D0 power state transition +* @dev: Device to transition +* +* Common helper to perform D3hot->D0 transition for PM-based reset methods. +* Handles IOMMU preparation, state transition, and waiting for device ready. +*/ +static int pci_do_d3hot_transition (struct pci_dev *dev) +{ +u16 csr; +int ret; + +if (dev->current_state !=PCI_D0 ) +return -EINVAL ; + + ret =pci_dev_reset_iommu_prepare (dev); +if (ret) { +pci_err (dev,"failed to stop IOMMU for a PCI reset: %d\n" , ret); +return ret; + } + +pci_read_config_word (dev, dev->pm_cap +PCI_PM_CTRL , &csr); + csr &= ~PCI_PM_CTRL_STATE_MASK ; + csr |= PCI_D3hot; +pci_write_config_word (dev, dev->pm_cap +PCI_PM_CTRL , csr); +pci_dev_d3_sleep (dev); + + csr &= ~PCI_PM_CTRL_STATE_MASK ; + csr |=PCI_D0 ; +pci_write_config_word (dev, dev->pm_cap +PCI_PM_CTRL , csr); +pci_dev_d3_sleep (dev); + + ret =pci_dev_wait (dev,"PM D3hot->D0" ,PCIE_RESET_READY_POLL_MS ); +pci_dev_reset_iommu_done (dev); +return ret; +} +/** * pci_pm_reset - Put device into PCI_D3 and back into PCI_D0. * @dev: Device to reset.
@@ -4455,7 +4492,6 @@ static int pci_af_flr(struct pci_dev *dev, bool probe)static int pci_pm_reset (struct pci_dev *dev,bool probe) {u16 csr; -int ret;if (!dev->pm_cap || dev->dev_flags &PCI_DEV_FLAGS_NO_PM_RESET )return -ENOTTY ;
@@ -4467,28 +4503,7 @@ static int pci_pm_reset(struct pci_dev *dev, bool probe)if (probe)return 0 ; -if (dev->current_state !=PCI_D0 ) -return -EINVAL ; - - ret =pci_dev_reset_iommu_prepare (dev); -if (ret) { -pci_err (dev,"failed to stop IOMMU for a PCI reset: %d\n" , ret); -return ret; - } - - csr &= ~PCI_PM_CTRL_STATE_MASK ; - csr |= PCI_D3hot; -pci_write_config_word (dev, dev->pm_cap +PCI_PM_CTRL , csr); -pci_dev_d3_sleep (dev); - - csr &= ~PCI_PM_CTRL_STATE_MASK ; - csr |=PCI_D0 ; -pci_write_config_word (dev, dev->pm_cap +PCI_PM_CTRL , csr); -pci_dev_d3_sleep (dev); - - ret =pci_dev_wait (dev,"PM D3hot->D0" ,PCIE_RESET_READY_POLL_MS ); -pci_dev_reset_iommu_done (dev); -return ret; +return pci_do_d3hot_transition (dev); }/**
@@ -4530,6 +4545,42 @@ static int pci_d3cold_reset(struct pci_dev *dev, bool probe)return pci_set_power_state (dev,PCI_D0 ); } +/** +* pci_soft_reset - Software-initiated reset via D3hot as last resort +* @dev: PCI device to reset +* @probe: if true, check if soft reset is supported; if false, perform reset +* +* Attempt a software-initiated reset via D3hot->D0 transition as an absolute +* last resort when all other reset methods have failed. This method only +* becomes available if the device has PM capability, pci_pm_reset() is blocked +* (typically by NoSoftRst+), and pci_d3cold_reset() is not available. +* +* Some devices incorrectly advertise NoSoftRst+ but D3hot transition does +* provide sufficient reset for certain use cases (e.g., VFIO passthrough). +* This method provides a "better than nothing" option when the device would +* otherwise have no reset capability. +* +* Returns 0 if device can be/was reset this way, -ENOTTY if a better reset +* method is available (pm or d3cold) or device lacks PM capability, or other +* negative error code on failure. +*/ +static int pci_soft_reset (struct pci_dev *dev,bool probe) +{ +if (pci_pm_reset (dev, true) ==0 ) +return -ENOTTY ; + +if (pci_d3cold_reset (dev, true) ==0 ) +return -ENOTTY ; + +if (!dev->pm_cap) +return -ENOTTY ; + +if (probe) +return 0 ; + +return pci_do_d3hot_transition (dev); +} +/** * pcie_wait_for_link_status - Wait for link status change * @pdev: Device whose link to wait for.
