linux-stable.git/arch/powerpc/kernel/eeh_driver.c, branch linux-5.2.y

powerpc/eeh: Clean up EEH PEs after recovery finishes

2019-10-07T16:59:17+00:00

[ Upstream commit 799abe283e5103d48e079149579b4f167c95ea0e ]

When the last device in an eeh_pe is removed the eeh_pe structure itself
(and any empty parents) are freed since they are no longer needed. This
results in a crash when a hotplug driver is involved since the following
may occur:

1. Device is suprise removed.
2. Driver performs an MMIO, which fails and queues and eeh_event.
3. Hotplug driver receives a hotplug interrupt and removes any
   pci_devs that were under the slot.
4. pci_dev is torn down and the eeh_pe is freed.
5. The EEH event handler thread processes the eeh_event and crashes
   since the eeh_pe pointer in the eeh_event structure is no
   longer valid.

Crashing is generally considered poor form. Instead of doing that use
the fact PEs are marked as EEH_PE_INVALID to keep them around until the
end of the recovery cycle, at which point we can safely prune any empty
PEs.

Signed-off-by: Oliver O'Halloran 
Signed-off-by: Michael Ellerman 
Link: https://lore.kernel.org/r/20190903101605.2890-2-oohall@gmail.com
Signed-off-by: Sasha Levin

powerpc/eeh: Clear stale EEH_DEV_NO_HANDLER flag

2019-10-07T16:59:11+00:00

[ Upstream commit aa06e3d60e245284d1e55497eb3108828092818d ]

The EEH_DEV_NO_HANDLER flag is used by the EEH system to prevent the
use of driver callbacks in drivers that have been bound part way
through the recovery process. This is necessary to prevent later stage
handlers from being called when the earlier stage handlers haven't,
which can be confusing for drivers.

However, the flag is set for all devices that are added after boot
time and only cleared at the end of the EEH recovery process. This
results in hot plugged devices erroneously having the flag set during
the first recovery after they are added (causing their driver's
handlers to be incorrectly ignored).

To remedy this, clear the flag at the beginning of recovery
processing. The flag is still cleared at the end of recovery
processing, although it is no longer really necessary.

Also clear the flag during eeh_handle_special_event(), for the same
reasons.

Signed-off-by: Sam Bobroff 
Signed-off-by: Michael Ellerman 
Link: https://lore.kernel.org/r/b8ca5629d27de74c957d4f4b250177d1b6fc4bbd.1565930772.git.sbobroff@linux.ibm.com
Signed-off-by: Sasha Levin

powerpc/eeh: Improve recovery of passed-through devices

2019-02-05T00:55:44+00:00

Currently, the EEH recovery process considers passed-through devices
as if they were not EEH-aware, which can cause them to be removed as
part of recovery.  Because device removal requires cooperation from
the guest, this may lead to the process stalling or deadlocking.
Also, if devices are removed on the host side, they will be removed
from their IOMMU group, making recovery in the guest impossible.

Therefore, alter the recovery process so that passed-through devices
are not removed but are instead left frozen (and marked isolated)
until the guest performs it's own recovery.  If firmware thaws a
passed-through PE because it's parent PE has been thawed (because it
was not passed through), re-freeze it.

Signed-off-by: Sam Bobroff 
Signed-off-by: Michael Ellerman

powerpc/eeh: Add include_passed to eeh_clear_pe_frozen_state()

2019-02-05T00:55:44+00:00

Add a parameter to eeh_clear_pe_frozen_state() that allows
passed-through PEs to be excluded. Update callers to always pass true
so that there is no change in behaviour.

This is to prepare for follow-up work for passed-through devices.

Signed-off-by: Sam Bobroff 
Signed-off-by: Michael Ellerman

powerpc/eeh: Add include_passed to eeh_pe_state_clear()

2019-02-05T00:55:43+00:00

Add a parameter to eeh_pe_state_clear() that allows passed-through PEs
to be excluded. Update callers to always pass true so that there is no
change in behaviour.

Also refactor to use direct traversal, to allow the removal of some
boilerplate.

This is to prepare for follow-up work for passed-through devices.

Signed-off-by: Sam Bobroff 
Signed-off-by: Michael Ellerman

powerpc/eeh: remove sw_state from eeh_unfreeze_pe()

2019-02-05T00:55:42+00:00

eeh_unfreeze_pe() performs two operations: unfreezing a PE (which may
cause firmware to unfreeze child PEs as well) and de-isolating the PE
and it's children.

To simplify this and support future work, separate out the
de-isolation and perform it at the call sites (when necessary).

There should be no change in behaviour.

Signed-off-by: Sam Bobroff 
Signed-off-by: Michael Ellerman

powerpc/eeh: Cleanup eeh_pe_clear_frozen_state()

2019-02-05T00:55:41+00:00

The 'clear_sw_state' parameter for eeh_pe_clear_frozen_state() is
redundant because it has no effect (except in the rare case of a
hardware error part way through unfreezing a tree of PEs, where it
would dangerously allow partial de-isolation before returning
failure).

It is passed down to __eeh_pe_clear_frozen_state(), and from there to
eeh_unfreeze_pe(), where it causes EEH_PE_ISOLATED to be removed
from the state of each PE during the traversal.  However, when the
traversal finishes, EEH_PE_ISOLATED is unconditionally removed by a
call to eeh_pe_state_clear() regardless of the parameter's value.

So remove the flag and pass false to eeh_unfreeze_pe() (to avoid the
rare case described above, as it was before the flag was introduced).
Also, perform the recursion directly in the function and eliminate a
bit of boilerplate.

There should be no change in functionality, except as mentioned above.

Signed-off-by: Sam Bobroff 
Signed-off-by: Michael Ellerman

powerpc/eeh: Declare pci_ers_result_name() as static

2018-11-25T06:11:21+00:00

Function pci_ers_result_name() is a static function, although not declared
as such. This was detected by sparse in the following warning

	arch/powerpc/kernel/eeh_driver.c:63:12: warning: symbol 'pci_ers_result_name' was not declared. Should it be static?

This patch simply declares the function a static.

Signed-off-by: Breno Leitao 
Signed-off-by: Michael Ellerman

powerpc/eeh: Cleanup control flow in eeh_handle_normal_event()

2018-10-13T11:21:25+00:00

Rather than mixing "if (state)" blocks and gotos, convert entirely to
"if (state)" blocks to make the state machine behaviour clearer.

Signed-off-by: Sam Bobroff 
Signed-off-by: Michael Ellerman

powerpc/eeh: Cleanup eeh_ops.wait_state()

2018-10-13T11:21:25+00:00

The wait_state member of eeh_ops does not need to be platform
dependent; it's just logic around eeh_ops.get_state(). Therefore,
merge the two (slightly different!) platform versions into a new
function, eeh_wait_state() and remove the eeh_ops member.

While doing this, also correct:
* The wait logic, so that it never waits longer than max_wait.
* The wait logic, so that it never waits less than
  EEH_STATE_MIN_WAIT_TIME.
* One call site where the result is treated like a bit field before
  it's checked for negative error values.
* In pseries_eeh_get_state(), rename the "state" parameter to "delay"
  because that's what it is.

Signed-off-by: Sam Bobroff 
Signed-off-by: Michael Ellerman