PCI: Expose PCIe Resizable BAR support via sysfs

Add a simple sysfs interface to Resizable BAR support, largely for the
purposes of assigning such devices to a VM through VFIO.  Resizable BARs
present a difficult feature to expose to a VM through emulation, as
resizing a BAR is done on the host.  It can fail, and often does, but we
have no means via emulation of a PCIe REBAR capability to handle the error
cases.

A vfio-pci specific ioctl interface is also cumbersome as there are often
multiple devices within the same bridge aperture and handling them is a
challenge.  In the interface proposed here, expanding a BAR potentially
requires such devices to be soft-removed during the resize operation and
rescanned after, in order for all the necessary resources to be released.
A pci-sysfs interface is also more universal than a vfio specific
interface.

Please see the ABI documentation update for usage.

Link: https://lore.kernel.org/r/166336088796.3597940.14973499936692558556.stgit@omen
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Cc: Krzysztof Wilczyński <kw@linux.com>

Documentation/ABI/testing/sysfs-bus-pci

@@ -457,3 +457,36 @@ Description:
 
 		The file is writable if the PF is bound to a driver that
 		implements ->sriov_set_msix_vec_count().
+
+What:		/sys/bus/pci/devices/.../resourceN_resize
+Date:		September 2022
+Contact:	Alex Williamson <alex.williamson@redhat.com>
+Description:
+		These files provide an interface to PCIe Resizable BAR support.
+		A file is created for each BAR resource (N) supported by the
+		PCIe Resizable BAR extended capability of the device.  Reading
+		each file exposes the bitmap of available resource sizes:
+
+		# cat resource1_resize
+		00000000000001c0
+
+		The bitmap represents supported resource sizes for the BAR,
+		where bit0 = 1MB, bit1 = 2MB, bit2 = 4MB, etc.  In the above
+		example the device supports 64MB, 128MB, and 256MB BAR sizes.
+
+		When writing the file, the user provides the bit position of
+		the desired resource size, for example:
+
+		# echo 7 > resource1_resize
+
+		This indicates to set the size value corresponding to bit 7,
+		128MB.  The resulting size is 2 ^ (bit# + 20).  This definition
+		matches the PCIe specification of this capability.
+
+		In order to make use of resource resizing, all PCI drivers must
+		be unbound from the device and peer devices under the same
+		parent bridge may need to be soft removed.  In the case of
+		VGA devices, writing a resize value will remove low level
+		console drivers from the device.  Raw users of pci-sysfs
+		resourceN attributes must be terminated prior to resizing.
+		Success of the resizing operation is not guaranteed.

drivers/pci/pci-sysfs.c

@@ -28,6 +28,7 @@
 #include <linux/pm_runtime.h>
 #include <linux/msi.h>
 #include <linux/of.h>
+#include <linux/aperture.h>
 #include "pci.h"
 
 static int sysfs_initialized;	/* = 0 */
@@ -1373,6 +1374,112 @@ static const struct attribute_group pci_dev_reset_attr_group = {
 	.is_visible = pci_dev_reset_attr_is_visible,
 };
 
+#define pci_dev_resource_resize_attr(n)					\
+static ssize_t resource##n##_resize_show(struct device *dev,		\
+					 struct device_attribute *attr,	\
+					 char * buf)			\
+{									\
+	struct pci_dev *pdev = to_pci_dev(dev);				\
+	ssize_t ret;							\
+									\
+	pci_config_pm_runtime_get(pdev);				\
+									\
+	ret = sysfs_emit(buf, "%016llx\n",				\
+			 (u64)pci_rebar_get_possible_sizes(pdev, n));	\
+									\
+	pci_config_pm_runtime_put(pdev);				\
+									\
+	return ret;							\
+}									\
+									\
+static ssize_t resource##n##_resize_store(struct device *dev,		\
+					  struct device_attribute *attr,\
+					  const char *buf, size_t count)\
+{									\
+	struct pci_dev *pdev = to_pci_dev(dev);				\
+	unsigned long size, flags;					\
+	int ret, i;							\
+	u16 cmd;							\
+									\
+	if (kstrtoul(buf, 0, &size) < 0)				\
+		return -EINVAL;						\
+									\
+	device_lock(dev);						\
+	if (dev->driver) {						\
+		ret = -EBUSY;						\
+		goto unlock;						\
+	}								\
+									\
+	pci_config_pm_runtime_get(pdev);				\
+									\
+	if ((pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA) {		\
+		ret = aperture_remove_conflicting_pci_devices(pdev,	\
+						"resourceN_resize");	\
+		if (ret)						\
+			goto pm_put;					\
+	}								\
+									\
+	pci_read_config_word(pdev, PCI_COMMAND, &cmd);			\
+	pci_write_config_word(pdev, PCI_COMMAND,			\
+			      cmd & ~PCI_COMMAND_MEMORY);		\
+									\
+	flags = pci_resource_flags(pdev, n);				\
+									\
+	pci_remove_resource_files(pdev);				\
+									\
+	for (i = 0; i < PCI_STD_NUM_BARS; i++) {			\
+		if (pci_resource_len(pdev, i) &&			\
+		    pci_resource_flags(pdev, i) == flags)		\
+			pci_release_resource(pdev, i);			\
+	}								\
+									\
+	ret = pci_resize_resource(pdev, n, size);			\
+									\
+	pci_assign_unassigned_bus_resources(pdev->bus);			\
+									\
+	if (pci_create_resource_files(pdev))				\
+		pci_warn(pdev, "Failed to recreate resource files after BAR resizing\n");\
+									\
+	pci_write_config_word(pdev, PCI_COMMAND, cmd);			\
+pm_put:									\
+	pci_config_pm_runtime_put(pdev);				\
+unlock:									\
+	device_unlock(dev);						\
+									\
+	return ret ? ret : count;					\
+}									\
+static DEVICE_ATTR_RW(resource##n##_resize)
+
+pci_dev_resource_resize_attr(0);
+pci_dev_resource_resize_attr(1);
+pci_dev_resource_resize_attr(2);
+pci_dev_resource_resize_attr(3);
+pci_dev_resource_resize_attr(4);
+pci_dev_resource_resize_attr(5);
+
+static struct attribute *resource_resize_attrs[] = {
+	&dev_attr_resource0_resize.attr,
+	&dev_attr_resource1_resize.attr,
+	&dev_attr_resource2_resize.attr,
+	&dev_attr_resource3_resize.attr,
+	&dev_attr_resource4_resize.attr,
+	&dev_attr_resource5_resize.attr,
+	NULL,
+};
+
+static umode_t resource_resize_is_visible(struct kobject *kobj,
+					  struct attribute *a, int n)
+{
+	struct pci_dev *pdev = to_pci_dev(kobj_to_dev(kobj));
+
+	return pci_rebar_get_current_size(pdev, n) < 0 ? 0 : a->mode;
+}
+
+static const struct attribute_group pci_dev_resource_resize_group = {
+	.attrs = resource_resize_attrs,
+	.is_visible = resource_resize_is_visible,
+};
+
 int __must_check pci_create_sysfs_dev_files(struct pci_dev *pdev)
 {
 	if (!sysfs_initialized)
@@ -1494,6 +1601,7 @@ const struct attribute_group *pci_dev_groups[] = {
 #ifdef CONFIG_ACPI
 	&pci_dev_acpi_attr_group,
 #endif
+	&pci_dev_resource_resize_group,
 	NULL,
 };