Message-ID: <20240613224125.GA1087289@bhelgaas>
Date: Thu, 13 Jun 2024 17:41:25 -0500
From: Bjorn Helgaas <helgaas@...nel.org>
To: Frank Li <Frank.li@....com>
Cc: Robin Murphy <robin.murphy@....com>, Richard Zhu <hongxing.zhu@....com>,
	Lucas Stach <l.stach@...gutronix.de>,
	Lorenzo Pieralisi <lpieralisi@...nel.org>,
	Krzysztof Wilczyński <kw@...ux.com>,
	Rob Herring <robh@...nel.org>, Bjorn Helgaas <bhelgaas@...gle.com>,
	Shawn Guo <shawnguo@...nel.org>,
	Sascha Hauer <s.hauer@...gutronix.de>,
	Pengutronix Kernel Team <kernel@...gutronix.de>,
	Fabio Estevam <festevam@...il.com>,
	NXP Linux Team <linux-imx@....com>,
	Philipp Zabel <p.zabel@...gutronix.de>,
	Liam Girdwood <lgirdwood@...il.com>,
	Mark Brown <broonie@...nel.org>,
	Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>,
	Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
	Conor Dooley <conor+dt@...nel.org>, linux-pci@...r.kernel.org,
	imx@...ts.linux.dev, linux-arm-kernel@...ts.infradead.org,
	linux-kernel@...r.kernel.org, bpf@...r.kernel.org,
	devicetree@...r.kernel.org, Will Deacon <will@...nel.org>,
	Joerg Roedel <joro@...tes.org>, Jason Gunthorpe <jgg@...pe.ca>,
	Alyssa Rosenzweig <alyssa@...enzweig.io>,
	Marc Zyngier <maz@...nel.org>
Subject: Re: [PATCH v5 08/12] PCI: imx6: Config look up table(LUT) to support
 MSI ITS and IOMMU for i.MX95

