From: Nicolin Chen <nicolinc@nvidia.com>
The swiotlb does not support a mapping size > swiotlb_max_mapping_size().
On the other hand, with a 64KB PAGE_SIZE configuration, it's observed that
an NVME device can map a size between 300KB~512KB, which certainly failed
the swiotlb mappings, though the default pool of swiotlb has many slots:
systemd[1]: Started Journal Service.
=> nvme 0000:00:01.0: swiotlb buffer is full (sz: 327680 bytes), total 32768 (slots), used 32 (slots)
note: journal-offline[392] exited with irqs disabled
note: journal-offline[392] exited with preempt_count 1
Call trace:
[ 3.099918] swiotlb_tbl_map_single+0x214/0x240
[ 3.099921] iommu_dma_map_page+0x218/0x328
[ 3.099928] dma_map_page_attrs+0x2e8/0x3a0
[ 3.101985] nvme_prep_rq.part.0+0x408/0x878 [nvme]
[ 3.102308] nvme_queue_rqs+0xc0/0x300 [nvme]
[ 3.102313] blk_mq_flush_plug_list.part.0+0x57c/0x600
[ 3.102321] blk_add_rq_to_plug+0x180/0x2a0
[ 3.102323] blk_mq_submit_bio+0x4c8/0x6b8
[ 3.103463] __submit_bio+0x44/0x220
[ 3.103468] submit_bio_noacct_nocheck+0x2b8/0x360
[ 3.103470] submit_bio_noacct+0x180/0x6c8
[ 3.103471] submit_bio+0x34/0x130
[ 3.103473] ext4_bio_write_folio+0x5a4/0x8c8
[ 3.104766] mpage_submit_folio+0xa0/0x100
[ 3.104769] mpage_map_and_submit_buffers+0x1a4/0x400
[ 3.104771] ext4_do_writepages+0x6a0/0xd78
[ 3.105615] ext4_writepages+0x80/0x118
[ 3.105616] do_writepages+0x90/0x1e8
[ 3.105619] filemap_fdatawrite_wbc+0x94/0xe0
[ 3.105622] __filemap_fdatawrite_range+0x68/0xb8
[ 3.106656] file_write_and_wait_range+0x84/0x120
[ 3.106658] ext4_sync_file+0x7c/0x4c0
[ 3.106660] vfs_fsync_range+0x3c/0xa8
[ 3.106663] do_fsync+0x44/0xc0
Since untrusted devices might go down the swiotlb pathway with dma-iommu,
these devices should not map a size larger than swiotlb_max_mapping_size.
To fix this bug, add iommu_dma_max_mapping_size() for untrusted devices to
take into account swiotlb_max_mapping_size() v.s. iova_rcache_range() from
the iommu_dma_opt_mapping_size().
Fixes: 82612d66d51d ("iommu: Allow the dma-iommu api to use bounce buffers")
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Link: https://lore.kernel.org/r/ee51a3a5c32cf885b18f6416171802669f4a718a.1707851466.git.nicolinc@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
---
drivers/iommu/dma-iommu.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
index 50ccc4f1ef81..7d1a20da6d94 100644
--- a/drivers/iommu/dma-iommu.c
+++ b/drivers/iommu/dma-iommu.c
@@ -1706,6 +1706,13 @@ static size_t iommu_dma_opt_mapping_size(void)
return iova_rcache_range();
}
+static size_t iommu_dma_max_mapping_size(struct device *dev)
+{
+ if (is_swiotlb_active(dev) && dev_is_untrusted(dev))
+ return swiotlb_max_mapping_size(dev);
+ return SIZE_MAX;
+}
+
static const struct dma_map_ops iommu_dma_ops = {
.flags = DMA_F_PCI_P2PDMA_SUPPORTED,
.alloc = iommu_dma_alloc,
@@ -1728,6 +1735,7 @@ static const struct dma_map_ops iommu_dma_ops = {
.unmap_resource = iommu_dma_unmap_resource,
.get_merge_boundary = iommu_dma_get_merge_boundary,
.opt_mapping_size = iommu_dma_opt_mapping_size,
+ .max_mapping_size = iommu_dma_max_mapping_size,
};
/*
--
2.44.0.rc0.258.g7320e95886-googOn Wed, Feb 21, 2024 at 11:35:04AM +0000, Will Deacon wrote:
> +static size_t iommu_dma_max_mapping_size(struct device *dev)
> +{
> + if (is_swiotlb_active(dev) && dev_is_untrusted(dev))
> + return swiotlb_max_mapping_size(dev);
Curious: do we really need both checks here? If swiotlb is active
for a device (for whatever reason), aren't we then always bound
by the max size? If not please add a comment explaining it.
