parport_pc: set properly the dma_mask for parport_pc device
FUJITA Tomonori [Mon, 22 Jun 2009 15:54:27 +0000 (16:54 +0100)]
parport_pc_probe_port() creates the own 'parport_pc' device if the
device argument is NULL. Then parport_pc_probe_port() doesn't
initialize the dma_mask and coherent_dma_mask of the device and calls
dma_alloc_coherent with it. dma_alloc_coherent fails because
dma_alloc_coherent() doesn't accept the uninitialized dma_mask:

http://lkml.org/lkml/2009/6/16/150

Long ago, X86_32 and X86_64 had the own dma_alloc_coherent
implementations; X86_32 accepted a device having dma_mask that is not
initialized however X86_64 didn't. When we merged them, we chose to
prohibit a device having dma_mask that is not initialized. I think
that it's good to require drivers to set up dma_mask (and
coherent_dma_mask) properly if the drivers want DMA.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Reported-by: Malcom Blaney <malcolm.blaney@maptek.com.au>
Tested-by: Malcom Blaney <malcolm.blaney@maptek.com.au>
Cc: stable@kernel.org
Signed-off-by: Alan Cox <alan@linux.intel.com>
Acked-by: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/parport/parport_pc.c

index 7f1cca7..1032d5f 100644 (file)
@@ -2290,6 +2290,9 @@ struct parport *parport_pc_probe_port(unsigned long int base,
                if (IS_ERR(pdev))
                        return NULL;
                dev = &pdev->dev;
+
+               dev->coherent_dma_mask = DMA_BIT_MASK(24);
+               dev->dma_mask = &dev->coherent_dma_mask;
        }
 
        ops = kmalloc(sizeof(struct parport_operations), GFP_KERNEL);