3.5.7.2 -stable review patch. If anyone has any objections, please let me know.
------------------ From: Johannes Berg <johannes.b...@intel.com> commit 7c34158231b2eda8dcbd297be2bb1559e69cb433 upstream. The RX replenish code doesn't handle DMA mapping failures, which will cause issues if there actually is a failure. This was reported by Shuah Khan who found a DMA mapping framework warning ("device driver failed to check map error"). Reported-by: Shuah Khan <shuah.k...@hp.com> Reviewed-by: Emmanuel Grumbach <emmanuel.grumb...@intel.com> Signed-off-by: Johannes Berg <johannes.b...@intel.com> [ herton: adapt for 3.5: * code to patch is on iwl-trans-pcie-rx.c * first function patched was named iwlagn_rx_allocate * adjust context ] Signed-off-by: Herton Ronaldo Krzesinski <herton.krzesin...@canonical.com> --- drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c | 23 ++++++++++++++++++++-- 1 file changed, 21 insertions(+), 2 deletions(-) diff --git a/drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c b/drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c index 5e18ff9..f774778 100644 --- a/drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c +++ b/drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c @@ -318,6 +318,14 @@ static void iwlagn_rx_allocate(struct iwl_trans *trans, gfp_t priority) rxb->page_dma = dma_map_page(trans->dev, page, 0, PAGE_SIZE << trans_pcie->rx_page_order, DMA_FROM_DEVICE); + if (dma_mapping_error(trans->dev, rxb->page_dma)) { + rxb->page = NULL; + spin_lock_irqsave(&rxq->lock, flags); + list_add(&rxb->list, &rxq->rx_used); + spin_unlock_irqrestore(&rxq->lock, flags); + __free_pages(page, trans_pcie->rx_page_order); + return; + } /* dma address must be no more than 36 bits */ BUG_ON(rxb->page_dma & ~DMA_BIT_MASK(36)); /* and also 256 byte aligned! */ @@ -467,8 +475,19 @@ static void iwl_rx_handle_rxbuf(struct iwl_trans *trans, dma_map_page(trans->dev, rxb->page, 0, PAGE_SIZE << trans_pcie->rx_page_order, DMA_FROM_DEVICE); - list_add_tail(&rxb->list, &rxq->rx_free); - rxq->free_count++; + if (dma_mapping_error(trans->dev, rxb->page_dma)) { + /* + * free the page(s) as well to not break + * the invariant that the items on the used + * list have no page(s) + */ + __free_pages(rxb->page, trans_pcie->rx_page_order); + rxb->page = NULL; + list_add_tail(&rxb->list, &rxq->rx_used); + } else { + list_add_tail(&rxb->list, &rxq->rx_free); + rxq->free_count++; + } } else list_add_tail(&rxb->list, &rxq->rx_used); spin_unlock_irqrestore(&rxq->lock, flags); -- 1.7.9.5 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/