Skip to contentSkip to conversation

ata: ahci: Respect bus DMA constraints

RFC PATCH New [email protected] 4 participants 18 messages Mar 7, 2019 – Apr 9, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

From: Marek Vasut <[email protected]>

Since commit 6c2fb2ea7636 ("of/device: Set bus DMA mask as appropriate"), the upstream bus can constraint device DMA range. Respect that constraint and do not change the device DMA masks if they were already set.

This is applicable e.g. on systems where the PCIe controller cannot expose the full address space range. Such a system may have a 64bit CPU with DRAM mapped both below and above the 32bit address space, yet the PCIe devices can not perform DMA directly to/from the DRAM range above the 32bit limit. Hence, for such setup to work, all the buffers must exist below the 32bit limit.

Signed-off-by: Marek Vasut <marek.vasut+renesas at gmail.com>
Cc: Christoph Hellwig <hch at lst.de>
Cc: Geert Uytterhoeven <geert+renesas at glider.be>
Cc: Jens Axboe <axboe at fb.com>
Cc: Jens Axboe <axboe at kernel.dk>
Cc: Keith Busch <keith.busch at intel.com>
Cc: Robin Murphy <robin.murphy at arm.com>
Cc: Sagi Grimberg <sagi at grimberg.me>
Cc: Wolfram Sang <wsa+renesas at sang-engineering.com>
Cc: linux-renesas-soc at vger.kernel.org
To: linux-ide at vger.kernel.org
To: linux-nvme at lists.infradead.org

---
 drivers/ata/ahci.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/ata/ahci.c +7 -0
index 021ce46e2e57..2acce056dd8c 100644
@@ -926,6 +926,13 @@ static int ahci_configure_dma_masks(struct pci_dev *pdev, int using_dac)
 	if (pdev->dma_mask && pdev->dma_mask < DMA_BIT_MASK(32))
 		return 0;
 
+	/*
+	 * The upstream device could have applied DMA constraints already,
+	 * respect those and do not change the DMA masks.
+	 */
+	if (pdev->dev.dma_mask && pdev->dev.coherent_dma_mask)
+		return 0;
+
 	if (using_dac &&
 	    !dma_set_mask(&pdev->dev, DMA_BIT_MASK(64))) {
 		rc = dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(64));
Signature · 1 line
2.20.1
#1 lore raw

RFC PATCH 2/2 nvme-pci: Respect bus DMA constraints

From: Marek Vasut <[email protected]>

Since commit 6c2fb2ea7636 ("of/device: Set bus DMA mask as appropriate"), the upstream bus can constraint device DMA range. Respect that constraint and do not change the device DMA masks if they were already set.

This is applicable e.g. on systems where the PCIe controller cannot expose the full address space range. Such a system may have a 64bit CPU with DRAM mapped both below and above the 32bit address space, yet the PCIe devices can not perform DMA directly to/from the DRAM range above the 32bit limit. Hence, for such setup to work, all the buffers must exist below the 32bit limit.

Signed-off-by: Marek Vasut <marek.vasut+renesas at gmail.com>
Cc: Christoph Hellwig <hch at lst.de>
Cc: Geert Uytterhoeven <geert+renesas at glider.be>
Cc: Jens Axboe <axboe at fb.com>
Cc: Jens Axboe <axboe at kernel.dk>
Cc: Keith Busch <keith.busch at intel.com>
Cc: Robin Murphy <robin.murphy at arm.com>
Cc: Sagi Grimberg <sagi at grimberg.me>
Cc: Wolfram Sang <wsa+renesas at sang-engineering.com>
Cc: linux-renesas-soc at vger.kernel.org
To: linux-ide at vger.kernel.org
To: linux-nvme at lists.infradead.org

---
 drivers/nvme/host/pci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/nvme/host/pci.c +2 -1
index 92bad1c810ac..8ec873fb86b1 100644
@@ -2323,7 +2323,8 @@ static int nvme_pci_enable(struct nvme_dev *dev)
 
 	pci_set_master(pdev);
 
-	if (dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(64)) &&
+	if (!dev->dev->coherent_dma_mask &&
+	    dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(64)) &&
 	    dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32)))
 		goto disable;
 
Signature · 1 line
2.20.1
#2 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

In reply to [email protected]
Quoted 34 lines · On 2019-03-07 12:04 am, [email protected] wrote:

From: Marek Vasut <marek.vasut+renesas at gmail.com>

Since commit 6c2fb2ea7636 ("of/device: Set bus DMA mask as appropriate"), the upstream bus can constraint device DMA range. Respect that constraint and do not change the device DMA masks if they were already set.

