* [PATCH v4 1/3] PCI: Preserve the initial bus number in pci_bus
2026-10-27 16:22 [PATCH v4 0/3] Fix amdgpu detection of pci=realloc Mario Limonciello
@ 2026-10-27 16:22 ` Mario Limonciello
2026-10-27 22:41 ` Bjorn Helgaas
2026-10-27 16:22 ` [PATCH v4 2/3] drm/amdgpu: Match VFCT using initial PCI bus number Mario Limonciello
2026-10-27 16:22 ` [PATCH v4 3/3] drm/radeon: " Mario Limonciello
2 siblings, 1 reply; 5+ messages in thread
From: Mario Limonciello @ 2026-10-27 16:22 UTC (permalink / raw)
To: Bjorn Helgaas, Alex Deucher, Christian König
Cc: Oz Tiram, linux-pci, linux-kernel, amd-gfx, dri-devel, Mario Limonciello
Bridge scans are split into two passes when bus numbers need reassigning.
Save the firmware secondary bus number on the bridge in pass 0, before it
is disabled. Use it to initialize number_initial when the child bus is
created in pass 1, before bus registration callbacks run.
Keep the saved secondary number on the bridge so a later rescan does not
replace the child's initial number with its reassigned runtime number.
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
---
drivers/pci/probe.c | 6 ++++++
include/linux/pci.h | 3 +++
2 files changed, 9 insertions(+)
diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index 27008e2ea5afc..3c03ac22223d9 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c
@@ -1012,6 +1012,7 @@ static int pci_register_host_bridge(struct pci_host_bridge *bridge)
bus->sysdata = bridge->sysdata;
bus->ops = bridge->ops;
bus->number = bus->busn_res.start = bridge->busnr;
+ bus->number_initial = bus->number;
#ifdef CONFIG_PCI_DOMAINS_GENERIC
if (bridge->domain_nr == PCI_DOMAIN_NR_NOT_SET)
bus->domain_nr = pci_bus_find_domain_nr(bus, parent);
@@ -1232,6 +1233,8 @@ static struct pci_bus *pci_alloc_child_bus(struct pci_bus *parent,
/* Set up the primary, secondary and subordinate bus numbers */
child->number = child->busn_res.start = busnr;
+ child->number_initial = bridge && bridge->bridge_secondary_busnr ?
+ bridge->bridge_secondary_busnr : busnr;
child->primary = parent->busn_res.start;
child->busn_res.end = 0xff;
@@ -1418,6 +1421,8 @@ static int pci_scan_bridge_extend(struct pci_bus *bus, struct pci_dev *dev,
pci_dbg(dev, "scanning [bus %02x-%02x] behind bridge, pass %d\n",
secondary, subordinate, pass);
+ if (!pass && !dev->bridge_secondary_busnr)
+ dev->bridge_secondary_busnr = secondary;
if (!primary && (primary != bus->number) && secondary && subordinate) {
pci_warn(dev, "Primary bus is hard wired to 0\n");
@@ -1470,6 +1475,7 @@ static int pci_scan_bridge_extend(struct pci_bus *bus, struct pci_dev *dev,
child = pci_add_new_bus(bus, dev, secondary);
if (!child)
goto out;
+ child->number_initial = secondary;
child->primary = primary;
pci_bus_insert_busn_res(child, secondary, subordinate);
child->bridge_ctl = bctl;
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 5ffcf5ce8c966..c607818a98a87 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -363,6 +363,8 @@ struct pci_dev {
struct pci_slot *slot; /* Physical slot this device is in */
unsigned int devfn; /* Encoded device & function index */
+ /* Firmware secondary bus number saved across bridge scan passes */
+ u8 bridge_secondary_busnr;
unsigned short vendor;
unsigned short device;
unsigned short subsystem_vendor;
@@ -724,6 +726,7 @@ struct pci_bus {
struct proc_dir_entry *procdir; /* Directory entry in /proc/bus/pci */
unsigned char number; /* Bus number */
+ unsigned char number_initial; /* Bus number before reassignment */
unsigned char primary; /* Number of primary bridge */
unsigned char max_bus_speed; /* enum pci_bus_speed */
unsigned char cur_bus_speed; /* enum pci_bus_speed */
--
2.43.0
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [PATCH v4 1/3] PCI: Preserve the initial bus number in pci_bus
2026-10-27 16:22 ` [PATCH v4 1/3] PCI: Preserve the initial bus number in pci_bus Mario Limonciello
@ 2026-10-27 22:41 ` Bjorn Helgaas
0 siblings, 0 replies; 5+ messages in thread
From: Bjorn Helgaas @ 2026-10-27 22:41 UTC (permalink / raw)
To: Mario Limonciello
Cc: Alex Deucher, Christian König, Oz Tiram, linux-pci,
linux-kernel, amd-gfx, dri-devel, Lijo Lazar
[+cc Lijo]
On Tue, Oct 27, 2026 at 11:22:25AM -0500, Mario Limonciello wrote:
> Bridge scans are split into two passes when bus numbers need reassigning.
