[PATCH 0/3] Adding offset keeping option when mapping data via SWIOTLB.*

Jianxiong Gao jxgao at google.com
Wed Jan 27 19:38:26 EST 2021


NVMe driver and other applications may depend on the data offset
to operate correctly. Currently when unaligned data is mapped via
SWIOTLB, the data is mapped as slab aligned with the SWIOTLB. This
patch adds an option to make sure the mapped data preserves its
offset of the orginal addrss.

Without the patch when creating xfs formatted disk on NVMe backends,
with swiotlb=force in kernel boot option, creates the following error:
meta-data=/dev/nvme2n1           isize=512    agcount=4, agsize=131072 blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=1        finobt=1, sparse=0, rmapbt=0, refl
ink=0
data     =                       bsize=4096   blocks=524288, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0 ftype=1
log      =internal log           bsize=4096   blocks=2560, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0
mkfs.xfs: pwrite failed: Input/output error

Jianxiong Gao (3):
  Adding page_offset_mask to device_dma_parameters
  Add swiotlb offset preserving mapping when
    dma_dma_parameters->page_offset_mask is non zero.
  Adding device_dma_parameters->offset_preserve_mask to NVMe driver.

 drivers/nvme/host/pci.c     |  4 ++++
 include/linux/device.h      |  1 +
 include/linux/dma-mapping.h | 17 +++++++++++++++++
 kernel/dma/swiotlb.c        | 16 +++++++++++++++-
 4 files changed, 37 insertions(+), 1 deletion(-)

-- 
2.27.0




More information about the Linux-nvme mailing list