On 27/02/2024 3:40 pm, Christoph Hellwig wrote:
> On Wed, Feb 21, 2024 at 11:35:04AM +0000, Will Deacon wrote:
>> +static size_t iommu_dma_max_mapping_size(struct device *dev)
>> +{
>> + if (is_swiotlb_active(dev) && dev_is_untrusted(dev))
>> + return swiotlb_max_mapping_size(dev);
>
> Curious: do we really need both checks here? If swiotlb is active
> for a device (for whatever reason), aren't we then always bound
> by the max size? If not please add a comment explaining it.
>
Oh, good point - if we have an untrusted device but SWIOTLB isn't
initialised for whatever reason, then it doesn't matter what
max_mapping_size returns because iommu_dma_map_page() is going to bail
out regardless.
Thanks,
Robin.
On Tue, Feb 27, 2024 at 03:53:05PM +0000, Robin Murphy wrote:
> On 27/02/2024 3:40 pm, Christoph Hellwig wrote:
> > On Wed, Feb 21, 2024 at 11:35:04AM +0000, Will Deacon wrote:
> > > +static size_t iommu_dma_max_mapping_size(struct device *dev)
> > > +{
> > > + if (is_swiotlb_active(dev) && dev_is_untrusted(dev))
> > > + return swiotlb_max_mapping_size(dev);
> >
> > Curious: do we really need both checks here? If swiotlb is active
> > for a device (for whatever reason), aren't we then always bound
> > by the max size? If not please add a comment explaining it.
> >
>
> Oh, good point - if we have an untrusted device but SWIOTLB isn't
> initialised for whatever reason, then it doesn't matter what
> max_mapping_size returns because iommu_dma_map_page() is going to bail out
> regardless.
Makes sense. Since this is all internal to the IOMMU DMA code, I can just
drop the first part of the check.
I'll get a v5 out shortly.
Will
On 21/02/2024 11:35 am, Will Deacon wrote:
> From: Nicolin Chen <nicolinc@nvidia.com>
>
> The swiotlb does not support a mapping size > swiotlb_max_mapping_size().
> On the other hand, with a 64KB PAGE_SIZE configuration, it's observed that
> an NVME device can map a size between 300KB~512KB, which certainly failed
> the swiotlb mappings, though the default pool of swiotlb has many slots:
> systemd[1]: Started Journal Service.