> Save the firmware secondary bus number on the bridge in pass 0, before it
> is disabled. Use it to initialize number_initial when the child bus is
> created in pass 1, before bus registration callbacks run.
>
> Keep the saved secondary number on the bridge so a later rescan does not
> replace the child's initial number with its reassigned runtime number.
>
> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> ---
> drivers/pci/probe.c | 6 ++++++
> include/linux/pci.h | 3 +++
> 2 files changed, 9 insertions(+)
>
> diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
> index 27008e2ea5afc..3c03ac22223d9 100644
> --- a/drivers/pci/probe.c
> +++ b/drivers/pci/probe.c
> @@ -1012,6 +1012,7 @@ static int pci_register_host_bridge(struct pci_host_bridge *bridge)
> bus->sysdata = bridge->sysdata;
> bus->ops = bridge->ops;
> bus->number = bus->busn_res.start = bridge->busnr;
> + bus->number_initial = bus->number;
> #ifdef CONFIG_PCI_DOMAINS_GENERIC
> if (bridge->domain_nr == PCI_DOMAIN_NR_NOT_SET)
> bus->domain_nr = pci_bus_find_domain_nr(bus, parent);
> @@ -1232,6 +1233,8 @@ static struct pci_bus *pci_alloc_child_bus(struct pci_bus *parent,
>
> /* Set up the primary, secondary and subordinate bus numbers */
> child->number = child->busn_res.start = busnr;
> + child->number_initial = bridge && bridge->bridge_secondary_busnr ?
> + bridge->bridge_secondary_busnr : busnr;
> child->primary = parent->busn_res.start;
> child->busn_res.end = 0xff;
>
> @@ -1418,6 +1421,8 @@ static int pci_scan_bridge_extend(struct pci_bus *bus, struct pci_dev *dev,
>
> pci_dbg(dev, "scanning [bus %02x-%02x] behind bridge, pass %d\n",
> secondary, subordinate, pass);
> + if (!pass && !dev->bridge_secondary_busnr)
> + dev->bridge_secondary_busnr = secondary;
I think I was mistaken to suggest that pci_bus was the right place to
store this.
It seems like we definitely need something in the pci_dev because we
may clear the initial bus numbers in the first pass, so we need to
stash the secondary bus number somewhere until the second pass
allocates the pci_bus and we can save it here.
Since we have to have it in the pci_dev, I suspect maybe it would be
enough to have it *only* there. Maybe it could be set in
pci_read_bridge_windows(), and drivers could find it via
pdev->bus->self?
