Eric Lee / smarc-fsl-linux-kernel

06 Nov, 2015

1 commit

  • 2c302e7e4   Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc ... Browse Code »

    Pull sparc updates from David Miller:
    "Just a couple of fixes/cleanups:

    - Correct NUMA latency calculations on sparc64, from Nitin Gupta.

    - ASI_ST_BLKINIT_MRU_S value was wrong, from Rob Gardner.

    - Fix non-faulting load handling of non-quad values, also from Rob
    Gardner.

    - Cleanup VISsave assembler, from Sam Ravnborg.

    - Fix iommu-common code so it doesn't emit rediculous warnings on
    some architectures, particularly ARM"

    * git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
    sparc64: Fix numa distance values
    sparc64: Don't restrict fp regs for no-fault loads
    iommu-common: Fix error code used in iommu_tbl_range_{alloc,free}().
    sparc64: use ENTRY/ENDPROC in VISsave
    sparc64: Fix incorrect ASI_ST_BLKINIT_MRU_S value

    Linus Torvalds
     

05 Nov, 2015

1 commit

  • d618382ba   iommu-common: Fix error code used in iommu_tbl_range_{alloc,free}(). ... Browse Code »

    The value returned from iommu_tbl_range_alloc() (and the one passed
    in as a fourth argument to iommu_tbl_range_free) is not a DMA address,
    it is rather an index into the IOMMU page table.

    Therefore using DMA_ERROR_CODE is not appropriate.

    Use a more type matching error code define, IOMMU_ERROR_CODE, and
    update all users of this interface.

    Reported-by: Andre Przywara
    Signed-off-by: David S. Miller

    David S. Miller
     

23 Sep, 2015

1 commit

  • d046b770c   lib/iommu-common.c: do not try to deref a null iommu->lazy_flush() pointer when n < pool->hint ... Browse Code »

    The check for invoking iommu->lazy_flush() from iommu_tbl_range_alloc()
    has to be refactored so that we only call ->lazy_flush() if it is
    non-null.

    I had a sparc kernel that was crashing when I was trying to process some
    very large perf.data files- the crash happens when the scsi driver calls
    into dma_4v_map_sg and thus the iommu_tbl_range_alloc().

    Signed-off-by: Sowmini Varadhan
    Cc: Benjamin Herrenschmidt
    Cc: Guenter Roeck
    Cc: David S. Miller
    Cc:
    Signed-off-by: Andrew Morton
    Signed-off-by: Linus Torvalds

    Sowmini Varadhan
     

07 Aug, 2015

1 commit

21 Apr, 2015

2 commits

  • 7b3372d4c   iommu-common: rename iommu_pool_hash to iommu_hash_common ... Browse Code »

    When CONFIG_DEBUG_FORCE_WEAK_PER_CPU is set, the DEFINE_PER_CPU_SECTION
    macro will define an extern __pcpu_unique_##name variable that could
    conflict with the same definition in powerpc at this time. Avoid that
    conflict by renaming iommu_pool_hash in iommu-common.c

    Thanks to Guenter Roeck for catching this, and helping to test the fix.

    Signed-off-by: Sowmini Varadhan
    Tested-by: Guenter Roeck
    Reviewed-by: Guenter Roeck
    Signed-off-by: David S. Miller

    Sowmini Varadhan
     
  • b0cc836d3   iommu-common: fix x86_64 compiler warnings ... Browse Code »

    Declare iommu_large_alloc as static. Remove extern definition for
    iommu_tbl_pool_init().

    Signed-off-by: Sowmini Varadhan
    Tested-by: Guenter Roeck
    Reviewed-by: Guenter Roeck
    Signed-off-by: David S. Miller

    Sowmini Varadhan
     

19 Apr, 2015

3 commits

  • 2f0c0fdc0   iommu-common: Fix PARISC compile-time warnings ... Browse Code »

    Fixes warnings due to
    - no DMA_ERROR_CODE on PARISC,
    - sizeof (unsigned long) == 4 bytes on PARISC.

    Signed-off-by: Sowmini Varadhan
    Signed-off-by: David S. Miller

    Sowmini Varadhan
     
  • ff7d37a50   Break up monolithic iommu table/lock into finer graularity pools and lock ... Browse Code »

    Investigation of multithreaded iperf experiments on an ethernet
    interface show the iommu->lock as the hottest lock identified by
    lockstat, with something of the order of 21M contentions out of
    27M acquisitions, and an average wait time of 26 us for the lock.
    This is not efficient. A more scalable design is to follow the ppc
    model, where the iommu_map_table has multiple pools, each stretching
    over a segment of the map, and with a separate lock for each pool.
    This model allows for better parallelization of the iommu map search.

    This patch adds the iommu range alloc/free function infrastructure.

    Signed-off-by: Sowmini Varadhan
    Acked-by: Benjamin Herrenschmidt
    Signed-off-by: David S. Miller

    Sowmini Varadhan
     
  • c12f048ff   sparc: Revert generic IOMMU allocator. ... Browse Code »

    I applied the wrong version of this patch series, V4 instead
    of V10, due to a patchwork bundling snafu.

    Signed-off-by: David S. Miller

    David S. Miller
     

18 Apr, 2015

1 commit

17 Apr, 2015

1 commit

  • 10b88a4b1   sparc: Break up monolithic iommu table/lock into finer graularity pools and lock ... Browse Code »

    Investigation of multithreaded iperf experiments on an ethernet
    interface show the iommu->lock as the hottest lock identified by
    lockstat, with something of the order of 21M contentions out of
    27M acquisitions, and an average wait time of 26 us for the lock.
    This is not efficient. A more scalable design is to follow the ppc
    model, where the iommu_table has multiple pools, each stretching
    over a segment of the map, and with a separate lock for each pool.
    This model allows for better parallelization of the iommu map search.

    This patch adds the iommu range alloc/free function infrastructure.

    Signed-off-by: Sowmini Varadhan
    Signed-off-by: David S. Miller

    Sowmini Varadhan