[PATCH v3] PCI: Fix BAR resize for devices on a root bus

[PATCH v3] PCI: Fix BAR resize for devices on a root bus

[Liz Fong-Jones via B4 Relay]

From: Liz Fong-Jones <lizf@honeycomb.io>

pci_do_resource_release_and_resize() releases the device BARs that
share a bridge window with the BAR being resized, but when the device
sits directly on a root bus (pdev->bus->self == NULL) it then skips
resource assignment entirely and returns success, leaving the BARs it
just released unassigned (IORESOURCE_UNSET).

Skipping pbus_reassign_bridge_resources() is correct in that case --
there is no bridge window to adjust -- but the device BARs still have
to be reassigned. Before the BAR release was consolidated into the PCI
core, this case worked for amdgpu because the driver released the BARs
itself and then called pci_assign_unassigned_bus_resources()
unconditionally after the resize, which assigns unassigned device BARs
also on a root bus. Commit db92e3fef53e ("drm/amdgpu: Remove driver
side BAR release before resize") removed that call, so nothing assigns
the released BARs anymore.

This breaks amdgpu completely on the SolidRun HoneyComb LX2K (NXP
LX2160A, arm64, ACPI), where the GPU endpoint is enumerated directly
on the root bus of its segment (there is no root port device, so
pdev->bus->self is NULL):

  amdgpu 0004:01:00.0: BAR 0 [mem 0xa400000000-0xa40fffffff 64bit pref]: releasing
  amdgpu 0004:01:00.0: BAR 2 [mem 0xa410000000-0xa4101fffff 64bit pref]: releasing
  amdgpu 0004:01:00.0: sw_init of IP block <gmc_v8_0> failed -19
  amdgpu 0004:01:00.0: amdgpu_device_ip_init failed
  amdgpu 0004:01:00.0: Fatal error during GPU init

No error is logged because the resize path reports success; amdgpu
then finds BAR0 IORESOURCE_UNSET and bails out with -ENODEV.

Assign the released BARs directly from the root bus windows when there
is no upstream bridge. On failure, roll back through the existing
restore path exactly as in the bridged case.

The root bus path also had a locking bug that any fix here necessarily
touches: the old "goto out" jumped to up_read(&pci_bus_sem) without a
matching down_read() (as does the "goto restore" taken when
pci_dev_res_add_to_list() fails in the release loop). Take pci_bus_sem
before the BAR release loop so every path through the function holds
it exactly once.

Use pci_upstream_bridge() rather than testing pdev->bus->self
directly to decide whether the device has an upstream bridge. The two
are usually equivalent, but pci_is_root_bus() -- which
pci_upstream_bridge() uses internally -- documents that "virtual"
buses added for SR-IOV via virtfn_add_bus() also have bus->self ==
NULL without being root buses. pci_do_resource_release_and_resize()
is reachable for VF BAR resizes (see the pci_resource_is_iov() checks
elsewhere in this file and in rebar.c), so testing bus->self directly
could misroute a VF resize into the root-bus assignment path added
above. Nothing about the bug being fixed here involves a VF; this is
purely for correctness on that path.

Fixes: 337b1b566db0 ("PCI: Fix restoring BARs on BAR resize rollback path")
Cc: stable@vger.kernel.org
Link: https://bugs.launchpad.net/ubuntu/+source/linux-hwe-7.0/+bug/2159596
Assisted-by: Claude:claude-fable-5 checkpatch
Assisted-by: Claude:claude-sonnet-5
Reviewed-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Liz Fong-Jones <lizf@honeycomb.io>
---
#regzbot introduced: 337b1b566db0

Observed at runtime on Ubuntu's linux-hwe-7.0 (7.0.0-14, broken) vs
linux-hwe-6.17 (working), but nothing here is distro-specific: Ubuntu
carries this code unmodified, and the affected function is identical
to current mainline. By source inspection the regression window is
v6.18 (old code paths) to v6.19 (consolidation). Workaround for
affected users: amdgpu.rebar=0.

On the topology question raised in review: this is physical hardware,
not a virtualized guest, running the UEFI/ACPI firmware variant of
this board (SolidRun also ships a U-Boot/devicetree variant of the
same hardware). Deferring to Jon Nettleton (CC'd, SolidRun) on the
exact root complex wiring; see the hardware manual's block diagram:
https://dev.solid-run.com/nxp/lx2160a/com-som/lx2160a-com-hardware-user-manual#block-diagram
---
Changes in v3:
- Add Reviewed-by from Krzysztof Wilczyński, received on v2.
- Use pci_upstream_bridge() instead of testing pdev->bus->self
  directly, per Bjorn Helgaas's review: bus->self == NULL doesn't
  necessarily mean a root bus (SR-IOV virtfn_add_bus() buses are the
  same), and this function is reachable for VF BAR resizes elsewhere
  in this file. Doesn't affect the bug being fixed here (not a VF)
  but is more correct in general. Reviewed-by kept from v2 -- this is
  a narrow, mechanical change to a path this device doesn't exercise,
  not a substantive rework of what was reviewed.
- Rebase onto current mainline.
- Add Jacob Martin to Cc: he independently triaged and accepted the
  corresponding Ubuntu report (LP: #2159596) on the Canonical kernel
  team, so he has direct interest in this landing.
- Move Ilpo Järvinen to To: for final review, per Bjorn.
- Link to v2: https://patch.msgid.link/20260712-pci-rebar-root-bus-v2-1-a1b9107a82dc@honeycomb.io