This is applicable e.g. on systems where the PCIe controller cannot expose the full address space range. Such a system may have a 64bit CPU with DRAM mapped both below and above the 32bit address space, yet the PCIe devices can not perform DMA directly to/from the DRAM range above the 32bit limit. Hence, for such setup to work, all the buffers must exist below the 32bit limit.

Signed-off-by: Marek Vasut <marek.vasut+renesas at gmail.com>
Cc: Christoph Hellwig <hch at lst.de>
Cc: Geert Uytterhoeven <geert+renesas at glider.be>
Cc: Jens Axboe <axboe at fb.com>
Cc: Jens Axboe <axboe at kernel.dk>
Cc: Keith Busch <keith.busch at intel.com>
Cc: Robin Murphy <robin.murphy at arm.com>
Cc: Sagi Grimberg <sagi at grimberg.me>
Cc: Wolfram Sang <wsa+renesas at sang-engineering.com>
Cc: linux-renesas-soc at vger.kernel.org
To: linux-ide at vger.kernel.org
To: linux-nvme at lists.infradead.org

---
  drivers/ata/ahci.c | 7 +++++++
  1 file changed, 7 insertions(+)
drivers/ata/ahci.c +0 -0
index 021ce46e2e57..2acce056dd8c 100644

drivers/ata/ahci.c static int ahci_configure_dma_masks(struct pci_dev *pdev, int using_dac) · line 934

Quoted 9 lines
drivers/ata/ahci.c +6 -0
@@ -926,6 +926,13 @@ static int ahci_configure_dma_masks(struct pci_dev *pdev, int using_dac)
  	if (pdev->dma_mask && pdev->dma_mask < DMA_BIT_MASK(32))
  		return 0;
 
+	/*
+	 * The upstream device could have applied DMA constraints already,
+	 * respect those and do not change the DMA masks.
+	 */
+	if (pdev->dev.dma_mask && pdev->dev.coherent_dma_mask)
+		return 0;

At least for DT platforms, the device masks are always going to be set to some initial value, which will most commonly just be the 32-bit default - that should not prevent the driver from setting wider masks if that's what the device really supports (in fact there are some patches queued in which we're now starting to formalise that properly).

Are you seeing a problem with a DMA API backend failing to respect bus_dma_mask?

Robin.

Quoted 5 lines
drivers/ata/ahci.c +1 -0
+
  	if (using_dac &&
  	    !dma_set_mask(&pdev->dev, DMA_BIT_MASK(64))) {
  		rc = dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(64));
#3 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 54 lines · Robin Murphy
Quoted 40 lines · On 2019-03-07 12:04 am, [email protected] wrote:

From: Marek Vasut <marek.vasut+renesas at gmail.com>

Since commit 6c2fb2ea7636 ("of/device: Set bus DMA mask as appropriate"), the upstream bus can constraint device DMA range. Respect that constraint and do not change the device DMA masks if they were already set.

This is applicable e.g. on systems where the PCIe controller cannot expose
the full address space range. Such a system may have a 64bit CPU with DRAM
mapped both below and above the 32bit address space, yet the PCIe devices can not perform DMA directly to/from the DRAM range above the 32bit limit.
Hence, for such setup to work, all the buffers must exist below the 32bit limit.

Signed-off-by: Marek Vasut <marek.vasut+renesas at gmail.com>
Cc: Christoph Hellwig <hch at lst.de>
Cc: Geert Uytterhoeven <geert+renesas at glider.be>
Cc: Jens Axboe <axboe at fb.com>
Cc: Jens Axboe <axboe at kernel.dk>
Cc: Keith Busch <keith.busch at intel.com>
Cc: Robin Murphy <robin.murphy at arm.com>
Cc: Sagi Grimberg <sagi at grimberg.me>
Cc: Wolfram Sang <wsa+renesas at sang-engineering.com>
Cc: linux-renesas-soc at vger.kernel.org
To: linux-ide at vger.kernel.org
To: linux-nvme at lists.infradead.org
---
? drivers/ata/ahci.c | 7 +++++++
? 1 file changed, 7 insertions(+)

drivers/ata/ahci.c +0 -0
index 021ce46e2e57..2acce056dd8c 100644
@@ -926,6 +926,13 @@ static int ahci_configure_dma_masks(struct
 pci_dev *pdev, int using_dac)
 ????? if (pdev->dma_mask && pdev->dma_mask < DMA_BIT_MASK(32))
 ????????? return 0;
Quoted 6 lines

? +??? /*
+???? * The upstream device could have applied DMA constraints already, +???? * respect those and do not change the DMA masks.
+???? */
+??? if (pdev->dev.dma_mask && pdev->dev.coherent_dma_mask) +??????? return 0;

At least for DT platforms, the device masks are always going to be set to some initial value, which will most commonly just be the 32-bit default - that should not prevent the driver from setting wider masks if that's what the device really supports (in fact there are some patches queued in which we're now starting to formalise that properly).