@@ -5105,6 +5156,7 @@ const struct pci_reset_fn_method pci_reset_fn_methods[] = { { pci_reset_bus_function, .name = "bus" }, { cxl_reset_bus_function, .name = "cxl_bus" }, { pci_d3cold_reset, .name = "d3cold" }, + { pci_soft_reset, .name = "soft" }, };/**
include/linux/pci.h +1 -1
@@ -51,7 +51,7 @@ PCI_STATUS_PARITY)/* Number of reset methods used in pci_reset_fn_methods array in pci.c */ -#define PCI_NUM_RESET_METHODS 9 +#define PCI_NUM_RESET_METHODS 10#define PCI_RESET_PROBE true#define PCI_RESET_DO_RESET false
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v4 2/3] PCI: Add soft reset method as last resort
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [Critical] pci_soft_reset() falsely reports a successful reset for NoSoftRst+ devices, destroying…
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Critical] pci_soft_reset() falsely reports a successful reset for NoSoftRst+ devices, destroying hardware isolation guarantees for VFIO passthrough.
- [High] pci_soft_reset() bypasses the PCI_DEV_FLAGS_NO_PM_RESET quirk, forcing unprotected hardware power state transitions on vulnerable devices. --
commit 9f864e1927975542a72967bb27f8a02e42ade61b
Author: Jose Ignacio Tornos Martinez <[email protected]>
PCI: Add soft reset method as last resort
This commit adds a software-initiated reset method that attempts a D3hot to D0 transition as a last resort when all other reset methods fail. It extracts the transition logic into a shared helper function used by both the standard power management reset and the new fallback soft reset.
Quoted 4 lines
drivers/pci/pci.c +0
-0
[ ... ]
Quoted 6 lines
drivers/pci/pci.c +2
-0@@ -4491,6 +4474,38 @@ static int pci_pm_reset(struct pci_dev *dev, bool probe)return ret; } +/** +* pci_pm_reset - Put device into PCI_D3 and back into PCI_D0.
[ ... ]
Quoted 6 lines
drivers/pci/pci.c +6
-0+static int pci_pm_reset (struct pci_dev *dev,bool probe) +{ +u16 csr; + +if (!dev->pm_cap || dev->dev_flags &PCI_DEV_FLAGS_NO_PM_RESET ) +return -ENOTTY ;
[ ... ]
Quoted 6 lines
drivers/pci/pci.c +2
-0@@ -4530,6 +4545,42 @@ static int pci_d3cold_reset(struct pci_dev *dev, bool probe)return pci_set_power_state (dev,PCI_D0 ); } +/** +* pci_soft_reset - Software-initiated reset via D3hot as last resort
[ ... ]
Quoted 6 lines
drivers/pci/pci.c +6
-0+static int pci_soft_reset (struct pci_dev *dev,bool probe) +{ +if (pci_pm_reset (dev, true) ==0 ) +return -ENOTTY ; + +if (pci_d3cold_reset (dev, true) ==0 )
drivers/pci/pci.c static int pci_d3cold_reset(struct pci_dev *dev, bool probe) · line 4582
Quoted 10 lines
drivers/pci/pci.c +10
-0+return -ENOTTY ; + +if (!dev->pm_cap) +return -ENOTTY ; + +if (probe) +return 0 ; + +return pci_do_d3hot_transition (dev); +}
Does this sequence introduce a regression by bypassing the PCI_DEV_FLAGS_NO_PM_RESET quirk?
