linux-stable.git/drivers/pci/devres.c, branch v6.11

PCI: Fix potential deadlock in pcim_intx()

2024-09-12T12:52:50+00:00

25216afc9db5 ("PCI: Add managed pcim_intx()") moved the allocation step for
pci_intx()'s device resource from pcim_enable_device() to pcim_intx(). As
before, pcim_enable_device() sets pci_dev.is_managed to true; and it is
never set to false again.

Due to the lifecycle of a struct pci_dev, it can happen that a second
driver obtains the same pci_dev after a first driver ran.  If one driver
uses pcim_enable_device() and the other doesn't, this causes the other
driver to run into managed pcim_intx(), which will try to allocate when
called for the first time.

Allocations might sleep, so calling pci_intx() while holding spinlocks
becomes then invalid, which causes lockdep warnings and could cause
deadlocks:

  ========================================================
  WARNING: possible irq lock inversion dependency detected
  6.11.0-rc6+ #59 Tainted: G        W
  --------------------------------------------------------
  CPU 0/KVM/1537 just changed the state of lock:
  ffffa0f0cff965f0 (&vdev->irqlock){-...}-{2:2}, at:
  vfio_intx_handler+0x21/0xd0 [vfio_pci_core] but this lock took another,
  HARDIRQ-unsafe lock in the past: (fs_reclaim){+.+.}-{0:0}

and interrupts could create inverse lock ordering between them.

other info that might help us debug this:

  Possible interrupt unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(fs_reclaim);
			       local_irq_disable();
			       lock(&vdev->irqlock);
			       lock(fs_reclaim);
  
    lock(&vdev->irqlock);

  *** DEADLOCK ***

Have pcim_enable_device()'s release function, pcim_disable_device(), set
pci_dev.is_managed to false so that subsequent drivers using the same
struct pci_dev do not implicitly run into managed code.

Link: https://lore.kernel.org/r/20240905072556.11375-2-pstanner@redhat.com
Fixes: 25216afc9db5 ("PCI: Add managed pcim_intx()")
Reported-by: Alex Williamson 
Closes: https://lore.kernel.org/all/20240903094431.63551744.alex.williamson@redhat.com/
Suggested-by: Alex Williamson 
Signed-off-by: Philipp Stanner 
Signed-off-by: Bjorn Helgaas 
Tested-by: Alex Williamson 
Reviewed-by: Damien Le Moal

PCI: Add managed pcim_iomap_range()

2024-07-11T21:20:15+00:00

The only managed mapping function currently is pcim_iomap() which doesn't
allow for mapping an area starting at a certain offset, which many drivers
want.

Add pcim_iomap_range() as an exported function.

Link: https://lore.kernel.org/r/20240613115032.29098-13-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
Signed-off-by: Krzysztof Wilczyński 
Signed-off-by: Bjorn Helgaas

PCI: Remove legacy pcim_release()

2024-07-11T21:20:15+00:00

Thanks to preceding cleanup steps, pcim_release() is now not needed
anymore and can be replaced by pcim_disable_device(), which is the exact
counterpart to pcim_enable_device().

This permits removing further parts of the old PCI devres implementation.

Replace pcim_release() with pcim_disable_device().  Remove the now unused
function get_pci_dr().  Remove the struct pci_devres from pci.h.

Link: https://lore.kernel.org/r/20240613115032.29098-12-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
Signed-off-by: Krzysztof Wilczyński 
Signed-off-by: Bjorn Helgaas

PCI: Add managed pcim_intx()

2024-07-11T21:20:01+00:00

pci_intx() is a "hybrid" function, i.e., it is managed if
pcim_enable_device() has been called, but unmanaged otherwise.

Add pcim_intx(), which is always managed, and implement pci_intx() using
it.

Remove the now-unused struct pci_devres.orig_intx and .restore_intx and
find_pci_dr().

Link: https://lore.kernel.org/r/20240613115032.29098-11-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
[kwilczynski: squashed in
https://lore.kernel.org/r/426645d40776198e0fcc942f4a6cac4433c7a9aa.camel@redhat.com
to fix problem reported and tested by Ashish Kalra :
https://lore.kernel.org/r/20240708214656.4721-1-Ashish.Kalra@amd.com
https://lore.kernel.org/r/8c4634e9-4f02-4c54-9c89-d75e2f4bf026@amd.com/]
Signed-off-by: Krzysztof Wilczyński 
[bhelgaas: commit log]
Signed-off-by: Bjorn Helgaas

PCI: Give pcim_set_mwi() its own devres cleanup callback

2024-07-10T04:20:17+00:00

Managing pci_set_mwi() with devres can easily be done with its own
callback, without the necessity to store any state about it in a
device-related struct.