Changes in v2:
- Add Assisted-by tags (missing from v1; required per
  Documentation/process/coding-assistants.rst)
- Add Link: to the corresponding Ubuntu bug report
- Drop the "# v6.19+" annotation on Cc: stable; unnecessary noise
  given the Fixes: tag already lets the stable team derive applicable
  versions (per stable-kernel-rules.rst)
- Link to v1: https://patch.msgid.link/20260705-pci-rebar-root-bus-v1-1-55df70cbdd88@honeycomb.io

To: Bjorn Helgaas <bhelgaas@google.com>
To: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Cc: linux-pci@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: regressions@lists.linux.dev
Cc: amd-gfx@lists.freedesktop.org
Cc: Jon Nettleton <jon@solid-run.com>
Cc: Jon Nettleton <jon.nettleton@gmail.com>
Cc: Jacob Martin <jacob.martin@canonical.com>
---
 drivers/pci/setup-bus.c | 26 ++++++++++++++++++++------
 1 file changed, 20 insertions(+), 6 deletions(-)

diff --git a/drivers/pci/setup-bus.c b/drivers/pci/setup-bus.c
index c0a949f2c9956..d50f3d57a7791 100644
--- a/drivers/pci/setup-bus.c
+++ b/drivers/pci/setup-bus.c
@@ -2380,6 +2380,7 @@ int pci_do_resource_release_and_resize(struct pci_dev *pdev, int resno, int size
 	struct resource *res = pci_resource_n(pdev, resno);
 	struct pci_dev_resource *dev_res;
 	struct pci_bus *bus = pdev->bus;
+	struct pci_dev *bridge = pci_upstream_bridge(pdev);
 	struct resource *b_win, *r;
 	LIST_HEAD(saved);
 	unsigned int i;
@@ -2397,6 +2398,8 @@ int pci_do_resource_release_and_resize(struct pci_dev *pdev, int resno, int size
 	if (ret)
 		return ret;
 
+	down_read(&pci_bus_sem);
+
 	pci_dev_for_each_resource(pdev, r, i) {
 		if (i >= PCI_BRIDGE_RESOURCES)
 			break;
@@ -2415,13 +2418,24 @@ int pci_do_resource_release_and_resize(struct pci_dev *pdev, int resno, int size
 
 	pci_resize_resource_set_size(pdev, resno, size);
 
-	if (!bus->self)
-		goto out;
+	if (bridge) {
+		ret = pbus_reassign_bridge_resources(bus, res, &saved);
+		if (ret)
+			goto restore;
+	} else {
+		/*
+		 * A device on a root bus has no bridge windows to adjust.
+		 * Assign the BARs released above directly from the root bus
+		 * windows.
+		 */
+		list_for_each_entry(dev_res, &saved, list) {
+			i = pci_resource_num(pdev, dev_res->res);
 
-	down_read(&pci_bus_sem);
-	ret = pbus_reassign_bridge_resources(bus, res, &saved);
-	if (ret)
-		goto restore;
+			ret = pci_assign_resource(pdev, i);
+			if (ret)
+				goto restore;
+		}
+	}
 
 out:
 	up_read(&pci_bus_sem);

---
base-commit: 8ba098e6b6ff0db8edf28528d1552be261af30d4
change-id: 20260704-pci-rebar-root-bus-f3123fe10fc7

Best regards,
--  
Liz Fong-Jones <lizf@honeycomb.io>

Re: [PATCH v3] PCI: Fix BAR resize for devices on a root bus

[sashiko-bot]

> From: Liz Fong-Jones <lizf@honeycomb.io>
> 
> pci_do_resource_release_and_resize() releases the device BARs that
> share a bridge window with the BAR being resized, but when the device
> sits directly on a root bus (pdev->bus->self == NULL) it then skips
> resource assignment entirely and returns success, leaving the BARs it
> just released unassigned (IORESOURCE_UNSET).
> 
> Skipping pbus_reassign_bridge_resources() is correct in that case --
> there is no bridge window to adjust -- but the device BARs still have
> to be reassigned. Before the BAR release was consolidated into the PCI
> core, this case worked for amdgpu because the driver released the BARs
> itself and then called pci_assign_unassigned_bus_resources()
> unconditionally after the resize, which assigns unassigned device BARs
> also on a root bus. Commit db92e3fef53e ("drm/amdgpu: Remove driver
> [ ... ]
> Link: https://bugs.launchpad.net/ubuntu/+source/linux-hwe-7.0/+bug/2159596
> Assisted-by: Claude:claude-fable-5 checkpatch
> Assisted-by: Claude:claude-sonnet-5
> Reviewed-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
> Signed-off-by: Liz Fong-Jones <lizf@honeycomb.io>

Sashiko has reviewed this patch and found no issues. It looks great!

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260731-pci-rebar-root-bus-v3-1-6732e233fc99@honeycomb.io?part=1