When pci_pm_reset() is called with probe set to true, it returns -ENOTTY if the device has the PCI_DEV_FLAGS_NO_PM_RESET flag set. Since -ENOTTY is not 0, pci_soft_reset() will proceed and unconditionally execute pci_do_d3hot_transition().
Could this force a power state transition on vulnerable devices that were explicitly blacklisted from power management resets?
Additionally, does bypassing the NoSoftRst check globally introduce a regression in device sanitization for compliant hardware?
The PCI specification defines the NoSoftRst bit to indicate that a device retains its internal state during a D3hot to D0 transition. If a compliant device correctly implements NoSoftRst set to 1, this fallback method will falsely report a successful reset without actually scrubbing the device context.
Since security boundaries like VFIO passthrough rely on pci_reset_function() to guarantee complete device sanitization between VMs, can this global fallback allow state contamination or data leaks?
Would it be safer to handle devices that incorrectly advertise NoSoftRst via device-specific quirks instead of a global bypass?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2
Re: [PATCH v4 2/3] PCI: Add soft reset method as last resort
In reply to Jose Ignacio Tornos MartinezQuoted 29 lines · On Mon, 18 May 2026 14:48:34 +0200 Jose Ignacio Tornos Martinez…
Add a software-initiated "soft" reset method that attempts
D3hot->D0transition as an absolute last resort when all other reset methods have failed.Some devices incorrectly advertise NoSoftRst+ (blocking PM reset) but the D3hot transition does provide sufficient reset for certain use cases, particularly VFIO passthrough scenarios. This method provides a "better than nothing" option when the device would otherwise have no reset capability.
The method only becomes available when:
-pci_pm_reset()is unavailable (typically blocked by NoSoftRst+)
-pci_d3cold_reset()is unavailable (no platform_PR3support)
- Device has PM capability (required for D3hot transition)Extract the D3hot transition logic into a shared helper function (
pci_do_d3hot_transition) used by bothpci_pm_resetandpci_soft_reset.Reset hierarchy with this change:
1.device_specific
2. acpi
3. flr
4.af_flr
5. pm (proper method, checks NoSoftRst)
6. bus
7.cxl_bus
8. d3cold (requires_PR3)
9. soft (NEW - D3hot without NoSoftRst check, absolute last resort)
Quoted 8 lines
Signed-off-by Jose Ignacio Tornos Martinez <[email protected]>---
v4: Implements D3hot transition as last resort when pm/d3cold unavailable
v3: https://lore.kernel.org/all/[email protected]/drivers/pci/pci.c | 98 ++++++++++++++++++++++++++++++++++----------- include/linux/pci.h | 2 +- 2 files changed, 76 insertions(+), 24 deletions(-)
NAK. This cannot happen as a general case, it will cause vfio-pci to report reset capabilities for essentially all devices, whether validated or not.
The suggestion was that for devices where this has proven "better than nothing", we could think about a device specific version of this, matching devices IDs, not a fall-through for any device.