Quoted 2 lines

Are you seeing a problem with a DMA API backend failing to respect bus_dma_mask?

Yes, the DMA mask gets overridden here to 64bit one, which on the R-Car Gen3 with PCI with 32bit addressing limitation makes the AHCI driver fail (and NVMe driver, and xHCI PCI etc). All those PCI devices fail the same way because they override the DMA mask.

Signature · 2 lines
Best regards,
Marek Vasut
#4 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 55 lines · Marek Vasut
Quoted 54 lines · Robin Murphy
Quoted 37 lines · On 2019-03-07 12:04 am, [email protected] wrote:

From: Marek Vasut <marek.vasut+renesas at gmail.com>

Since commit 6c2fb2ea7636 ("of/device: Set bus DMA mask as appropriate"), the upstream bus can constraint device DMA range. Respect that constraint and do not change the device DMA masks if they were already set.

This is applicable e.g. on systems where the PCIe controller cannot expose
the full address space range. Such a system may have a 64bit CPU with DRAM
mapped both below and above the 32bit address space, yet the PCIe devices can not perform DMA directly to/from the DRAM range above the 32bit limit.
Hence, for such setup to work, all the buffers must exist below the 32bit limit.

Signed-off-by: Marek Vasut <marek.vasut+renesas at gmail.com>
Cc: Christoph Hellwig <hch at lst.de>
Cc: Geert Uytterhoeven <geert+renesas at glider.be>
Cc: Jens Axboe <axboe at fb.com>
Cc: Jens Axboe <axboe at kernel.dk>
Cc: Keith Busch <keith.busch at intel.com>
Cc: Robin Murphy <robin.murphy at arm.com>
Cc: Sagi Grimberg <sagi at grimberg.me>
Cc: Wolfram Sang <wsa+renesas at sang-engineering.com>
Cc: linux-renesas-soc at vger.kernel.org
To: linux-ide at vger.kernel.org
To: linux-nvme at lists.infradead.org

---
 ? drivers/ata/ahci.c | 7 +++++++

? 1 file changed, 7 insertions(+)

drivers/ata/ahci.c +0 -0
index 021ce46e2e57..2acce056dd8c 100644

drivers/ata/ahci.c static int ahci_configure_dma_masks(struct · line 934

Quoted 9 lines
drivers/ata/ahci.c +5 -0
@@ -926,6 +926,13 @@ static int ahci_configure_dma_masks(struct
 pci_dev *pdev, int using_dac)
 ????? if (pdev->dma_mask && pdev->dma_mask < DMA_BIT_MASK(32))
 ????????? return 0;
 ? +??? /*
+???? * The upstream device could have applied DMA constraints already,
+???? * respect those and do not change the DMA masks.
+???? */
+??? if (pdev->dev.dma_mask && pdev->dev.coherent_dma_mask)
+??????? return 0;

At least for DT platforms, the device masks are always going to be set to some initial value, which will most commonly just be the 32-bit default - that should not prevent the driver from setting wider masks if that's what the device really supports (in fact there are some patches queued in which we're now starting to formalise that properly).

Quoted 7 lines
Quoted 2 lines

Are you seeing a problem with a DMA API backend failing to respect bus_dma_mask?

Yes, the DMA mask gets overridden here to 64bit one, which on the R-Car Gen3 with PCI with 32bit addressing limitation makes the AHCI driver fail (and NVMe driver, and xHCI PCI etc). All those PCI devices fail the same way because they override the DMA mask.

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Robin.

#5 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 63 lines · Robin Murphy
Quoted 62 lines · Marek Vasut
Quoted 60 lines · Robin Murphy
Quoted 44 lines · On 2019-03-07 12:04 am, [email protected] wrote:

From: Marek Vasut <marek.vasut+renesas at gmail.com>

Since commit 6c2fb2ea7636 ("of/device: Set bus DMA mask as appropriate"),
the upstream bus can constraint device DMA range. Respect that constraint
and do not change the device DMA masks if they were already set.