On Thu, Jun 06, 2024 at 04:24:17PM -0400, Frank Li wrote:
> On Mon, Jun 03, 2024 at 04:07:55PM -0400, Frank Li wrote:
> > On Mon, Jun 03, 2024 at 01:56:27PM -0500, Bjorn Helgaas wrote:
> > > On Mon, Jun 03, 2024 at 02:42:45PM -0400, Frank Li wrote:
> > > > On Mon, Jun 03, 2024 at 12:19:21PM -0500, Bjorn Helgaas wrote:
> > > > > On Fri, May 31, 2024 at 03:58:49PM +0100, Robin Murphy wrote:
> > > > > > On 2024-05-31 12:08 am, Bjorn Helgaas wrote:
> > > > > > > [+cc IOMMU and pcie-apple.c folks for comment]
> > > > > > > 
> > > > > > > On Tue, May 28, 2024 at 03:39:21PM -0400, Frank Li wrote:
> > > > > > > > For the i.MX95, configuration of a LUT is necessary to convert Bus Device
> > > > > > > > Function (BDF) to stream IDs, which are utilized by both IOMMU and ITS.
> > > > > > > > This involves examining the msi-map and smmu-map to ensure consistent
> > > > > > > > mapping of PCI BDF to the same stream IDs. Subsequently, LUT-related
> > > > > > > > registers are configured. In the absence of an msi-map, the built-in MSI
> > > > > > > > controller is utilized as a fallback.
> > > > > > > > 
> > > > > > > > Additionally, register a PCI bus notifier to trigger imx_pcie_add_device()
> > > > > > > > upon the appearance of a new PCI device and when the bus is an iMX6 PCI
> > > > > > > > controller. This function configures the correct LUT based on Device Tree
> > > > > > > > Settings (DTS).
> > > > > > > 
> > > > > > > This scheme is pretty similar to apple_pcie_bus_notifier().  If we
> > > > > > > have to do this, I wish it were *more* similar, i.e., copy the
> > > > > > > function names, bitmap tracking, code structure, etc.
> > > > > > > 
> > > > > > > I don't really know how stream IDs work, but I assume they are used on
> > > > > > > most or all arm64 platforms, so I'm a little surprised that of all the
> > > > > > > PCI host drivers used on arm64, only pcie-apple.c and pci-imx6.c need
> > > > > > > this notifier.
> > > > > > 
> > > > > > This is one of those things that's mostly at the mercy of the PCIe root
> > > > > > complex implementation. Typically the SMMU StreamID and/or GIC ITS DeviceID
> > > > > > is derived directly from the PCI RID, sometimes with additional high-order
> > > > > > bits hard-wired to disambiguate PCI segments. I believe this RID-translation
> > > > > > LUT is a particular feature of the the Synopsys IP - I know there's also one
> > > > > > on the NXP Layerscape platforms, but on those it's programmed by the
> > > > > > bootloader, which also generates the appropriate "msi-map" and "iommu-map"
> > > > > > properties to match. Ideally that's what i.MX should do as well, but hey.
> > > > > 
> > > > > Maybe this RID-translation is a feature of i.MX, not of Synopsys?  I
> > > > > see that the LUT CSR accesses use IMX95_* definitions.
> > > > 
> > > > Yes, it convert 16bit RID to 6bit stream id.
> > > 
> > > IIUC, you're saying this is not a Synopsys feature, it's an i.MX
> > > feature.
> > 
> > Yes, it is i.MX feature. But I think other vendor should have similar
> > situation if use old arm smmu.
> > 
> > > 
> > > > > > If it's really necessary to do this programming from Linux, then there's
> > > > > > still no point in it being dynamic - the mappings cannot ever change, since
> > > > > > the rest of the kernel believes that what the DT said at boot time was
> > > > > > already a property of the hardware. It would be a lot more logical, and
> > > > > > likely simpler, for the driver to just read the relevant map property and
> > > > > > program the entire LUT to match, all in one go at controller probe time.
> > > > > > Rather like what's already commonly done with the parsing of "dma-ranges" to
> > > > > > program address-translation LUTs for inbound windows.
> > > > > > 
> > > > > > Plus that would also give a chance of safely dealing with bad DTs specifying
> > > > > > invalid ID mappings (by refusing to probe at all). As it is, returning an
> > > > > > error from a child's BUS_NOTIFY_ADD_DEVICE does nothing except prevent any
> > > > > > further notifiers from running at that point - the device will still be
> > > > > > added, allowed to bind a driver, and able to start sending DMA/MSI traffic
> > > > > > without the controller being correctly programmed, which at best won't work
> > > > > > and at worst may break the whole system.
> > > > > 
> > > > > Frank, could the imx LUT be programmed once at boot-time instead of at
> > > > > device-add time?  I'm guessing maybe not because apparently there is a
> > > > > risk of running out of LUT entries?
> > > > 
> > > > It is not good idea to depend on boot loader so much.
> > > 
> > > I meant "could this be programmed once when the Linux imx host
> > > controller driver is probed?"  But from the below, it sounds like
> > > that's not possible in general because you don't have enough stream
> > > IDs to do that.
> > 
> > Oh! sorry miss understand what your means. It is possible like what I did
> > at v3 version. But I think it is not good enough. 
> > 
> > > 
> > > > Some hot plug devics
> > > > (SD7.0) may plug after system boot. Two PCIe instances shared one set
> > > > of 6bits stream id (total 64). Assume total 16 assign to two PCIe
> > > > controllers. each have 8 stream id. If use uboot assign it static, each
> > > > PCIe controller have below 8 devices.  It will be failrue one controller
> > > > connect 7, another connect 9. but if dynamtic alloc when devices add, both
> > > > controller can work.
> > > > 
> > > > Although we have not so much devices now,  this way give us possility to
> > > > improve it in future.
> > > > 
> > > > > It sounds like the consequences of running out of LUT entries are
> > > > > catastrophic, e.g., memory corruption from mis-directed DMA?  If
> > > > > that's possible, I think we need to figure out how to prevent the
> > > > > device from being used, not just dev_warn() about it.
> > > > 
> > > > Yes, but so far, we have not met such problem now. We can improve it when
> > > > we really face such problem.
> > > 
> > > If this controller can only support DMA from a limited number of
> > > endpoints below it, I think we should figure out how to enforce that
> > > directly.  Maybe we can prevent drivers from enabling bus mastering or
> > > something.  I'm not happy with the idea of waiting for and debugging a
> > > report of data corruption.
> > 
> > It may add a pre-add hook function to pci bridge. let me do more research.
> 
> Hi Bjorn:
> 
> int pci_setup_device(struct pci_dev *dev)
> {
> 	dev->error_state = pci_channel_io_normal;
> 	...
> 	pci_fixup_device(pci_fixup_early, dev);
> 
> 	^^^ I can add fixup hook for pci_fixup_early. If not resource, 
> I can set dev->error_state to pci_channel_io_frozen or
> pci_channel_io_perm_failure
> 	
> 	And add below check here after call hook function.
> 
> 	if (dev->error_state != pci_channel_io_normal)
> 		return -EIO;
> 		
> }
> 
> How do you think this method? If you agree, I can continue search device
> remove hook up.

I think this would mean the device would not appear to be enumerated
at all, right?  I.e., it wouldn't show up in lspci?  And we couldn't
use even a pure programmed IO driver with no DMA or MSI?

I wonder if we should have a function pointer in struct
pci_host_bridge, kind of like the existing ->map_irq(), where we could
do host bridge-specific setup when enumerating a PCI device.

We'd still have to solve the issue of preventing DMA, but a hook like
that might avoid the need for a quirk or the bus notifier approach.

Bjorn

Powered by blists - more mailing lists