um: add PCI over virtio emulation driver

To support testing of PCI/PCIe drivers in UML, add a PCI bus
support driver. This driver uses virtio, which in UML is really
just vhost-user, to talk to devices, and adds the devices to
the virtual PCI bus in the system.

Since virtio already allows DMA/bus mastering this really isn't
all that hard, of course we need the logic_iomem infrastructure
that was added by a previous patch.

The protocol to talk to the device is has a few fairly simple
messages for reading to/writing from config and IO spaces, and
messages for the device to send the various interrupts (INT#,
MSI/MSI-X and while suspended PME#).

Note that currently no offical virtio device ID is assigned for
this protocol, as a consequence this patch requires defining it
in the Kconfig, with a default that makes the driver refuse to
work at all.

Finally, in order to add support for MSI/MSI-X interrupts, some
small changes are needed in the UML IRQ code, it needs to have
more interrupts, changing NR_IRQS from 64 to 128 if this driver
is enabled, but not actually use them for anything so that the
generic IRQ domain/MSI infrastructure can allocate IRQ numbers.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Richard Weinberger <richard@nod.at>

arch/um/Kconfig

@@ -15,7 +15,7 @@ config UML
 	select HAVE_FUTEX_CMPXCHG if FUTEX
 	select HAVE_DEBUG_KMEMLEAK
 	select HAVE_DEBUG_BUGVERBOSE
-	select NO_DMA
+	select NO_DMA if !UML_DMA_EMULATION
 	select GENERIC_IRQ_SHOW
 	select GENERIC_CPU_DEVICES
 	select HAVE_GCC_PLUGINS
@@ -26,10 +26,21 @@ config MMU
 	bool
 	default y
 
+config UML_DMA_EMULATION
+	bool
+
 config NO_IOMEM
 	bool "disable IOMEM" if EXPERT
+	depends on !INDIRECT_IOMEM
 	default y
 
+config UML_IOMEM_EMULATION
+	bool
+	select INDIRECT_IOMEM
+	select GENERIC_PCI_IOMAP
+	select GENERIC_IOMAP
+	select NO_GENERIC_PCI_IOPORT_MAP
+
 config NO_IOPORT_MAP
 	def_bool y
 

arch/um/drivers/Kconfig

@@ -357,3 +357,23 @@ config UML_RTC
 	  rtcwake, especially in time-travel mode. This driver enables that
 	  by providing a fake RTC clock that causes a wakeup at the right
 	  time.
+
+config UML_PCI_OVER_VIRTIO
+	bool "Enable PCI over VIRTIO device simulation"
+	# in theory, just VIRTIO is enough, but that causes recursion
+	depends on VIRTIO_UML
+	select FORCE_PCI
+	select UML_IOMEM_EMULATION
+	select UML_DMA_EMULATION
+	select PCI_MSI
+	select PCI_MSI_IRQ_DOMAIN
+	select PCI_LOCKLESS_CONFIG
+
+config UML_PCI_OVER_VIRTIO_DEVICE_ID
+	int "set the virtio device ID for PCI emulation"
+	default -1
+	depends on UML_PCI_OVER_VIRTIO
+	help
+	  There's no official device ID assigned (yet), set the one you
+	  wish to use for experimentation here. The default of -1 is
+	  not valid and will cause the driver to fail at probe.

arch/um/drivers/Makefile

@@ -64,6 +64,7 @@ obj-$(CONFIG_BLK_DEV_COW_COMMON) += cow_user.o
 obj-$(CONFIG_UML_RANDOM) += random.o
 obj-$(CONFIG_VIRTIO_UML) += virtio_uml.o
 obj-$(CONFIG_UML_RTC) += rtc.o
+obj-$(CONFIG_UML_PCI_OVER_VIRTIO) += virt-pci.o
 
 # pcap_user.o must be added explicitly.
 USER_OBJS := fd.o null.o pty.o tty.o xterm.o slip_common.o pcap_user.o vde_user.o vector_user.o

arch/um/include/asm/Kbuild

@@ -18,7 +18,6 @@ generic-y += mcs_spinlock.h
 generic-y += mmiowb.h
 generic-y += module.lds.h
 generic-y += param.h
-generic-y += pci.h
 generic-y += percpu.h
 generic-y += preempt.h
 generic-y += softirq_stack.h

arch/um/include/asm/io.h

@@ -3,16 +3,23 @@
 #define _ASM_UM_IO_H
 #include <linux/types.h>
 
+/* get emulated iomem (if desired) */
+#include <asm-generic/logic_io.h>
+
+#ifndef ioremap
 #define ioremap ioremap
 static inline void __iomem *ioremap(phys_addr_t offset, size_t size)
 {
 	return NULL;
 }
+#endif /* ioremap */
 