Given the "partial reset" nature of this, even on the target device, I still wonder though whether userspace cannot already handle this by forcing the power state through sysfs prior to assigning. Thanks,
Alex
Quoted 150 lines
drivers/pci/pci.c +75
-23@@ -4437,6 +4437,43 @@ static int pci_af_flr(struct pci_dev *dev, bool probe)return ret; } +/** +* pci_do_d3hot_transition - Perform D3hot->D0 power state transition +* @dev: Device to transition +* +* Common helper to perform D3hot->D0 transition for PM-based reset methods. +* Handles IOMMU preparation, state transition, and waiting for device ready. +*/ +static int pci_do_d3hot_transition (struct pci_dev *dev) +{ +u16 csr; +int ret; + +if (dev->current_state !=PCI_D0 ) +return -EINVAL ; + + ret =pci_dev_reset_iommu_prepare (dev); +if (ret) { +pci_err (dev,"failed to stop IOMMU for a PCI reset: %d\n" , ret); +return ret; + } + +pci_read_config_word (dev, dev->pm_cap +PCI_PM_CTRL , &csr); + csr &= ~PCI_PM_CTRL_STATE_MASK ; + csr |= PCI_D3hot; +pci_write_config_word (dev, dev->pm_cap +PCI_PM_CTRL , csr); +pci_dev_d3_sleep (dev); + + csr &= ~PCI_PM_CTRL_STATE_MASK ; + csr |=PCI_D0 ; +pci_write_config_word (dev, dev->pm_cap +PCI_PM_CTRL , csr); +pci_dev_d3_sleep (dev); + + ret =pci_dev_wait (dev,"PM D3hot->D0" ,PCIE_RESET_READY_POLL_MS ); +pci_dev_reset_iommu_done (dev); +return ret; +} +/** * pci_pm_reset - Put device into PCI_D3 and back into PCI_D0. * @dev: Device to reset. @@ -4455,7 +4492,6 @@ static int pci_af_flr(struct pci_dev *dev, bool probe)static int pci_pm_reset (struct pci_dev *dev,bool probe) {u16 csr; -int ret;if (!dev->pm_cap || dev->dev_flags &PCI_DEV_FLAGS_NO_PM_RESET )return -ENOTTY ;@@ -4467,28 +4503,7 @@ static int pci_pm_reset(struct pci_dev *dev, bool probe)if (probe)return 0 ; -if (dev->current_state !=PCI_D0 ) -return -EINVAL ; - - ret =pci_dev_reset_iommu_prepare (dev); -if (ret) { -pci_err (dev,"failed to stop IOMMU for a PCI reset: %d\n" , ret); -return ret; - } - - csr &= ~PCI_PM_CTRL_STATE_MASK ; - csr |= PCI_D3hot; -pci_write_config_word (dev, dev->pm_cap +PCI_PM_CTRL , csr); -pci_dev_d3_sleep (dev); - - csr &= ~PCI_PM_CTRL_STATE_MASK ; - csr |=PCI_D0 ; -pci_write_config_word (dev, dev->pm_cap +PCI_PM_CTRL , csr); -pci_dev_d3_sleep (dev); - - ret =pci_dev_wait (dev,"PM D3hot->D0" ,PCIE_RESET_READY_POLL_MS ); -pci_dev_reset_iommu_done (dev); -return ret; +return pci_do_d3hot_transition (dev); }/** @@ -4530,6 +4545,42 @@ static int pci_d3cold_reset(struct pci_dev *dev, bool probe)return pci_set_power_state (dev,PCI_D0 ); } +/** +* pci_soft_reset - Software-initiated reset via D3hot as last resort +* @dev: PCI device to reset +* @probe: if true, check if soft reset is supported; if false, perform reset +* +* Attempt a software-initiated reset via D3hot->D0 transition as an absolute +* last resort when all other reset methods have failed. This method only +* becomes available if the device has PM capability, pci_pm_reset() is blocked +* (typically by NoSoftRst+), and pci_d3cold_reset() is not available. +* +* Some devices incorrectly advertise NoSoftRst+ but D3hot transition does +* provide sufficient reset for certain use cases (e.g., VFIO passthrough). +* This method provides a "better than nothing" option when the device would +* otherwise have no reset capability. +* +* Returns 0 if device can be/was reset this way, -ENOTTY if a better reset +* method is available (pm or d3cold) or device lacks PM capability, or other +* negative error code on failure. +*/ +static int pci_soft_reset (struct pci_dev *dev,bool probe) +{ +if (pci_pm_reset (dev, true) ==0 ) +return -ENOTTY ; + +if (pci_d3cold_reset (dev, true) ==0 ) +return -ENOTTY ; + +if (!dev->pm_cap) +return -ENOTTY ; + +if (probe) +return 0 ; + +return pci_do_d3hot_transition (dev); +} +/** * pcie_wait_for_link_status - Wait for link status change * @pdev: Device whose link to wait for. @@ -5105,6 +5156,7 @@ const struct pci_reset_fn_method pci_reset_fn_methods[] = { { pci_reset_bus_function, .name = "bus" }, { cxl_reset_bus_function, .name = "cxl_bus" }, { pci_d3cold_reset, .name = "d3cold" }, + { pci_soft_reset, .name = "soft" }, };/** include/linux/pci.h +1
-1@@ -51,7 +51,7 @@ PCI_STATUS_PARITY)/* Number of reset methods used in pci_reset_fn_methods array in pci.c */ -#define PCI_NUM_RESET_METHODS 9 +#define PCI_NUM_RESET_METHODS 10#define PCI_RESET_PROBE true#define PCI_RESET_DO_RESET false
Re: [PATCH v4 2/3] PCI: Add soft reset method as last resort
Hello Alex,
Thank you for the feedback. I understand the concern about reporting reset capabilities for all devices.
Regarding the sysfs power state approach: I tested this, but the challenge is that VFIO needs automatic reset during VM crash/reassignment. When a VM terminates uncleanly, VFIO calls the device reset path automatically before reassignment - there's no opportunity for userspace to manipulate sysfs power state in that flow.
For the device-specific approach you mentioned, what if I modify the "soft" method to require an explicit quirk flag? Something like:
static int pci_soft_reset(struct pci_dev *dev, bool probe)
{
/* Only available if device explicitly quirked for soft reset */
if (!(dev->dev_flags & PCI_DEV_FLAGS_ALLOW_SOFT_RESET))
return -ENOTTY;
if (pci_pm_reset(dev, true) == 0)
return -ENOTTY;
if (pci_d3cold_reset(dev, true) == 0)
return -ENOTTY;
if (!dev->pm_cap)
return -ENOTTY;
if (probe)
return 0;
return pci_do_d3hot_transition(dev);}
DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_QCOM, 0x1103, quirk_allow_soft_reset); DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_QCOM, 0x1107, quirk_allow_soft_reset); DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_QCOM, 0x0308, quirk_allow_soft_reset);
This way:
- Only proven device IDs get soft reset capability (no broad reset reporting)
- Device-specific version matching device IDs (as you suggested)
- Generic infrastructure but explicit opt-in per device
- Different from v2: new method with quirk, not modifying pm reset behavior
Would this approach be acceptable?
Thanks
Best regards
José Ignacio
Re: [PATCH v4 2/3] PCI: Add soft reset method as last resort
Quoted 20 lines · On Tue, 19 May 2026 07:35:50 +0200 Jose Ignacio Tornos Martinez…
Hello Alex,
Thank you for the feedback. I understand the concern about reporting reset capabilities for all devices.
Regarding the sysfs power state approach: I tested this, but the challenge is that VFIO needs automatic reset during VM crash/reassignment. When a VM terminates uncleanly, VFIO calls the device reset path automatically before reassignment - there's no opportunity for userspace to manipulate sysfs power state in that flow.
For the device-specific approach you mentioned, what if I modify the "soft" method to require an explicit quirk flag? Something like:
static int
pci_soft_reset(struct pci_dev *dev, bool probe){ /* Only available if device explicitly quirked for soft reset */ if (!(dev->dev_flags & PCI_DEV_FLAGS_ALLOW_SOFT_RESET)) return -ENOTTY;
...
Quoted 2 lines
Would this approach be acceptable?