> if (!primary && (primary != bus->number) && secondary && subordinate) {
> pci_warn(dev, "Primary bus is hard wired to 0\n");
> @@ -1470,6 +1475,7 @@ static int pci_scan_bridge_extend(struct pci_bus *bus, struct pci_dev *dev,
> child = pci_add_new_bus(bus, dev, secondary);
> if (!child)
> goto out;
> + child->number_initial = secondary;
> child->primary = primary;
> pci_bus_insert_busn_res(child, secondary, subordinate);
> child->bridge_ctl = bctl;
> diff --git a/include/linux/pci.h b/include/linux/pci.h
> index 5ffcf5ce8c966..c607818a98a87 100644
> --- a/include/linux/pci.h
> +++ b/include/linux/pci.h
> @@ -363,6 +363,8 @@ struct pci_dev {
> struct pci_slot *slot; /* Physical slot this device is in */
>
> unsigned int devfn; /* Encoded device & function index */
> + /* Firmware secondary bus number saved across bridge scan passes */
> + u8 bridge_secondary_busnr;
I think you're right to use "secondary" here so we don't perpetuate
the "subordinate" misuse. Maybe also include "initial" and make it
"unsigned char" so it's the same as pci_bus.number?
> unsigned short vendor;
> unsigned short device;
> unsigned short subsystem_vendor;
> @@ -724,6 +726,7 @@ struct pci_bus {
> struct proc_dir_entry *procdir; /* Directory entry in /proc/bus/pci */
>
> unsigned char number; /* Bus number */
> + unsigned char number_initial; /* Bus number before reassignment */
> unsigned char primary; /* Number of primary bridge */
> unsigned char max_bus_speed; /* enum pci_bus_speed */
> unsigned char cur_bus_speed; /* enum pci_bus_speed */
>
> --
> 2.43.0
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH v4 2/3] drm/amdgpu: Match VFCT using initial PCI bus number
2026-10-27 16:22 [PATCH v4 0/3] Fix amdgpu detection of pci=realloc Mario Limonciello
2026-10-27 16:22 ` [PATCH v4 1/3] PCI: Preserve the initial bus number in pci_bus Mario Limonciello
@ 2026-10-27 16:22 ` Mario Limonciello
2026-10-27 16:22 ` [PATCH v4 3/3] drm/radeon: " Mario Limonciello
2 siblings, 0 replies; 5+ messages in thread
From: Mario Limonciello @ 2026-10-27 16:22 UTC (permalink / raw)
To: Bjorn Helgaas, Alex Deucher, Christian König
Cc: Oz Tiram, linux-pci, linux-kernel, amd-gfx, dri-devel,
Mario Limonciello, Reza Nikzad Chaleshori
VFCT records the PCI bus number from BIOS POST. The current bus number
can change after PCI core renumbers the bus, while accepting a mismatch
based only on device identity can select the wrong VFCT image when
matching devices are present. Compare against the bus number captured
from the firmware bus assignment before bridge renumbering.
Reported-by: Reza Nikzad Chaleshori <rezanikzadch@gmail.com>
Closes: https://lore.kernel.org/amd-gfx/20261014080523.186533-1-rezanikzadch@gmail.com/
Fixes: 11c141672045f ("drm/amdgpu: Fix VFCT bus number matching with soft filter")
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c | 17 +++--------------
1 file changed, 3 insertions(+), 14 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
index 08ed1eb695586..051be1905e296 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
@@ -386,10 +386,8 @@ static bool amdgpu_read_disabled_bios(struct amdgpu_device *adev)
* VFCT entries contain the PCI bus number as recorded during BIOS POST.
* On systems where the kernel renumbers PCI buses (e.g. pci=realloc or
* resource conflicts), the runtime bus number may differ from the POST
- * value. Match by device identity (vendor + device + function) and use
- * the bus number as a preference: exact bus match is preferred, but when
- * the bus numbers disagree we accept the entry if the device identity
- * matches.
+ * value. Match using the bus number captured when the device was
+ * discovered, before any bus renumbering.