This is applicable e.g. on systems where the PCIe controller cannot expose
the full address space range. Such a system may have a 64bit CPU with DRAM
mapped both below and above the 32bit address space, yet the PCIe devices
can not perform DMA directly to/from the DRAM range above the 32bit limit.
Hence, for such setup to work, all the buffers must exist below the 32bit
limit.

Signed-off-by: Marek Vasut <marek.vasut+renesas at gmail.com>
Cc: Christoph Hellwig <hch at lst.de>
Cc: Geert Uytterhoeven <geert+renesas at glider.be>
Cc: Jens Axboe <axboe at fb.com>
Cc: Jens Axboe <axboe at kernel.dk>
Cc: Keith Busch <keith.busch at intel.com>
Cc: Robin Murphy <robin.murphy at arm.com>
Cc: Sagi Grimberg <sagi at grimberg.me>
Cc: Wolfram Sang <wsa+renesas at sang-engineering.com>
Cc: linux-renesas-soc at vger.kernel.org
To: linux-ide at vger.kernel.org
To: linux-nvme at lists.infradead.org
---
?? drivers/ata/ahci.c | 7 +++++++
?? 1 file changed, 7 insertions(+)

drivers/ata/ahci.c +0 -0
index 021ce46e2e57..2acce056dd8c 100644
@@ -926,6 +926,13 @@ static int ahci_configure_dma_masks(struct
 pci_dev *pdev, int using_dac)
 ?????? if (pdev->dma_mask && pdev->dma_mask < DMA_BIT_MASK(32))
 ?????????? return 0;
Quoted 6 lines

?? +??? /*
+???? * The upstream device could have applied DMA constraints already, +???? * respect those and do not change the DMA masks.
+???? */
+??? if (pdev->dev.dma_mask && pdev->dev.coherent_dma_mask) +??????? return 0;

At least for DT platforms, the device masks are always going to be set to some initial value, which will most commonly just be the 32-bit default - that should not prevent the driver from setting wider masks if that's what the device really supports (in fact there are some patches queued in which we're now starting to formalise that properly).

Are you seeing a problem with a DMA API backend failing to respect bus_dma_mask?

Quoted 8 lines
Quoted 4 lines

Yes, the DMA mask gets overridden here to 64bit one, which on the R-Car Gen3 with PCI with 32bit addressing limitation makes the AHCI driver fail (and NVMe driver, and xHCI PCI etc). All those PCI devices fail the same way because they override the DMA mask.

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

Signature · 2 lines
Best regards,
Marek Vasut
#6 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 5 lines · Marek Vasut
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

#7 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 9 lines · Christoph Hellwig
Quoted 5 lines · Marek Vasut
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

Signature · 2 lines
Best regards,
Marek Vasut
#8 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 6 lines · Marek Vasut
Quoted 5 lines · Christoph Hellwig
Quoted 4 lines · Marek Vasut
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case. What are the exact symptoms you see? Does it involve swiotlb not kicking in, or iommu issues? What is the exact kernel version?

#9 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 12 lines · Christoph Hellwig
Quoted 11 lines · Marek Vasut
Quoted 9 lines · Christoph Hellwig
Quoted 5 lines · Marek Vasut
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 6 lines
Quoted 3 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case. What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

Quoted 1 line

What is the exact kernel version?

next/master from 20190306 (5.0.0 + next patches)

Signature · 2 lines
Best regards,
Marek Vasut
#10 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 29 lines · Marek Vasut
Quoted 12 lines · Christoph Hellwig
Quoted 11 lines · Marek Vasut
Quoted 9 lines · Christoph Hellwig
Quoted 5 lines · Marek Vasut
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 6 lines
Quoted 3 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case. What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Quoted 8 lines

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

Digging further ...

drivers/nvme/host/pci.c nvme_map_data() calls dma_map_sg_attrs() and the resulting sglist contains entry with >32bit PA. This is because dma_map_sg_attrs() calls dma_direct_map_sg(), which in turn calls dma_direct_map_sg(), then dma_direct_map_page() and that's where it goes weird.

dma_direct_map_page() does a dma_direct_possible() check before triggering swiotlb_map(). The check succeeds, so the later isn't executed.

dma_direct_possible() calls dma_capable() with dev->dma_mask = DMA_BIT_MASK(64) and dev->dma_bus_mask = 0, so min_not_zero(*dev->dma_mask, dev->bus_dma_mask) returns DMA_BIT_MASK(64).

