Re: [RFC][PATCH] Reset PCIe devices to address DMA problem on kdump with iommu

From: Don Dutile <ddutile@redhat.com>
To: Takao Indoh <indou.takao@jp.fujitsu.com>
Cc: kexec@lists.infradead.org, linux-kernel@vger.kernel.org,
	linux-pci@vger.kernel.org, bhelgaas@google.com,
	vgoyal@redhat.com, hbabu@us.ibm.com,
	ishii.hironobu@jp.fujitsu.com, martin.wilck@ts.fujitsu.com
Subject: Re: [RFC][PATCH] Reset PCIe devices to address DMA problem on kdump with iommu
Date: Mon, 06 Aug 2012 00:09:43 -0400	[thread overview]
Message-ID: <501F4387.3010404@redhat.com> (raw)
In-Reply-To: <501BB4EF.7080909@jp.fujitsu.com>

On 08/03/2012 07:24 AM, Takao Indoh wrote:
> Hi all,
> 
> This patch adds kernel parameter "reset_pcie_devices" which resets PCIe
> devices at boot time to address DMA problem on kdump with iommu. When
> this parameter is specified, a hot reset is triggered on each PCIe root
> port and downstream port to reset its downstream endpoint.
> 
> Background:
> A kdump problem about DMA has been discussed for a long time. That is,
> when a kernel is switched to the kdump kernel DMA derived from first
> kernel affects second kernel. Recently this problem surfaces when iommu
> is used for PCI passthrough on KVM guest. In the case of the machine I
> use, when intel_iommu=on is specified, DMAR error is detected in kdump
> kernel and PCI SERR is also detected. Finally kdump fails because some
> devices does not work correctly.
> 
> The root cause is that ongoing DMA from first kernel causes DMAR fault
> because page table of DMAR is initialized while kdump kernel is booting
> up. Therefore to address this problem DMA needs to be stopped before DMAR
> is initialized at kdump kernel boot time. By this patch, PCIe devices
> are reset by hot reset and its DMA is stopped when reset_pcie_devices is
> specified. One problem of this solution is that VGA is reset and the
> monitor blacks out when the link between the port and VGA controller was
> reset. So this patch does not reset the port whose child endpoint is VGA
> device.
> 
> Any comments would be appreciated.
> 
> Signed-off-by: Takao Indoh<indou.takao@jp.fujitsu.com>
> ---
Have you considered something less disruptive such as clearing the 
Master Enable in each endpoint's PCI Command Register ?
That should prevent DMA transactions from endpoints during the kdump and
kexec, and when the driver for the device gets reconfigured,
Master Enable will be set back on, but after the driver has had the
opportunity to do a device-specific reset.

- Don

>   Documentation/kernel-parameters.txt |    4 +
>   drivers/pci/quirks.c                |   59 ++++++++++++++++++++++++++
>   2 files changed, 63 insertions(+)
> 
> diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
> index e714a02..e694e9f 100644
> --- a/Documentation/kernel-parameters.txt
> +++ b/Documentation/kernel-parameters.txt
> @@ -2489,6 +2489,10 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
>   	reset_devices	[KNL] Force drivers to reset the underlying device
>   			during initialization.
> 
> +	reset_pcie_devices
> +			[PCIE] Reset PCIe endpoint at boot time by sending a
> +			hot reset to root port and downstream port
> +
>   	resume=		[SWSUSP]
>   			Specify the partition device for software suspend
>   			Format:
> diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> index 5155317..7f7fc02 100644
> --- a/drivers/pci/quirks.c
> +++ b/drivers/pci/quirks.c
> @@ -32,6 +32,65 @@
>   #include "pci.h"
> 
>   /*
> + * Reset PCIe endpoint by sending a hot reset to root port and downstream port
> + */
> +unsigned int reset_pcie_devices;
> +EXPORT_SYMBOL(reset_pcie_devices);
> +static int __init set_reset_pcie_devices(char *str)
> +{
> +	reset_pcie_devices = 1;
> +	return 1;
> +}
> +__setup("reset_pcie_devices", set_reset_pcie_devices);
> +
> +static void __devinit quirk_pcie_device_reset(struct pci_dev *dev)
> +{
> +	struct pci_bus *subordinate;
> +	struct pci_dev *child;
> +	u16 ctrl;
> +
> +	if (!reset_pcie_devices || !pci_is_pcie(dev) || !dev->subordinate ||
> +	    ((dev->pcie_type != PCI_EXP_TYPE_ROOT_PORT)&&
> +	     (dev->pcie_type != PCI_EXP_TYPE_DOWNSTREAM)))
> +		return;
> +
> +	subordinate = dev->subordinate;
> +	list_for_each_entry(child,&subordinate->devices, bus_list) {
> +		if ((child->pcie_type == PCI_EXP_TYPE_UPSTREAM) ||
> +		    (child->pcie_type == PCI_EXP_TYPE_PCI_BRIDGE) ||
> +		    ((child->class>>  16) == PCI_BASE_CLASS_DISPLAY))
> +		/* Don't reset switch, bridge, VGA device */
> +		return;
> +	}
> +
> +	dev_info(&dev->dev, "Reset Secondary bus\n");
> +
> +	list_for_each_entry(child,&subordinate->devices, bus_list) {
> +		dev_info(&child->dev, "save state\n");
> +		pci_save_state(child);
> +	}
> +
> +	/* Assert Secondary Bus Reset */
> +	pci_read_config_word(dev, PCI_BRIDGE_CONTROL,&ctrl);
> +	ctrl |= PCI_BRIDGE_CTL_BUS_RESET;
> +	pci_write_config_word(dev, PCI_BRIDGE_CONTROL, ctrl);
> +
> +	msleep(2);
> +
> +	/* De-assert Secondary Bus Reset */
> +	ctrl&= ~PCI_BRIDGE_CTL_BUS_RESET;
> +	pci_write_config_word(dev, PCI_BRIDGE_CONTROL, ctrl);
> +
> +	msleep(200);
> +
> +	list_for_each_entry(child,&subordinate->devices, bus_list) {
> +		dev_info(&child->dev, "restore state\n");
> +		pci_restore_state(child);
> +	}
> +}
> +DECLARE_PCI_FIXUP_FINAL(PCI_ANY_ID, PCI_ANY_ID, quirk_pcie_device_reset);
> +
> +/*
>    * Decoding should be disabled for a PCI device during BAR sizing to avoid
>    * conflict. But doing so may cause problems on host bridge and perhaps other
>    * key system devices. For devices that need to have mmio decoding always-on,
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-pci" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html