+#ifndef iounmap
 #define iounmap iounmap
 static inline void iounmap(void __iomem *addr)
 {
 }
+#endif /* iounmap */
 
 #include <asm-generic/io.h>
 

arch/um/include/asm/irq.h

@@ -31,7 +31,13 @@
 
 #endif
 
-#define NR_IRQS			64
+#define UM_LAST_SIGNAL_IRQ	64
+/* If we have (simulated) PCI MSI, allow 64 more interrupt numbers for it */
+#ifdef CONFIG_PCI_MSI
+#define NR_IRQS			(UM_LAST_SIGNAL_IRQ + 64)
+#else
+#define NR_IRQS			UM_LAST_SIGNAL_IRQ
+#endif /* CONFIG_PCI_MSI */
 
 #include <asm-generic/irq.h>
 #endif

arch/um/include/asm/msi.h

+#include <asm-generic/msi.h>

arch/um/include/asm/pci.h

+/* SPDX-License-Identifier: GPL-2.0-only */
+#ifndef __ASM_UM_PCI_H
+#define __ASM_UM_PCI_H
+#include <linux/types.h>
+#include <asm/io.h>
+
+#define PCIBIOS_MIN_IO		0
+#define PCIBIOS_MIN_MEM		0
+
+#define pcibios_assign_all_busses() 1
+
+extern int isa_dma_bridge_buggy;
+
+#ifdef CONFIG_PCI
+static inline int pci_get_legacy_ide_irq(struct pci_dev *dev, int channel)
+{
+	/* no legacy IRQs */
+	return -ENODEV;
+}
+#endif
+
+#ifdef CONFIG_PCI_DOMAINS
+static inline int pci_proc_domain(struct pci_bus *bus)
+{
+	/* always show the domain in /proc */
+	return 1;
+}
+#endif  /* CONFIG_PCI */
+
+#ifdef CONFIG_PCI_MSI_IRQ_DOMAIN
+/*
+ * This is a bit of an annoying hack, and it assumes we only have
+ * the virt-pci (if anything). Which is true, but still.
+ */
+void *pci_root_bus_fwnode(struct pci_bus *bus);
+#define pci_root_bus_fwnode	pci_root_bus_fwnode
+#endif
+
+#endif  /* __ASM_UM_PCI_H */

arch/um/kernel/Makefile

@@ -23,6 +23,7 @@ obj-$(CONFIG_BLK_DEV_INITRD) += initrd.o
 obj-$(CONFIG_GPROF)	+= gprof_syms.o
 obj-$(CONFIG_EARLY_PRINTK) += early_printk.o
 obj-$(CONFIG_STACKTRACE) += stacktrace.o
+obj-$(CONFIG_GENERIC_PCI_IOMAP) += ioport.o
 
 USER_OBJS := config.o
 

arch/um/kernel/ioport.c

+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2021 Intel Corporation
+ * Author: Johannes Berg <johannes@sipsolutions.net>
+ */
+#include <asm/iomap.h>
+#include <asm-generic/pci_iomap.h>
+
+void __iomem *__pci_ioport_map(struct pci_dev *dev, unsigned long port,
+			       unsigned int nr)
+{
+	return NULL;
+}

arch/um/kernel/irq.c

@@ -56,7 +56,7 @@ struct irq_entry {
 
 static DEFINE_SPINLOCK(irq_lock);
 static LIST_HEAD(active_fds);
-static DECLARE_BITMAP(irqs_allocated, NR_IRQS);
+static DECLARE_BITMAP(irqs_allocated, UM_LAST_SIGNAL_IRQ);
 static bool irqs_suspended;
 
 static void irq_io_loop(struct irq_reg *irq, struct uml_pt_regs *regs)
@@ -419,7 +419,8 @@ unsigned int do_IRQ(int irq, struct uml_pt_regs *regs)
 
 void um_free_irq(int irq, void *dev)
 {
-	if (WARN(irq < 0 || irq > NR_IRQS, "freeing invalid irq %d", irq))
+	if (WARN(irq < 0 || irq > UM_LAST_SIGNAL_IRQ,
+		 "freeing invalid irq %d", irq))
 		return;
 
 	free_irq_by_irq_and_dev(irq, dev);
@@ -648,7 +649,7 @@ void __init init_IRQ(void)
 
 	irq_set_chip_and_handler(TIMER_IRQ, &alarm_irq_type, handle_edge_irq);
 