Surely enough, if I hack dma_direct_possible() to return 0, swiotlb_map() kicks in and the nvme driver starts working fine.

I presume the question here is, why is dev->bus_dma_mask = 0 ?

Signature · 2 lines
Best regards,
Marek Vasut
#11 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Hi Marek,

Quoted 50 lines · Marek Vasut
Quoted 29 lines · Marek Vasut
Quoted 12 lines · Christoph Hellwig
Quoted 11 lines · Marek Vasut
Quoted 9 lines · Christoph Hellwig
Quoted 5 lines · Marek Vasut
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 6 lines
Quoted 3 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case. What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Quoted 8 lines

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

Digging further ...

drivers/nvme/host/pci.c nvme_map_data() calls dma_map_sg_attrs() and the resulting sglist contains entry with >32bit PA. This is because dma_map_sg_attrs() calls dma_direct_map_sg(), which in turn calls dma_direct_map_sg(), then dma_direct_map_page() and that's where it goes weird.

dma_direct_map_page() does a dma_direct_possible() check before triggering swiotlb_map(). The check succeeds, so the later isn't executed.

Quoted 8 lines

dma_direct_possible() calls dma_capable() with dev->dma_mask = DMA_BIT_MASK(64) and dev->dma_bus_mask = 0, so min_not_zero(*dev->dma_mask, dev->bus_dma_mask) returns DMA_BIT_MASK(64).

Surely enough, if I hack dma_direct_possible() to return 0, swiotlb_map() kicks in and the nvme driver starts working fine.

I presume the question here is, why is dev->bus_dma_mask = 0 ?

Because that's the default, and almost no code overrides that?

$ git grep "\<bus_dma_mask ="
arch/mips/pci/fixup-sb1250.c:           dev->dev.bus_dma_mask =
DMA_BIT_MASK(32);
arch/x86/kernel/pci-dma.c:      pdev->dev.bus_dma_mask = DMA_BIT_MASK(32);
drivers/acpi/arm64/iort.c:              dev->bus_dma_mask = mask;
drivers/of/device.c:            dev->bus_dma_mask = mask;

dev is the nvme PCI device, I assume? So you can ignore the last match.

The first two seem to be related to platforms that cannot do >32 bit DMA on PCI. So that's a hint on how to fix this...

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#12 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 1 line · Geert Uytterhoeven

Hi Marek,

Hi,

Quoted 55 lines · Marek Vasut
Quoted 54 lines · Marek Vasut
Quoted 29 lines · Christoph Hellwig
Quoted 12 lines · Marek Vasut
Quoted 11 lines · Christoph Hellwig
Quoted 9 lines · Marek Vasut
Quoted 5 lines
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 6 lines
Quoted 3 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case. What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Quoted 8 lines

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

Digging further ...

drivers/nvme/host/pci.c nvme_map_data() calls dma_map_sg_attrs() and the resulting sglist contains entry with >32bit PA. This is because dma_map_sg_attrs() calls dma_direct_map_sg(), which in turn calls dma_direct_map_sg(), then dma_direct_map_page() and that's where it goes weird.

dma_direct_map_page() does a dma_direct_possible() check before triggering swiotlb_map(). The check succeeds, so the later isn't executed.

dma_direct_possible() calls dma_capable() with dev->dma_mask = DMA_BIT_MASK(64) and dev->dma_bus_mask = 0, so min_not_zero(*dev->dma_mask, dev->bus_dma_mask) returns DMA_BIT_MASK(64).

Quoted 6 lines
Quoted 4 lines

Surely enough, if I hack dma_direct_possible() to return 0, swiotlb_map() kicks in and the nvme driver starts working fine.

I presume the question here is, why is dev->bus_dma_mask = 0 ?

Because that's the default, and almost no code overrides that?

But shouldn't drivers/of/device.c set that for the PCIe controller ?

Quoted 11 lines
$ git grep "\<bus_dma_mask ="
arch/mips/pci/fixup-sb1250.c:           dev->dev.bus_dma_mask =
DMA_BIT_MASK(32);
arch/x86/kernel/pci-dma.c:      pdev->dev.bus_dma_mask = DMA_BIT_MASK(32);
drivers/acpi/arm64/iort.c:              dev->bus_dma_mask = mask;
drivers/of/device.c:            dev->bus_dma_mask = mask;

dev is the nvme PCI device, I assume? So you can ignore the last match.

The first two seem to be related to platforms that cannot do >32 bit DMA on PCI. So that's a hint on how to fix this...

