From patchwork Mon May 25 12:25:41 2009 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Michael S. Tsirkin" X-Patchwork-Id: 25816 Received: from vger.kernel.org (vger.kernel.org [209.132.176.167]) by demeter.kernel.org (8.14.2/8.14.2) with ESMTP id n4PCTSuS006645 for ; Mon, 25 May 2009 12:29:28 GMT Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751985AbZEYM3X (ORCPT ); Mon, 25 May 2009 08:29:23 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752866AbZEYM3X (ORCPT ); Mon, 25 May 2009 08:29:23 -0400 Received: from mx2.redhat.com ([66.187.237.31]:33543 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752354AbZEYM3U (ORCPT ); Mon, 25 May 2009 08:29:20 -0400 Received: from int-mx2.corp.redhat.com (int-mx2.corp.redhat.com [172.16.27.26]) by mx2.redhat.com (8.13.8/8.13.8) with ESMTP id n4PCQfFb022832; Mon, 25 May 2009 08:26:41 -0400 Received: from ns3.rdu.redhat.com (ns3.rdu.redhat.com [10.11.255.199]) by int-mx2.corp.redhat.com (8.13.1/8.13.1) with ESMTP id n4PCQeso029431; Mon, 25 May 2009 08:26:40 -0400 Received: from redhat.com (vpn-10-50.str.redhat.com [10.32.10.50]) by ns3.rdu.redhat.com (8.13.8/8.13.8) with ESMTP id n4PCQaRa014945; Mon, 25 May 2009 08:26:37 -0400 Date: Mon, 25 May 2009 15:25:41 +0300 From: "Michael S. Tsirkin" To: Paul Brook , Avi Kivity , qemu-devel@nongnu.org, Carsten Otte , kvm@vger.kernel.org, Rusty Russell , virtualization@lists.linux-foundation.org, Christian Borntraeger , Blue Swirl Subject: [PATCH 05/11] qemu: MSI-X support functions Message-ID: <20090525122541.GF5046@redhat.com> References: MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.18 (2008-05-17) X-Scanned-By: MIMEDefang 2.58 on 172.16.27.26 Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org Add functions implementing MSI-X support. First user will be virtio-pci. Note that platform must set a flag to declare MSI supported. For PC this will be set by APIC. Signed-off-by: Michael S. Tsirkin --- Makefile.target | 2 +- hw/msix.c | 423 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ hw/msix.h | 35 +++++ hw/pci.h | 20 +++ 4 files changed, 479 insertions(+), 1 deletions(-) create mode 100644 hw/msix.c create mode 100644 hw/msix.h diff --git a/Makefile.target b/Makefile.target index 664a1e3..87b2859 100644 --- a/Makefile.target +++ b/Makefile.target @@ -486,7 +486,7 @@ endif #CONFIG_BSD_USER ifndef CONFIG_USER_ONLY OBJS=vl.o osdep.o monitor.o pci.o loader.o isa_mmio.o machine.o \ - gdbstub.o gdbstub-xml.o + gdbstub.o gdbstub-xml.o msix.o # virtio has to be here due to weird dependency between PCI and virtio-net. # need to fix this properly OBJS+=virtio-blk.o virtio-balloon.o virtio-net.o virtio-console.o diff --git a/hw/msix.c b/hw/msix.c new file mode 100644 index 0000000..4db2fc1 --- /dev/null +++ b/hw/msix.c @@ -0,0 +1,423 @@ +/* + * MSI-X device support + * + * This module includes support for MSI-X in pci devices. + * + * Author: Michael S. Tsirkin + * + * Copyright (c) 2009, Red Hat Inc, Michael S. Tsirkin (mst@redhat.com) + * + * This work is licensed under the terms of the GNU GPL, version 2. See + * the COPYING file in the top-level directory. + */ + +#include "hw.h" +#include "msix.h" +#include "pci.h" + +/* Declaration from linux/pci_regs.h */ +#define PCI_CAP_ID_MSIX 0x11 /* MSI-X */ +#define PCI_MSIX_FLAGS 2 /* Table at lower 11 bits */ +#define PCI_MSIX_FLAGS_QSIZE 0x7FF +#define PCI_MSIX_FLAGS_ENABLE (1 << 15) +#define PCI_MSIX_FLAGS_BIRMASK (7 << 0) + +/* MSI-X capability structure */ +#define MSIX_TABLE_OFFSET 4 +#define MSIX_PBA_OFFSET 8 +#define MSIX_CAP_LENGTH 12 + +/* MSI enable bit is in byte 1 in FLAGS register */ +#define MSIX_ENABLE_OFFSET (PCI_MSIX_FLAGS + 1) +#define MSIX_ENABLE_MASK (PCI_MSIX_FLAGS_ENABLE >> 8) + +/* MSI-X table format */ +#define MSIX_MSG_ADDR 0 +#define MSIX_MSG_UPPER_ADDR 4 +#define MSIX_MSG_DATA 8 +#define MSIX_VECTOR_CTRL 12 +#define MSIX_ENTRY_SIZE 16 +#define MSIX_VECTOR_MASK 0x1 + +/* How much space does an MSIX table need. */ +/* The spec requires giving the table structure + * a 4K aligned region all by itself. Align it to + * target pages so that drivers can do passthrough + * on the rest of the region. */ +#define MSIX_PAGE_SIZE TARGET_PAGE_ALIGN(0x1000) +/* Reserve second half of the page for pending bits */ +#define MSIX_PAGE_PENDING (MSIX_PAGE_SIZE / 2) +#define MSIX_MAX_ENTRIES 32 + + +#ifdef MSIX_DEBUG +#define DEBUG(fmt, ...) \ + do { \ + fprintf(stderr, "%s: " fmt, __func__ , __VA_ARGS__); \ + } while (0) +#else +#define DEBUG(fmt, ...) do { } while(0) +#endif + +/* Flag to globally disable MSI-X support */ +int msix_disable; + +/* Flag for interrupt controller to declare MSI-X support */ +int msix_supported; + +/* Add MSI-X capability to the config space for the device. */ +/* Given a bar and its size, add MSI-X table on top of it + * and fill MSI-X capability in the config space. + * Original bar size must be a power of 2 or 0. + * New bar size is returned. */ +static int msix_add_config(struct PCIDevice *pdev, unsigned short nentries, + unsigned bar_nr, unsigned bar_size) +{ + int config_offset; + uint8_t *config; + uint32_t new_size; + + if (nentries < 1 || nentries > PCI_MSIX_FLAGS_QSIZE + 1) + return -EINVAL; + if (bar_size > 0x80000000) + return -ENOSPC; + + /* Add space for MSI-X structures */ + if (!bar_size) + new_size = MSIX_PAGE_SIZE; + else if (bar_size < MSIX_PAGE_SIZE) { + bar_size = MSIX_PAGE_SIZE; + new_size = MSIX_PAGE_SIZE * 2; + } else + new_size = bar_size * 2; + + pdev->msix_bar_size = new_size; + config_offset = pci_add_capability(pdev, PCI_CAP_ID_MSIX, MSIX_CAP_LENGTH); + if (config_offset < 0) + return config_offset; + config = pdev->config + config_offset; + + pci_set_word(config + PCI_MSIX_FLAGS, nentries - 1); + /* Table on top of BAR */ + pci_set_long(config + MSIX_TABLE_OFFSET, bar_size | bar_nr); + /* Pending bits on top of that */ + pci_set_long(config + MSIX_PBA_OFFSET, (bar_size + MSIX_PAGE_PENDING) | + bar_nr); + pdev->msix_cap = config_offset; + /* Make flags bit writeable. */ + pdev->mask[config_offset + MSIX_ENABLE_OFFSET] |= MSIX_ENABLE_MASK; + return 0; +} + +static void msix_free_irq_entries(PCIDevice *dev) +{ + int vector; + + for (vector = 0; vector < dev->msix_entries_nr; ++vector) + dev->msix_entry_used[vector] = 0; +} + +/* Handle MSI-X capability config write. */ +void msix_write_config(PCIDevice *dev, uint32_t addr, + uint32_t val, int len) +{ + unsigned enable_pos = dev->msix_cap + MSIX_ENABLE_OFFSET; + if (addr + len <= enable_pos || addr > enable_pos) + return; + + if (msix_enabled(dev)) + qemu_set_irq(dev->irq[0], 0); +} + +static uint32_t msix_mmio_readl(void *opaque, target_phys_addr_t addr) +{ + PCIDevice *dev = opaque; + unsigned int offset = addr & (MSIX_PAGE_SIZE - 1); + void *page = dev->msix_table_page; + uint32_t val = 0; + + memcpy(&val, (void *)((char *)page + offset), 4); + + return val; +} + +static uint32_t msix_mmio_read_unallowed(void *opaque, target_phys_addr_t addr) +{ + fprintf(stderr, "MSI-X: only dword read is allowed!