> => nvme 0000:00:01.0: swiotlb buffer is full (sz: 327680 bytes), total 32768 (slots), used 32 (slots)
> note: journal-offline[392] exited with irqs disabled
> note: journal-offline[392] exited with preempt_count 1
>
> Call trace:
> [ 3.099918] swiotlb_tbl_map_single+0x214/0x240
> [ 3.099921] iommu_dma_map_page+0x218/0x328
> [ 3.099928] dma_map_page_attrs+0x2e8/0x3a0
> [ 3.101985] nvme_prep_rq.part.0+0x408/0x878 [nvme]
> [ 3.102308] nvme_queue_rqs+0xc0/0x300 [nvme]
> [ 3.102313] blk_mq_flush_plug_list.part.0+0x57c/0x600
> [ 3.102321] blk_add_rq_to_plug+0x180/0x2a0
> [ 3.102323] blk_mq_submit_bio+0x4c8/0x6b8
> [ 3.103463] __submit_bio+0x44/0x220
> [ 3.103468] submit_bio_noacct_nocheck+0x2b8/0x360
> [ 3.103470] submit_bio_noacct+0x180/0x6c8
> [ 3.103471] submit_bio+0x34/0x130
> [ 3.103473] ext4_bio_write_folio+0x5a4/0x8c8
> [ 3.104766] mpage_submit_folio+0xa0/0x100
> [ 3.104769] mpage_map_and_submit_buffers+0x1a4/0x400
> [ 3.104771] ext4_do_writepages+0x6a0/0xd78
> [ 3.105615] ext4_writepages+0x80/0x118
> [ 3.105616] do_writepages+0x90/0x1e8
> [ 3.105619] filemap_fdatawrite_wbc+0x94/0xe0
> [ 3.105622] __filemap_fdatawrite_range+0x68/0xb8
> [ 3.106656] file_write_and_wait_range+0x84/0x120
> [ 3.106658] ext4_sync_file+0x7c/0x4c0
> [ 3.106660] vfs_fsync_range+0x3c/0xa8
> [ 3.106663] do_fsync+0x44/0xc0
>
> Since untrusted devices might go down the swiotlb pathway with dma-iommu,
> these devices should not map a size larger than swiotlb_max_mapping_size.
>
> To fix this bug, add iommu_dma_max_mapping_size() for untrusted devices to
> take into account swiotlb_max_mapping_size() v.s. iova_rcache_range() from
> the iommu_dma_opt_mapping_size().
On the basis that this is at least far closer to correct than doing nothing,
Acked-by: Robin Murphy <robin.murphy@arm.com>
TBH I'm scared to think about theoretical correctness for all the
interactions between the IOVA granule and min_align_mask, since just the
SWIOTLB stuff is bad enough, even before you realise the ways that the
IOVA allocation isn't necessarily right either. However I reckon as long
as we don't ever see a granule smaller than IO_TLB_SIZE, and/or a
min_align_mask larger than a granule, then this should probably work
well enough as-is.
Cheers,
Robin.
> Fixes: 82612d66d51d ("iommu: Allow the dma-iommu api to use bounce buffers")
> Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
> Link: https://lore.kernel.org/r/ee51a3a5c32cf885b18f6416171802669f4a718a.1707851466.git.nicolinc@nvidia.com
> Signed-off-by: Will Deacon <will@kernel.org>
> ---
> drivers/iommu/dma-iommu.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
> index 50ccc4f1ef81..7d1a20da6d94 100644
> --- a/drivers/iommu/dma-iommu.c
> +++ b/drivers/iommu/dma-iommu.c
> @@ -1706,6 +1706,13 @@ static size_t iommu_dma_opt_mapping_size(void)
> return iova_rcache_range();
> }
>
> +static size_t iommu_dma_max_mapping_size(struct device *dev)
> +{
> + if (is_swiotlb_active(dev) && dev_is_untrusted(dev))
> + return swiotlb_max_mapping_size(dev);
> + return SIZE_MAX;
> +}
> +
> static const struct dma_map_ops iommu_dma_ops = {
> .flags = DMA_F_PCI_P2PDMA_SUPPORTED,
> .alloc = iommu_dma_alloc,
> @@ -1728,6 +1735,7 @@ static const struct dma_map_ops iommu_dma_ops = {
> .unmap_resource = iommu_dma_unmap_resource,
> .get_merge_boundary = iommu_dma_get_merge_boundary,
> .opt_mapping_size = iommu_dma_opt_mapping_size,
> + .max_mapping_size = iommu_dma_max_mapping_size,
> };
>
> /*
© 2016 - 2026 Red Hat, Inc.