Re: [PoC] arm: dma-mapping: direct: Apply dma_pfn_offset only when it is valid

From: Peter Ujfalusi <peter.ujfalusi@ti.com>
To: Christoph Hellwig <hch@lst.de>
Cc: Robin Murphy <robin.murphy@arm.com>, <vigneshr@ti.com>,
	<konrad.wilk@oracle.com>, <linux@armlinux.org.uk>,
	<linux-kernel@vger.kernel.org>,
	<iommu@lists.linux-foundation.org>,
	<linux-arm-kernel@lists.infradead.org>, <rogerq@ti.com>,
	<robh@kernel.org>
Subject: Re: [PoC] arm: dma-mapping: direct: Apply dma_pfn_offset only when it is valid
Date: Thu, 30 Jan 2020 15:04:37 +0200	[thread overview]
Message-ID: <b2b1cb21-3aae-2181-fd79-f63701f283c0@ti.com> (raw)
In-Reply-To: <20200130075332.GA30735@lst.de>

On 30/01/2020 9.53, Christoph Hellwig wrote:
> [skipping the DT bits, as I'm everything but an expert on that..]
> 
> On Mon, Jan 27, 2020 at 04:00:30PM +0200, Peter Ujfalusi wrote:
>> I agree on the phys_to_dma(). It should fail for addresses which does
>> not fall into any of the ranges.
>> It is just a that we in Linux don't have the concept atm for ranges, we
>> have only _one_ range which applies to every memory address.
> 
> what does atm here mean?

struct device have only single dma_pfn_offset, one can not have multiple
ranges defined. If we have then only the first is taken and the physical
address and dma address is discarded, only the dma_pfn_offset is stored
and used.

> We have needed multi-range support for quite a while, as common broadcom
> SOCs do need it.  So patches for that are welcome at least from the
> DMA layer perspective (kinda similar to your pseudo code earlier)

But do they have dma_pfn_offset != 0?

>>> Nobody's disputing that the current dma_direct_supported()
>>> implementation is broken for the case where ZONE_DMA itself is offset
>>> from PA 0; the more pressing question is why Christoph's diff, which was
>>> trying to take that into account, still didn't work.
>>
>> I understand that this is a bit more complex than I interpret it, but
>> the k2g is broken and currently the simplest way to make it work is to
>> use the arm dma_ops in case the pfn_offset is not 0.
>> It will be easy to test dma-direct changes trying to address the issue
>> in hand, but will allow k2g to be usable at the same time.
> 
> Well, using the legacy arm dma ops means we can't use swiotlb if there
> is an offset, which is also wrong for lots of common cases, including
> the Rpi 4.  I'm still curious why my patch didn't work, as I thought
> it should.

The dma_pfn_offset is _still_ applied to the mask we are trying to set
(and validate) via dma-direct.

in dma_direct_supported:
mask == 0xffffffff // DMA_BIT_MASK(32)
dev->dma_pfn_offset == 0x780000 // Keystone 2
min_mask == 0xffffff

tmp_mask = __phys_to_dma(dev, min_mask);
tmp_mask == 0xff880ffffff

within __phys_to_dma() converts the min_mask to pfn and calls
pfn_to_dma() which does:
if (dev)
	pfn -= dev->dma_pfn_offset;

the returned pfn is then converted back to address.

the mask (0xffffffff) is well under the tmp_mask (0xff880ffffff) so
dma_direct_supported() will tell us that DMA is not supported for
DMA_BIT_MASK(32), which is not true, because DMA is supporting 32 bits.

> We'll need to find the minimum change to make it work
> for now without switching ops, even if it isn't the correct one, and
> then work from there.

Sure, but can we fix the regression by reverting to arm_ops for now only
if dma_pfn_offset is not 0? It used to work fine in the past at least it
appeared to work on K2 platforms.

- Péter

Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki.
Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki