1. 04 3月, 2013 5 次提交
  2. 28 2月, 2013 4 次提交
    • J
      iwlwifi: use coherent DMA memory for command header · 38c0f334
      Johannes Berg 提交于
      Recently in commit 8a964f44
      ("iwlwifi: always copy first 16 bytes of commands") we fixed
      the problem that the hardware writes back to the command and
      that could overwrite parts of the data that was still needed
      and would thus be corrupted.
      
      Investigating this problem more closely we found that this
      write-back isn't really ordered very well with respect to
      other DMA traffic. Therefore, it sometimes happened that the
      write-back occurred after unmapping the command again which
      is clearly an issue and could corrupt the next allocation
      that goes to that spot, or (better) cause IOMMU faults.
      
      To fix this, allocate coherent memory for the first 16 bytes
      of each command, containing the write-back part, and use it
      for all queues. All the dynamic DMA mappings only need to be
      TO_DEVICE then. This ensures that even when the write-back
      happens "too late" it can't hit memory that has been freed
      or a mapping that doesn't exist any more.
      
      Since now the actual command is no longer modified, we can
      also remove CMD_WANT_HCMD and get rid of the DMA sync that
      was necessary to update the scratch pointer.
      Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
      Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
      38c0f334
    • E
      iwlwifi: disable 8K A-MSDU by default · aed7d9ac
      Emmanuel Grumbach 提交于
      Supporting 8K A-MSDU means that we need to allocate order 1
      pages for every Rx packet. Even when there is no traffic.
      This adds stress on the memory manager. The handling of
      compound pages is also less trivial for the memory manager
      and not using them will make the allocation code run faster
      although I didn't really measure.
      Eric also pointed out that having huge buffers with little
      data in them is not very nice towards the TCP stack since
      the truesize of the skb is huge. This doesn't allow TCP
      to have a big Rx window.
      See https://patchwork.kernel.org/patch/2167711/ for details.
      
      Note that very few vendors will actually send A-MSDU.
      Disable this feature by default.
      Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
      Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
      aed7d9ac
    • J
      iwlwifi: rename IWL_MAX_CMD_TFDS to IWL_MAX_CMD_TBS_PER_TFD · 1afbfb60
      Johannes Berg 提交于
      The IWL_MAX_CMD_TFDS name for this constant is wrong, the
      constant really indicates how many TBs we can use in the
      driver for a single command TFD, rename the constant and
      also add a comment explaining it.
      Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
      Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
      1afbfb60
    • J
      iwlwifi: don't map complete commands bidirectionally · 98891754
      Johannes Berg 提交于
      The reason we mapped them bidirectionally was that not doing
      so had caused IOMMU exceptions, due to the fact that the HW
      writes back into the command. Now that the first part of the
      command including the write-back part is always in the first
      buffer, we don't need to map the remaining buffer(s) bidi
      and can get rid of the special-casing for commands.
      
      This is a requisite patch for another one to fix DMA mapping.
      Reviewed-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
      Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
      98891754
  3. 27 2月, 2013 4 次提交
  4. 23 2月, 2013 3 次提交
  5. 19 2月, 2013 19 次提交
  6. 18 2月, 2013 5 次提交