That doesn't feel right, it's not a platform limitation, but a PCIe IP limitation, so this fix should live somewhere in drivers/ I think ?

Signature · 2 lines
Best regards,
Marek Vasut
#13 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 60 lines · Marek Vasut
Quoted 1 line · Geert Uytterhoeven

Hi Marek,

Hi,

Quoted 55 lines · Marek Vasut
Quoted 54 lines · Marek Vasut
Quoted 29 lines · Christoph Hellwig
Quoted 12 lines · Marek Vasut
Quoted 11 lines · Christoph Hellwig
Quoted 9 lines · Marek Vasut
Quoted 5 lines
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 6 lines
Quoted 3 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case. What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Quoted 8 lines

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

Digging further ...

drivers/nvme/host/pci.c nvme_map_data() calls dma_map_sg_attrs() and the resulting sglist contains entry with >32bit PA. This is because dma_map_sg_attrs() calls dma_direct_map_sg(), which in turn calls dma_direct_map_sg(), then dma_direct_map_page() and that's where it goes weird.

dma_direct_map_page() does a dma_direct_possible() check before triggering swiotlb_map(). The check succeeds, so the later isn't executed.

dma_direct_possible() calls dma_capable() with dev->dma_mask = DMA_BIT_MASK(64) and dev->dma_bus_mask = 0, so min_not_zero(*dev->dma_mask, dev->bus_dma_mask) returns DMA_BIT_MASK(64).

Quoted 8 lines
Quoted 6 lines
Quoted 4 lines

Surely enough, if I hack dma_direct_possible() to return 0, swiotlb_map() kicks in and the nvme driver starts working fine.

I presume the question here is, why is dev->bus_dma_mask = 0 ?

Because that's the default, and almost no code overrides that?

But shouldn't drivers/of/device.c set that for the PCIe controller ?

Urgh, I really should have spotted the significance of "NVMe", but somehow it failed to click :(

Of course the existing code works fine for everything *except* PCI devices on DT-based systems... That's because of_dma_get_range() has never been made to work correctly with the trick we play of passing the host bridge of_node through of_dma_configure(). I've got at least 2 or 3 half-finished attempts at improving that, but they keep getting sidetracked into trying to clean up the various new of_dma_configure() hacks I find in drivers and/or falling down the rabbit-hole of starting to redesign the whole dma_pfn_offset machinery entirely. Let me dig one up and try to constrain it to solve just this most common "one single limited range" condition for the sake of making actual progress...

Robin.

Quoted 14 lines
Quoted 11 lines
$ git grep "\<bus_dma_mask ="
arch/mips/pci/fixup-sb1250.c:           dev->dev.bus_dma_mask =
DMA_BIT_MASK(32);
arch/x86/kernel/pci-dma.c:      pdev->dev.bus_dma_mask = DMA_BIT_MASK(32);
drivers/acpi/arm64/iort.c:              dev->bus_dma_mask = mask;
drivers/of/device.c:            dev->bus_dma_mask = mask;

dev is the nvme PCI device, I assume? So you can ignore the last match.

The first two seem to be related to platforms that cannot do >32 bit DMA on PCI. So that's a hint on how to fix this...

That doesn't feel right, it's not a platform limitation, but a PCIe IP limitation, so this fix should live somewhere in drivers/ I think ? >

#14 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 74 lines · Robin Murphy
Quoted 73 lines · Marek Vasut
Quoted 1 line · Geert Uytterhoeven

Hi Marek,

Hi,

Quoted 68 lines
Quoted 66 lines · Marek Vasut
Quoted 34 lines · Marek Vasut
Quoted 15 lines · Christoph Hellwig
Quoted 14 lines · Marek Vasut
Quoted 12 lines · Christoph Hellwig
Quoted 8 lines · Marek Vasut
Quoted 6 lines

Right, but whoever *interprets* the device masks after the driver has
overridden them should be taking the (smaller) bus mask into account as
well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 7 lines
Quoted 4 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit
address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case.? What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in
the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Quoted 8 lines

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

Digging further ...

drivers/nvme/host/pci.c nvme_map_data() calls dma_map_sg_attrs() and the
resulting sglist contains entry with >32bit PA. This is because dma_map_sg_attrs() calls dma_direct_map_sg(), which in turn calls dma_direct_map_sg(), then dma_direct_map_page() and that's where it goes
weird.

dma_direct_map_page() does a dma_direct_possible() check before triggering swiotlb_map(). The check succeeds, so the later isn't executed.

dma_direct_possible() calls dma_capable() with dev->dma_mask = DMA_BIT_MASK(64) and dev->dma_bus_mask = 0, so min_not_zero(*dev->dma_mask, dev->bus_dma_mask) returns DMA_BIT_MASK(64).

