Module edac_core.ko uses call_rcu() callbacks in edac_device.c, edac_mc.c and edac_pci.c. They all uses a wait_for_completion scheme, but this scheme it not 100% safe on multiple CPUs. See the _rcu_barrier() implementation which explains why extra precausion is needed. The patch adds a comment about rcu_barrier() and as a precausion calls rcu_barrier(). A maintainer needs to look at removing the wait_for_completion code. Signed-off-by: Jesper Dangaard Brouer <hawk@xxxxxxx> --- drivers/edac/edac_device.c | 5 +++++ drivers/edac/edac_mc.c | 5 +++++ drivers/edac/edac_pci.c | 5 +++++ 3 files changed, 15 insertions(+), 0 deletions(-) diff --git a/drivers/edac/edac_device.c b/drivers/edac/edac_device.c index b02a6a6..5e831c9 100644 --- a/drivers/edac/edac_device.c +++ b/drivers/edac/edac_device.c @@ -373,6 +373,11 @@ static void del_edac_device_from_global_list(struct edac_device_ctl_info init_completion(&edac_device->removal_complete); call_rcu(&edac_device->rcu, complete_edac_device_list_del); wait_for_completion(&edac_device->removal_complete); + + /* hawk@xxxxxxx 2009-06-22: I think that rcu_barrier() should + * be used instead of wait_for_completion, because + * rcu_barrier() take multiple CPUs into account */ + rcu_barrier(); } /* diff --git a/drivers/edac/edac_mc.c b/drivers/edac/edac_mc.c index 335b7eb..edcce41 100644 --- a/drivers/edac/edac_mc.c +++ b/drivers/edac/edac_mc.c @@ -428,6 +428,11 @@ static void del_mc_from_global_list(struct mem_ctl_info *mci) init_completion(&mci->complete); call_rcu(&mci->rcu, complete_mc_list_del); wait_for_completion(&mci->complete); + + /* hawk@xxxxxxx 2009-06-22: I think that rcu_barrier() should + * be used instead of wait_for_completion, because + * rcu_barrier() take multiple CPUs into account */ + rcu_barrier(); } /** diff --git a/drivers/edac/edac_pci.c b/drivers/edac/edac_pci.c index 30b585b..d0eb8c9 100644 --- a/drivers/edac/edac_pci.c +++ b/drivers/edac/edac_pci.c @@ -188,6 +188,11 @@ static void del_edac_pci_from_global_list(struct edac_pci_ctl_info *pci) init_completion(&pci->complete); call_rcu(&pci->rcu, complete_edac_pci_list_del); wait_for_completion(&pci->complete); + + /* hawk@xxxxxxx 2009-06-22: I think that rcu_barrier() should + * be used instead of wait_for_completion, because + * rcu_barrier() take multiple CPUs into account */ + rcu_barrier(); } #if 0 -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