\n"); + return 0; +} + +static uint8_t msix_pending_mask(int vector) +{ + return 1 << (vector % 8); +} + +static uint8_t *msix_pending_byte(PCIDevice *dev, int vector) +{ + return dev->msix_table_page + MSIX_PAGE_PENDING + vector / 8; +} + +static int msix_is_pending(PCIDevice *dev, int vector) +{ + return *msix_pending_byte(dev, vector) & msix_pending_mask(vector); +} + +static void msix_set_pending(PCIDevice *dev, int vector) +{ + *msix_pending_byte(dev, vector) |= msix_pending_mask(vector); +} + +static void msix_clr_pending(PCIDevice *dev, int vector) +{ + *msix_pending_byte(dev, vector) &= ~msix_pending_mask(vector); +} + +static int msix_is_masked(PCIDevice *dev, int vector) +{ + unsigned offset = vector * MSIX_ENTRY_SIZE + MSIX_VECTOR_CTRL; + return dev->msix_table_page[offset] & MSIX_VECTOR_MASK; +} + +static void msix_mmio_writel(void *opaque, target_phys_addr_t addr, + uint32_t val) +{ + PCIDevice *dev = opaque; + unsigned int offset = addr & (MSIX_PAGE_SIZE - 1); + int vector = offset / MSIX_ENTRY_SIZE; + memcpy(dev->msix_table_page + offset, &val, 4); + if (!msix_is_masked(dev, vector) && msix_is_pending(dev, vector)) { + msix_clr_pending(dev, vector); + msix_notify(dev, vector); + } +} + +static void msix_mmio_write_unallowed(void *opaque, target_phys_addr_t addr, + uint32_t val) +{ + fprintf(stderr, "MSI-X: only dword write is allowed!\n"); +} + +static CPUWriteMemoryFunc *msix_mmio_write[] = { + msix_mmio_write_unallowed, msix_mmio_write_unallowed, msix_mmio_writel +}; + +static CPUReadMemoryFunc *msix_mmio_read[] = { + msix_mmio_read_unallowed, msix_mmio_read_unallowed, msix_mmio_readl +}; + +/* Should be called from device's map method. */ +void msix_mmio_map(PCIDevice *d, int region_num, + uint32_t addr, uint32_t size, int type) +{ + uint8_t *config = d->config + d->msix_cap; + uint32_t table = pci_get_long(config + MSIX_TABLE_OFFSET); + uint32_t offset = table & ~(MSIX_PAGE_SIZE - 1); + /* TODO: for assigned devices, we'll want to make it possible to map + * pending bits separately in case they are in a separate bar. */ + int table_bir = table & PCI_MSIX_FLAGS_BIRMASK; + + if (table_bir != region_num) + return; + if (size <= offset) + return; + cpu_register_physical_memory(addr + offset, size - offset, + d->msix_mmio_index); +} + +/* Initialize the MSI-X structures. Note: if MSI-X is supported, BAR size is + * modified, it should be retrieved with msix_bar_size. */ +int msix_init(struct PCIDevice *dev, unsigned short nentries, + unsigned bar_nr, unsigned bar_size) +{ + int ret = -ENOMEM; + /* Nothing to do if MSI is not supported by interrupt controller */ + if (!msix_supported) + return -ENOTTY; + + if (nentries > MSIX_MAX_ENTRIES) + return -EINVAL; + + dev->msix_entry_used = qemu_mallocz(MSIX_MAX_ENTRIES * + sizeof *dev->msix_entry_used); + if (!dev->msix_entry_used) + goto err_used; + + dev->msix_table_page = qemu_mallocz(MSIX_PAGE_SIZE); + if (!dev->msix_table_page) + goto err_page; + + dev->msix_mmio_index = cpu_register_io_memory(0, msix_mmio_read, + msix_mmio_write, dev); + if (dev->msix_mmio_index == -1) { + ret = -EBUSY; + goto err_index; + } + + dev->msix_entries_nr = nentries; + dev->cap_supported |= QEMU_PCI_CAP_MSIX; + /* If disabled, stop here. User can later load confiuration with MSI-X + * enabled. */ + if (msix_disable) + return 0; + + ret = msix_add_config(dev, nentries, bar_nr, bar_size); + if (ret) + goto err_config; + + dev->cap_present |= QEMU_PCI_CAP_MSIX; + return 0; + +err_config: + cpu_unregister_io_memory(dev->msix_mmio_index); +err_index: + qemu_free(dev->msix_table_page); + dev->msix_table_page = NULL; +err_page: + qemu_free(dev->msix_entry_used); + dev->msix_entry_used = NULL; +err_used: + return ret; +} + +/* Clean up resources for the device. */ +int msix_uninit(PCIDevice *dev) +{ + if (!