Re: [PATCH v2] r8169: set coherent DMA mask as well as streaming DMA mask

From: David Miller
Date: Sat Oct 15 2016 - 17:31:22 EST


From: Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx>
Date: Fri, 14 Oct 2016 14:48:51 +0100

>
>> On 14 Oct 2016, at 14:42, David Laight <David.Laight@xxxxxxxxxx> wrote:
>>
>> From: Of Ard Biesheuvel
>>> Sent: 14 October 2016 14:41
>>> PCI devices that are 64-bit DMA capable should set the coherent
>>> DMA mask as well as the streaming DMA mask. On some architectures,
>>> these are managed separately, and so the coherent DMA mask will be
>>> left at its default value of 32 if it is not set explicitly. This
>>> results in errors such as
>>>
>>> r8169 Gigabit Ethernet driver 2.3LK-NAPI loaded
>>> hwdev DMA mask = 0x00000000ffffffff, dev_addr = 0x00000080fbfff000
>>> swiotlb: coherent allocation failed for device 0000:02:00.0 size=4096
>>> CPU: 0 PID: 1062 Comm: systemd-udevd Not tainted 4.8.0+ #35
>>> Hardware name: AMD Seattle/Seattle, BIOS 10:53:24 Oct 13 2016
>>>
>>> on systems without memory that is 32-bit addressable by PCI devices.
>>>
>>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx>
>>> ---
>>> v2: dropped the hunk that sets the coherent DMA mask to DMA_BIT_MASK(32),
>>> which is unnecessary given that it is the default
>>>
>>> drivers/net/ethernet/realtek/r8169.c | 3 ++-
>>> 1 file changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c
>>> index e55638c7505a..bf000d819a21 100644
>>> --- a/drivers/net/ethernet/realtek/r8169.c
>>> +++ b/drivers/net/ethernet/realtek/r8169.c
>>> @@ -8273,7 +8273,8 @@ static int rtl_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
>>> if ((sizeof(dma_addr_t) > 4) &&
>>> (use_dac == 1 || (use_dac == -1 && pci_is_pcie(pdev) &&
>>> tp->mac_version >= RTL_GIGA_MAC_VER_18)) &&
>>> - !pci_set_dma_mask(pdev, DMA_BIT_MASK(64))) {
>>> + !pci_set_dma_mask(pdev, DMA_BIT_MASK(64)) &&
>>> + !pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64))) {
>>
>> Isn't there a dma_set_mask_and_coherent() function ?
>>
>
> Not of the pci_xxx variety afaik

You can often use the "dev_*" variants intechangably with the pci_*()
ones.

In fact you'll find that for several architectures pci_*() is
implemented via calls to dev_*().