Remove the MWI state from struct pci_devres.  Give pcim_set_mwi() a
separate devres cleanup callback.

Link: https://lore.kernel.org/r/20240613115032.29098-10-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
Signed-off-by: Krzysztof Wilczyński 
Signed-off-by: Bjorn Helgaas

PCI: Move struct pci_devres.pinned bit to struct pci_dev

2024-07-10T04:20:10+00:00

The bit describing whether the PCI device is currently pinned is stored
in struct pci_devres. To clean up and simplify the PCI devres API, it's
better if this information is stored in struct pci_dev.

This will later permit simplifying pcim_enable_device().

Move the 'pinned' boolean bit to struct pci_dev.

Restructure bits in struct pci_dev so the pm / pme fields are next to
each other.

Link: https://lore.kernel.org/r/20240613115032.29098-9-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
Signed-off-by: Krzysztof Wilczyński 
Signed-off-by: Bjorn Helgaas

PCI: Remove struct pci_devres.enabled status bit

2024-07-10T04:20:06+00:00

The struct pci_devres has a separate boolean to track whether a device is
enabled. That, however, can easily be tracked in an agnostic manner through
the function pci_is_enabled().

Using it allows for simplifying the PCI devres implementation.

Replace the separate 'enabled' status bit from struct pci_devres with
calls to pci_is_enabled() at the appropriate places.

Link: https://lore.kernel.org/r/20240613115032.29098-8-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
Signed-off-by: Krzysztof Wilczyński 
Signed-off-by: Bjorn Helgaas

PCI: Add managed pcim_request_region()

2024-07-10T04:19:56+00:00

These existing functions:

  pci_request_region()
  pci_request_selected_regions()
  pci_request_selected_regions_exclusive()

are "hybrid" functions built on __pci_request_region() and are managed if
pcim_enable_device() has been called, but unmanaged otherwise.

Add these new functions:

  pcim_request_region()
  pcim_request_region_exclusive()

These are *always* managed and use the new pcim_addr_devres tracking
infrastructure instead of find_pci_dr() and struct pci_devres.region_mask.

Implement the hybrid functions using the new "pure" functions and remove
struct pci_devres.region_mask, which is no longer needed.

Link: https://lore.kernel.org/r/20240613115032.29098-6-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
Signed-off-by: Krzysztof Wilczyński 
[bhelgaas: commit log]
Signed-off-by: Bjorn Helgaas

PCI: Deprecate pcim_iomap_table(), pcim_iomap_regions_request_all()

2024-07-10T04:19:46+00:00

Deprecate pcim_iomap_table().  It returns a pointer to a table of
ioremapped BARs, or NULL if it fails.  This makes uses like this:

  addr = pcim_iomap_table(pdev)[0];

problematic because it causes a NULL pointer dereference on failure.
Callers should use pcim_iomap() instead.

Deprecate pcim_iomap_regions_request_all() because it is built on
__pci_request_region() and is managed if pcim_enable_device() has been
called, but unmanaged otherwise, which is prone to errors.

Callers should either use pcim_iomap_regions() to request and map BARs, or
use pcim_request_region() followed by pcim_iomap().

Link: https://lore.kernel.org/r/20240613115032.29098-5-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
Signed-off-by: Krzysztof Wilczyński 
[bhelgaas: commit log, sphinx markup]
Signed-off-by: Bjorn Helgaas

PCI: Add managed partial-BAR request and map infrastructure

2024-07-10T04:19:35+00:00

The pcim_iomap_devres table tracks entire-BAR mappings, so we can't use it
to build a managed version of pci_iomap_range(), which maps partial BARs.

Add struct pcim_addr_devres, which can track request and mapping of both
entire BARs and partial BARs.

Add the following internal devres functions based on struct
pcim_addr_devres:

  pcim_iomap_region()               # request & map entire BAR
  pcim_iounmap_region()             # unmap & release entire BAR
  pcim_request_region()             # request entire BAR
  pcim_release_region()             # release entire BAR
  pcim_request_all_regions()        # request all entire BARs
  pcim_release_all_regions()        # release all entire BARs

Rework the following public interfaces using the new infrastructure
listed above:

  pcim_iomap()                      # map partial BAR
  pcim_iounmap()                    # unmap partial BAR
  pcim_iomap_regions()              # request & map specified BARs
  pcim_iomap_regions_request_all()  # request all BARs, map specified BARs
  pcim_iounmap_regions()            # unmap & release specified BARs

Link: https://lore.kernel.org/r/20240613115032.29098-4-pstanner@redhat.com
Signed-off-by: Philipp Stanner 
Signed-off-by: Krzysztof Wilczyński 
[bhelgaas: commit log]
Signed-off-by: Bjorn Helgaas