Skip to contentSkip to conversation

iommu/amd: fix a crash in iova_magazine_free_pfns

PATCH In review Qian Cai 4 participants 4 messages Jul 11, 2019 – Jul 22, 2019 IOMMU

lore
Share & tools
lore

Qian Cai

author
#0 lore raw

PATCH v2 1/1 iommu/amd: fix a crash in iova_magazine_free_pfns

The commit b3aa14f02254 ("iommu: remove the mapping_error dma_map_ops method") incorrectly changed the checking from dma_ops_alloc_iova() in map_sg() causes a crash under memory pressure as dma_ops_alloc_iova() never return DMA_MAPPING_ERROR on failure but 0, so the error handling is all wrong.

kernel BUG at drivers/iommu/iova.c:801!

    Workqueue: kblockd blk_mq_run_work_fn
    RIP: 0010:iova_magazine_free_pfns+0x7d/0xc0
    Call Trace:
     free_cpu_cached_iovas+0xbd/0x150
     alloc_iova_fast+0x8c/0xba
     dma_ops_alloc_iova.isra.6+0x65/0xa0
     map_sg+0x8c/0x2a0
     scsi_dma_map+0xc6/0x160
     pqi_aio_submit_io+0x1f6/0x440 [smartpqi]
     pqi_scsi_queue_command+0x90c/0xdd0 [smartpqi]
     scsi_queue_rq+0x79c/0x1200
     blk_mq_dispatch_rq_list+0x4dc/0xb70
     blk_mq_sched_dispatch_requests+0x249/0x310
     __blk_mq_run_hw_queue+0x128/0x200
     blk_mq_run_work_fn+0x27/0x30
     process_one_work+0x522/0xa10
     worker_thread+0x63/0x5b0
     kthread+0x1d2/0x1f0
     ret_from_fork+0x22/0x40
Fixes b3aa14f02254 ("iommu: remove the mapping_error dma_map_ops method")

---

v2: Fix the offensive commit directly.

 drivers/iommu/amd_iommu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iommu/amd_iommu.c +1 -1
index 73740b969e62..b607a92791d3 100644
@@ -2533,7 +2533,7 @@ static int map_sg(struct device *dev, struct scatterlist *sglist,
 	npages = sg_num_pages(dev, sglist, nelems);
 
 	address = dma_ops_alloc_iova(dev, dma_dom, npages, dma_mask);
-	if (address == DMA_MAPPING_ERROR)
+	if (!address)
 		goto out_err;
 
 	prot = dir2prot(direction);
Signature · 1 line
1.8.3.1
#2 lore raw
In reply to Qian Cai
Quoted 6 lines · Qian Cai

The commit b3aa14f02254 ("iommu: remove the mapping_error dma_map_ops method") incorrectly changed the checking from dma_ops_alloc_iova() in map_sg() causes a crash under memory pressure as dma_ops_alloc_iova() never return DMA_MAPPING_ERROR on failure but 0, so the error handling is all wrong.

This one seems to have fallen through the cracks.

Applied directly.

Maybe it's hiding in some fixes tree that I haven't gotten a pull request for yet?

Linus

#3 lore raw

Hi Linus,

Quoted 3 lines · Linus Torvalds

This one seems to have fallen through the cracks.

Applied directly.

Thanks!

Quoted 2 lines

Maybe it's hiding in some fixes tree that I haven't gotten a pull request for yet?

No, I havn't had it applied anywhere yet. I usually don't pay close attention to patches sent to me during the merge window, as I won't update my tree anyway before -rc1.

I only take care of important fixes, but this one seems to have fallen through my heuristic. Thanks for applying it directly.

Regards,

Joerg