-	for (i = 1; i < NR_IRQS; i++)
+	for (i = 1; i < UM_LAST_SIGNAL_IRQ; i++)
 		irq_set_chip_and_handler(i, &normal_irq_type, handle_edge_irq);
 	/* Initialize EPOLL Loop */
 	os_setup_epoll();

include/uapi/linux/virtio_pcidev.h

+/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */
+/*
+ * Copyright (C) 2021 Intel Corporation
+ * Author: Johannes Berg <johannes@sipsolutions.net>
+ */
+#ifndef _UAPI_LINUX_VIRTIO_PCIDEV_H
+#define _UAPI_LINUX_VIRTIO_PCIDEV_H
+#include <linux/types.h>
+
+/**
+ * enum virtio_pcidev_ops - virtual PCI device operations
+ * @VIRTIO_PCIDEV_OP_CFG_READ: read config space, size is 1, 2, 4 or 8;
+ *	the @data field should be filled in by the device (in little endian).
+ * @VIRTIO_PCIDEV_OP_CFG_WRITE: write config space, size is 1, 2, 4 or 8;
+ *	the @data field contains the data to write (in little endian).
+ * @VIRTIO_PCIDEV_OP_BAR_READ: read BAR mem/pio, size can be variable;
+ *	the @data field should be filled in by the device (in little endian).
+ * @VIRTIO_PCIDEV_OP_BAR_WRITE: write BAR mem/pio, size can be variable;
+ *	the @data field contains the data to write (in little endian).
+ * @VIRTIO_PCIDEV_OP_MMIO_MEMSET: memset MMIO, size is variable but
+ *	the @data field only has one byte (unlike @VIRTIO_PCIDEV_OP_MMIO_WRITE)
+ * @VIRTIO_PCIDEV_OP_INT: legacy INTx# pin interrupt, the addr field is 1-4 for
+ *	the number
+ * @VIRTIO_PCIDEV_OP_MSI: MSI(-X) interrupt, this message basically transports
+ *	the 16- or 32-bit write that would otherwise be done into memory,
+ *	analogous to the write messages (@VIRTIO_PCIDEV_OP_MMIO_WRITE) above
+ * @VIRTIO_PCIDEV_OP_PME: Dummy message whose content is ignored (and should be
+ *	all zeroes) to signal the PME# pin.
+ */
+enum virtio_pcidev_ops {
+	VIRTIO_PCIDEV_OP_RESERVED = 0,
+	VIRTIO_PCIDEV_OP_CFG_READ,
+	VIRTIO_PCIDEV_OP_CFG_WRITE,
+	VIRTIO_PCIDEV_OP_MMIO_READ,
+	VIRTIO_PCIDEV_OP_MMIO_WRITE,
+	VIRTIO_PCIDEV_OP_MMIO_MEMSET,
+	VIRTIO_PCIDEV_OP_INT,
+	VIRTIO_PCIDEV_OP_MSI,
+	VIRTIO_PCIDEV_OP_PME,
+};
+
+/**
+ * struct virtio_pcidev_msg - virtio PCI device operation
+ * @op: the operation to do
+ * @bar: the bar (only with BAR read/write messages)
+ * @reserved: reserved
+ * @size: the size of the read/write (in bytes)
+ * @addr: the address to read/write
+ * @data: the data, normally @size long, but just one byte for
+ *	%VIRTIO_PCIDEV_OP_MMIO_MEMSET
+ *
+ * Note: the fields are all in native (CPU) endian, however, the
+ * @data values will often be in little endian (see the ops above.)
+ */
+struct virtio_pcidev_msg {
+	__u8 op;
+	__u8 bar;
+	__u16 reserved;
+	__u32 size;
+	__u64 addr;
+	__u8 data[];
+};
+
+#endif /* _UAPI_LINUX_VIRTIO_PCIDEV_H */