Surely enough, if I hack dma_direct_possible() to return 0, swiotlb_map() kicks in and the nvme driver starts working fine.

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Quoted 1 line

I presume the question here is, why is dev->bus_dma_mask = 0 ?

Because that's the default, and almost no code overrides that?

But shouldn't drivers/of/device.c set that for the PCIe controller ?

Urgh, I really should have spotted the significance of "NVMe", but somehow it failed to click :(

Good thing it did now :-)

Quoted 10 lines

Of course the existing code works fine for everything *except* PCI devices on DT-based systems... That's because of_dma_get_range() has never been made to work correctly with the trick we play of passing the host bridge of_node through of_dma_configure(). I've got at least 2 or 3 half-finished attempts at improving that, but they keep getting sidetracked into trying to clean up the various new of_dma_configure() hacks I find in drivers and/or falling down the rabbit-hole of starting to redesign the whole dma_pfn_offset machinery entirely. Let me dig one up and try to constrain it to solve just this most common "one single limited range" condition for the sake of making actual progress...

That'd be nice, thank you. I'm happy to test it on various devices here.

Signature · 2 lines
Best regards,
Marek Vasut
#15 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 97 lines · Marek Vasut
Quoted 74 lines · Robin Murphy
Quoted 73 lines · Marek Vasut
Quoted 1 line · Geert Uytterhoeven

Hi Marek,

Hi,

Quoted 68 lines
Quoted 66 lines · Marek Vasut
Quoted 34 lines · Marek Vasut
Quoted 15 lines · Christoph Hellwig
Quoted 14 lines · Marek Vasut
Quoted 12 lines · Christoph Hellwig
Quoted 8 lines · Marek Vasut
Quoted 6 lines

Right, but whoever *interprets* the device masks after the driver has
overridden them should be taking the (smaller) bus mask into account as
well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 7 lines
Quoted 4 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit
address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case.? What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in
the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Quoted 8 lines

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

Digging further ...

drivers/nvme/host/pci.c nvme_map_data() calls dma_map_sg_attrs() and the
resulting sglist contains entry with >32bit PA. This is because dma_map_sg_attrs() calls dma_direct_map_sg(), which in turn calls dma_direct_map_sg(), then dma_direct_map_page() and that's where it goes
weird.

dma_direct_map_page() does a dma_direct_possible() check before triggering swiotlb_map(). The check succeeds, so the later isn't executed.

dma_direct_possible() calls dma_capable() with dev->dma_mask = DMA_BIT_MASK(64) and dev->dma_bus_mask = 0, so min_not_zero(*dev->dma_mask, dev->bus_dma_mask) returns DMA_BIT_MASK(64).

Surely enough, if I hack dma_direct_possible() to return 0, swiotlb_map() kicks in and the nvme driver starts working fine.

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Quoted 1 line

I presume the question here is, why is dev->bus_dma_mask = 0 ?

Because that's the default, and almost no code overrides that?

But shouldn't drivers/of/device.c set that for the PCIe controller ?

Urgh, I really should have spotted the significance of "NVMe", but somehow it failed to click :(

Good thing it did now :-)

Quoted 10 lines

Of course the existing code works fine for everything *except* PCI devices on DT-based systems... That's because of_dma_get_range() has never been made to work correctly with the trick we play of passing the host bridge of_node through of_dma_configure(). I've got at least 2 or 3 half-finished attempts at improving that, but they keep getting sidetracked into trying to clean up the various new of_dma_configure() hacks I find in drivers and/or falling down the rabbit-hole of starting to redesign the whole dma_pfn_offset machinery entirely. Let me dig one up and try to constrain it to solve just this most common "one single limited range" condition for the sake of making actual progress...

Quoted 1 line

That'd be nice, thank you. I'm happy to test it on various devices here.

Just curious, no stress, did you get anywhere with this patch(set) yet?

