diff options
author | Sam Bobroff <sbobroff@linux.ibm.com> | 2018-11-29 14:16:41 +1100 |
---|---|---|
committer | Michael Ellerman <mpe@ellerman.id.au> | 2019-02-05 11:55:44 +1100 |
commit | 1ef52073fd25ea97090eaff2c8b528ebf401a12a (patch) | |
tree | 85d99a82f5dbd6386015ee73b704c744113b41f7 /arch/powerpc/kernel/eeh.c | |
parent | 4d8e325d9df32ef00136d7885f0c65bf124edd22 (diff) | |
download | linux-1ef52073fd25ea97090eaff2c8b528ebf401a12a.tar.bz2 |
powerpc/eeh: Improve recovery of passed-through devices
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 <sbobroff@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Diffstat (limited to 'arch/powerpc/kernel/eeh.c')
-rw-r--r-- | arch/powerpc/kernel/eeh.c | 47 |
1 files changed, 37 insertions, 10 deletions
diff --git a/arch/powerpc/kernel/eeh.c b/arch/powerpc/kernel/eeh.c index 8d32587b07dc..416d1ef49762 100644 --- a/arch/powerpc/kernel/eeh.c +++ b/arch/powerpc/kernel/eeh.c @@ -877,6 +877,24 @@ static void *eeh_set_dev_freset(struct eeh_dev *edev, void *flag) return NULL; } +static void eeh_pe_refreeze_passed(struct eeh_pe *root) +{ + struct eeh_pe *pe; + int state; + + eeh_for_each_pe(root, pe) { + if (eeh_pe_passed(pe)) { + state = eeh_ops->get_state(pe, NULL); + if (state & + (EEH_STATE_MMIO_ACTIVE | EEH_STATE_MMIO_ENABLED)) { + pr_info("EEH: Passed-through PE PHB#%x-PE#%x was thawed by reset, re-freezing for safety.\n", + pe->phb->global_number, pe->addr); + eeh_pe_set_option(pe, EEH_OPT_FREEZE_PE); + } + } + } +} + /** * eeh_pe_reset_full - Complete a full reset process on the indicated PE * @pe: EEH PE @@ -889,7 +907,7 @@ static void *eeh_set_dev_freset(struct eeh_dev *edev, void *flag) * * This function will attempt to reset a PE three times before failing. */ -int eeh_pe_reset_full(struct eeh_pe *pe) +int eeh_pe_reset_full(struct eeh_pe *pe, bool include_passed) { int reset_state = (EEH_PE_RESET | EEH_PE_CFG_BLOCKED); int type = EEH_RESET_HOT; @@ -911,11 +929,11 @@ int eeh_pe_reset_full(struct eeh_pe *pe) /* Make three attempts at resetting the bus */ for (i = 0; i < 3; i++) { - ret = eeh_pe_reset(pe, type); + ret = eeh_pe_reset(pe, type, include_passed); if (ret) break; - ret = eeh_pe_reset(pe, EEH_RESET_DEACTIVATE); + ret = eeh_pe_reset(pe, EEH_RESET_DEACTIVATE, include_passed); if (ret) break; @@ -936,6 +954,12 @@ int eeh_pe_reset_full(struct eeh_pe *pe) __func__, state, pe->phb->global_number, pe->addr, (i + 1)); } + /* Resetting the PE may have unfrozen child PEs. If those PEs have been + * (potentially) passed through to a guest, re-freeze them: + */ + if (!include_passed) + eeh_pe_refreeze_passed(pe); + eeh_pe_state_clear(pe, reset_state, true); return ret; } @@ -1611,13 +1635,12 @@ int eeh_pe_get_state(struct eeh_pe *pe) } EXPORT_SYMBOL_GPL(eeh_pe_get_state); -static int eeh_pe_reenable_devices(struct eeh_pe *pe) +static int eeh_pe_reenable_devices(struct eeh_pe *pe, bool include_passed) { struct eeh_dev *edev, *tmp; struct pci_dev *pdev; int ret = 0; - /* Restore config space */ eeh_pe_restore_bars(pe); /* @@ -1638,9 +1661,13 @@ static int eeh_pe_reenable_devices(struct eeh_pe *pe) } /* The PE is still in frozen state */ - ret = eeh_unfreeze_pe(pe); + if (include_passed || !eeh_pe_passed(pe)) { + ret = eeh_unfreeze_pe(pe); + } else + pr_info("EEH: Note: Leaving passthrough PHB#%x-PE#%x frozen.\n", + pe->phb->global_number, pe->addr); if (!ret) - eeh_pe_state_clear(pe, EEH_PE_ISOLATED, true); + eeh_pe_state_clear(pe, EEH_PE_ISOLATED, include_passed); return ret; } @@ -1654,7 +1681,7 @@ static int eeh_pe_reenable_devices(struct eeh_pe *pe) * indicated type, either fundamental reset or hot reset. * PE reset is the most important part for error recovery. */ -int eeh_pe_reset(struct eeh_pe *pe, int option) +int eeh_pe_reset(struct eeh_pe *pe, int option, bool include_passed) { int ret = 0; @@ -1668,11 +1695,11 @@ int eeh_pe_reset(struct eeh_pe *pe, int option) switch (option) { case EEH_RESET_DEACTIVATE: ret = eeh_ops->reset(pe, option); - eeh_pe_state_clear(pe, EEH_PE_CFG_BLOCKED, true); + eeh_pe_state_clear(pe, EEH_PE_CFG_BLOCKED, include_passed); if (ret) break; - ret = eeh_pe_reenable_devices(pe); + ret = eeh_pe_reenable_devices(pe, include_passed); break; case EEH_RESET_HOT: case EEH_RESET_FUNDAMENTAL: |