Device specific resets are made for this scenario. Look at pci_dev_specific_reset() and pci_dev_reset_methods[]. The supporting evidence that this performs a worthwhile reset is still a bit weak, but heuristically it seems better than nothing, which is what we're left with otherwise. Reset via D3hot for a device that does not expose NoSoftRst- is not something we should enable or endorse for any common use case. Thanks,
Alex
Re: [PATCH v4 2/3] PCI: Add soft reset method as last resort
Hello Alex,
Thank you for the guidance.
OK, using pci_dev_reset_methods[] makes sense.
I'll implement device_specific reset entries for the Qualcomm devices using pci_set_power_state(dev, D3cold), which automatically falls back to D3hot when the platform doesn't support D3cold. This handles both scenarios (platforms with and without _PR3).
I'll keep the general d3cold reset method (patch 1/3) as well, since it provides value for other devices that need strict _PR3-based power cycling without D3hot fallback. The device_specific entries are the Qualcomm-specific workaround, while d3cold remains general infrastructure.
I'll also keep quirk_no_bus_reset as a safety net to prevent the known-broken bus reset from being attempted. This ensures that if users override reset methods via sysfs, they can still use d3cold (when _PR3 is available) without accidentally falling back to the broken bus reset.
Thanks
Best regards
José Ignacio
PATCH v4 3/3 PCI: Disable broken bus reset on Qualcomm devices
In reply to Jose Ignacio Tornos MartinezSome Qualcomm PCIe devices do not properly support Secondary Bus Reset (SBR). These devices have no FLR capability and advertise NoSoftRst+ (blocking PM reset), leaving bus reset as the only available method. However, bus reset does not work reliably for these devices.
The problem manifests in VFIO passthrough scenarios with these affected devices:
- ath11k WiFi (17cb:1103): Normal VM operation works fine, including clean shutdown/reboot. However, when the VM terminates uncleanly (crash, force-off), VFIO attempts to reset the device before it can be assigned to another VM. Without a working reset method, the device remains in an undefined state, preventing reuse.
- ath12k WiFi (17cb:1107): Same behavior as ath11k.
- SDX62/SDX65 5G modems (17cb:0308): Never successfully initialize even on first VM assignment without proper reset capability.
Disable bus reset for these devices (following the pattern of other Atheros/Qualcomm devices) so alternative reset methods (d3cold or soft) are used instead, which provide working reset capability.
---
v4: Quirk code unchanged from v3
v3: https://lore.kernel.org/all/[email protected]/
drivers/pci/quirks.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/pci/quirks.c +3 -0
@@ -3789,6 +3789,9 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATHEROS, 0x0030, quirk_no_bus_reset);DECLARE_PCI_FIXUP_HEADER (PCI_VENDOR_ID_ATHEROS ,0x0032 , quirk_no_bus_reset);DECLARE_PCI_FIXUP_HEADER (PCI_VENDOR_ID_ATHEROS ,0x003c , quirk_no_bus_reset);DECLARE_PCI_FIXUP_HEADER (PCI_VENDOR_ID_ATHEROS ,0x0033 , quirk_no_bus_reset);DECLARE_PCI_FIXUP_HEADER (PCI_VENDOR_ID_ATHEROS ,0x0034 , quirk_no_bus_reset);DECLARE_PCI_FIXUP_HEADER (PCI_VENDOR_ID_ATHEROS ,0x003e , quirk_no_bus_reset); +DECLARE_PCI_FIXUP_HEADER (PCI_VENDOR_ID_QCOM ,0x1103 , quirk_no_bus_reset);/* ath11k */ +DECLARE_PCI_FIXUP_HEADER (PCI_VENDOR_ID_QCOM ,0x1107 , quirk_no_bus_reset);/* ath12k */ +DECLARE_PCI_FIXUP_HEADER (PCI_VENDOR_ID_QCOM ,0x0308 , quirk_no_bus_reset);/* SDX62/SDX65 */
/*
* Root port on some Cavium CN8xxx chips do not successfully complete a bus
--
2.53.0