Signature · 2 lines
Best regards,
Marek Vasut
#16 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

Quoted 98 lines · Marek Vasut
Quoted 97 lines · Marek Vasut
Quoted 74 lines · Robin Murphy
Quoted 73 lines · Marek Vasut
Quoted 1 line · Geert Uytterhoeven

Hi Marek,

Hi,

Quoted 68 lines
Quoted 66 lines · Marek Vasut
Quoted 34 lines · Marek Vasut
Quoted 15 lines · Christoph Hellwig
Quoted 14 lines · Marek Vasut
Quoted 12 lines · Christoph Hellwig
Quoted 8 lines · Marek Vasut
Quoted 6 lines

Right, but whoever *interprets* the device masks after the driver has
overridden them should be taking the (smaller) bus mask into account as
well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 7 lines
Quoted 4 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit
address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case.? What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in
the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Quoted 8 lines

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

Digging further ...

drivers/nvme/host/pci.c nvme_map_data() calls dma_map_sg_attrs() and the
resulting sglist contains entry with >32bit PA. This is because dma_map_sg_attrs() calls dma_direct_map_sg(), which in turn calls dma_direct_map_sg(), then dma_direct_map_page() and that's where it goes
weird.

dma_direct_map_page() does a dma_direct_possible() check before triggering swiotlb_map(). The check succeeds, so the later isn't executed.

dma_direct_possible() calls dma_capable() with dev->dma_mask = DMA_BIT_MASK(64) and dev->dma_bus_mask = 0, so min_not_zero(*dev->dma_mask, dev->bus_dma_mask) returns DMA_BIT_MASK(64).

Surely enough, if I hack dma_direct_possible() to return 0, swiotlb_map() kicks in and the nvme driver starts working fine.

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Quoted 1 line

I presume the question here is, why is dev->bus_dma_mask = 0 ?

Because that's the default, and almost no code overrides that?

But shouldn't drivers/of/device.c set that for the PCIe controller ?

Urgh, I really should have spotted the significance of "NVMe", but somehow it failed to click :(

Good thing it did now :-)

Quoted 10 lines

Of course the existing code works fine for everything *except* PCI devices on DT-based systems... That's because of_dma_get_range() has never been made to work correctly with the trick we play of passing the host bridge of_node through of_dma_configure(). I've got at least 2 or 3 half-finished attempts at improving that, but they keep getting sidetracked into trying to clean up the various new of_dma_configure() hacks I find in drivers and/or falling down the rabbit-hole of starting to redesign the whole dma_pfn_offset machinery entirely. Let me dig one up and try to constrain it to solve just this most common "one single limited range" condition for the sake of making actual progress...

Quoted 3 lines
Quoted 1 line

That'd be nice, thank you. I'm happy to test it on various devices here.

Just curious, no stress, did you get anywhere with this patch(set) yet?

Bump ?

Signature · 2 lines
Best regards,
Marek Vasut
#17 lore raw

RFC PATCH 1/2 ata: ahci: Respect bus DMA constraints

In reply to Marek Vasut

Hi Marek,

Quoted 29 lines · Marek Vasut
Quoted 12 lines · Christoph Hellwig
Quoted 11 lines · Marek Vasut
Quoted 9 lines · Christoph Hellwig
Quoted 5 lines · Marek Vasut
Quoted 3 lines

Right, but whoever *interprets* the device masks after the driver has overridden them should be taking the (smaller) bus mask into account as well, so the question is where is *that* not being done correctly?

Do you have a hint where I should look for that ?

If this a 32-bit ARM platform it might the complete lack of support for bus_dma_mask in arch/arm/mm/dma-mapping.c..

Quoted 6 lines
Quoted 3 lines

It's an ARM 64bit platform, just the PCIe controller is limited to 32bit address range, so the devices on the PCIe bus cannot read the host's DRAM above the 32bit limit.

arm64 should take the mask into account both for the swiotlb and iommu case. What are the exact symptoms you see?

With the nvme, the device is recognized, but cannot be used. It boils down to PCI BAR access being possible, since that's all below the 32bit boundary, but when the device tries to do any sort of DMA, that transfer returns nonsense data.

But when I call dma_set_mask_and_coherent(dev->dev, DMA_BIT_MASK(32) in the affected driver (thus far I tried this nvme, xhci-pci and ahci-pci drivers), it all starts to work fine.

Quoted 8 lines

Could it be that the driver overwrites the (coherent_)dma_mask and that's why the swiotlb/iommu code cannot take this into account ?

Quoted 2 lines

Does it involve
swiotlb not kicking in, or iommu issues?

How can I check ? I added printks into arch/arm64/mm/dma-mapping.c and drivers/iommu/dma-iommu.c , but I suspect I need to look elsewhere.

So far the IOMMU is disabled in upstream, as no devices are whitelisted in drivers/iommu/ipmmu-vmsa.c.

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds