Re: PCI DMA to small buffers on cache-incoherent arch

From: David S. Miller (davem@redhat.com)
Date: Sun Jun 09 2002 - 00:29:03 EST


   From: Roland Dreier <roland@topspin.com>
   Date: 08 Jun 2002 18:26:12 -0700

   Just to make sure I'm reading this correctly, you're saying that as
   long as a buffer is OK for DMA, it should be OK to use a
   sub-cache-line chunk as a DMA buffer via pci_map_single(), and
   accessing the rest of the cache line should be OK at any time before,
   during and after the DMA.
   
Yes.

       David> This means what MIPS is doing is wrong. For partial
       David> cacheline bits it can't do the invalidate thing.
   
   If I understand you, this means non-cache-coherent PPC is wrong as
   well -- pci_map_single() goes through consistent_sync() and turns
   into:
   
           case PCI_DMA_FROMDEVICE: /* invalidate only */
                   invalidate_dcache_range(start, end);
                   break;
   
   What alternate implementation are you proposing?

For non-cacheline aligned chunks in the range "start" to "end" you
must perform a cache writeback and invalidate. To preserve the data
outside of the DMA range.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sat Jun 15 2002 - 22:00:13 EST