*
* Returns: 0 on match, -ENODEV on no match
*/
@@ -405,16 +403,7 @@ static int amdgpu_acpi_vfct_match(struct amdgpu_device *adev,
vhdr->PCIFunction != PCI_FUNC(adev->pdev->devfn))
return -ENODEV;
- /* Exact bus number match - preferred */
- if (vhdr->PCIBus == adev->pdev->bus->number)
- return 0;
-
- /* Bus mismatch but device identity matches (PCI renumbering case) */
- dev_notice(adev->dev,
- "VFCT bus number mismatch: table %u != runtime %u, matching by device identity (vendor 0x%04x device 0x%04x)\n",
- vhdr->PCIBus, adev->pdev->bus->number,
- adev->pdev->vendor, adev->pdev->device);
- return 0;
+ return vhdr->PCIBus == adev->pdev->bus->number_initial ? 0 : -ENODEV;
}
static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
--
2.43.0
^ permalink raw reply [flat|nested] 5+ messages in thread* [PATCH v4 3/3] drm/radeon: Match VFCT using initial PCI bus number
2026-10-27 16:22 [PATCH v4 0/3] Fix amdgpu detection of pci=realloc Mario Limonciello
2026-10-27 16:22 ` [PATCH v4 1/3] PCI: Preserve the initial bus number in pci_bus Mario Limonciello
2026-10-27 16:22 ` [PATCH v4 2/3] drm/amdgpu: Match VFCT using initial PCI bus number Mario Limonciello
@ 2026-10-27 16:22 ` Mario Limonciello
2 siblings, 0 replies; 5+ messages in thread
From: Mario Limonciello @ 2026-10-27 16:22 UTC (permalink / raw)
To: Bjorn Helgaas, Alex Deucher, Christian König
Cc: Oz Tiram, linux-pci, linux-kernel, amd-gfx, dri-devel, Mario Limonciello
VFCT records the PCI bus number from BIOS POST. The current bus number
can change after PCI core renumbers the bus, while accepting a mismatch
based only on device identity can select the wrong VFCT image when
matching devices are present. Compare against the bus number captured
from the firmware bus assignment before bridge renumbering.
Fixes: 4059e2f02c8a ("drm/radeon: Fix VFCT bus number matching with soft filter")
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
---
drivers/gpu/drm/radeon/radeon_bios.c | 17 +++--------------
1 file changed, 3 insertions(+), 14 deletions(-)
diff --git a/drivers/gpu/drm/radeon/radeon_bios.c b/drivers/gpu/drm/radeon/radeon_bios.c
index 1da6bf1c17ff3..bdbca4fad6b8c 100644
--- a/drivers/gpu/drm/radeon/radeon_bios.c
+++ b/drivers/gpu/drm/radeon/radeon_bios.c
@@ -609,10 +609,8 @@ static bool radeon_read_disabled_bios(struct radeon_device *rdev)
* VFCT entries contain the PCI bus number as recorded during BIOS POST.
* On systems where the kernel renumbers PCI buses (e.g. pci=realloc or
* resource conflicts), the runtime bus number may differ from the POST
- * value. Match by device identity (vendor + device + function) and use
- * the bus number as a preference: exact bus match is preferred, but when
- * the bus numbers disagree we accept the entry if the device identity
- * matches.
+ * value. Match using the bus number captured when the device was
+ * discovered, before any bus renumbering.
*
* Returns: 0 on match, -ENODEV on no match
*/
@@ -628,16 +626,7 @@ static int radeon_acpi_vfct_match(struct radeon_device *rdev,
vhdr->PCIFunction != PCI_FUNC(rdev->pdev->devfn))
return -ENODEV;
- /* Exact bus number match - preferred */
- if (vhdr->PCIBus == rdev->pdev->bus->number)
- return 0;
-
- /* Bus mismatch but device identity matches (PCI renumbering case) */
- dev_notice(&rdev->pdev->dev,
- "VFCT bus number mismatch: table %u != runtime %u, matching by device identity (vendor 0x%04x device 0x%04x)\n",
- vhdr->PCIBus, rdev->pdev->bus->number,
- rdev->pdev->vendor, rdev->pdev->device);
- return 0;
+ return vhdr->PCIBus == rdev->pdev->bus->number_initial ? 0 : -ENODEV;
}
static bool radeon_acpi_vfct_bios(struct radeon_device *rdev)
--
2.43.0
^ permalink raw reply [flat|nested] 5+ messages in thread