(dev->cap_supported & QEMU_PCI_CAP_MSIX)) + return 0; + pci_del_capability(dev, PCI_CAP_ID_MSIX, MSIX_CAP_LENGTH); + dev->msix_cap = 0; + msix_free_irq_entries(dev); + dev->msix_entries_nr = 0; + cpu_unregister_io_memory(dev->msix_mmio_index); + qemu_free(dev->msix_table_page); + dev->msix_table_page = NULL; + qemu_free(dev->msix_entry_used); + dev->msix_entry_used = NULL; + dev->cap_present &= ~QEMU_PCI_CAP_MSIX; + dev->cap_supported &= ~QEMU_PCI_CAP_MSIX; + return 0; +} + +void msix_save(PCIDevice *dev, QEMUFile *f) +{ + unsigned nentries = (pci_get_word(dev->config + PCI_MSIX_FLAGS) & + PCI_MSIX_FLAGS_QSIZE) + 1; + qemu_put_buffer(f, dev->msix_table_page, nentries * MSIX_ENTRY_SIZE); + qemu_put_buffer(f, dev->msix_table_page + MSIX_PAGE_PENDING, + (nentries + 7) / 8); +} + +/* Should be called after restoring the config space. */ +int msix_load(PCIDevice *dev, QEMUFile *f) +{ + uint8_t offset = pci_find_capability(dev, PCI_CAP_ID_MSIX); + unsigned nentries; + + if (!!(dev->cap_present & QEMU_PCI_CAP_MSIX) == !!offset) { + fprintf(stderr, "MSI-X bit set but no capability is present\n"); + return -EINVAL; + } + + msix_free_irq_entries(dev); + + if (!dev->cap_present & QEMU_PCI_CAP_MSIX) + return 0; + + /* Sanity check: we probably could add more of these. */ + nentries = (pci_get_word(dev->config + PCI_MSIX_FLAGS) & + PCI_MSIX_FLAGS_QSIZE) + 1; + if (nentries > MSIX_MAX_ENTRIES) { + fprintf(stderr, "msix_load: nentries mismatch: %d > %d\n", + nentries, dev->msix_entries_nr); + return -EINVAL; + } + + /* Make flags bit writeable. */ + dev->mask[offset + MSIX_ENABLE_OFFSET] |= MSIX_ENABLE_MASK; + /* Reserve space used by this capability */ + pci_reserve_capability(dev, offset, MSIX_CAP_LENGTH); + /* Store the new offset */ + dev->msix_cap = offset; + + dev->msix_entries_nr = nentries; + + qemu_get_buffer(f, dev->msix_table_page, nentries * MSIX_ENTRY_SIZE); + qemu_get_buffer(f, dev->msix_table_page + MSIX_PAGE_PENDING, + (nentries + 7) / 8); + + return 0; +} + +/* Does device support MSI-X? */ +int msix_present(PCIDevice *dev) +{ + return dev->cap_present & QEMU_PCI_CAP_MSIX; +} + +/* Is MSI-X enabled? */ +int msix_enabled(PCIDevice *dev) +{ + return (dev->cap_present & QEMU_PCI_CAP_MSIX) && + (dev->config[dev->msix_cap + MSIX_ENABLE_OFFSET] & + MSIX_ENABLE_MASK); +} + +/* Size of bar where MSI-X table resides, or 0 if MSI-X not supported. */ +uint32_t msix_bar_size(PCIDevice *dev) +{ + return (dev->cap_present & QEMU_PCI_CAP_MSIX) ? + dev->msix_bar_size : 0; +} + +/* Send an MSI-X message */ +void msix_notify(PCIDevice *dev, unsigned vector) +{ + uint8_t *table_entry = dev->msix_table_page + vector * MSIX_ENTRY_SIZE; + uint64_t address; + uint32_t data; + + if (vector >= dev->msix_entries_nr || !dev->msix_entry_used[vector]) + return; + if (msix_is_masked(dev, vector)) { + msix_set_pending(dev, vector); + return; + } + + address = pci_get_long(table_entry + MSIX_MSG_UPPER_ADDR); + address = (address << 32) | pci_get_long(table_entry + MSIX_MSG_ADDR); + data = pci_get_long(table_entry + MSIX_MSG_DATA); + stl_phys(address, data); +} + +void msix_reset(PCIDevice *dev) +{ + if (!(dev->cap_present & QEMU_PCI_CAP_MSIX)) + return; + msix_free_irq_entries(dev); + dev->config[dev->msix_cap + MSIX_ENABLE_OFFSET] &= MSIX_ENABLE_MASK; + memset(dev->msix_table_page, 0, MSIX_PAGE_SIZE); +} + +/* PCI spec suggests that devices make it possible for software to configure + * less vectors than supported by the device, but does not specify a standard + * mechanism for devices to do so. + * + * We support this by asking devices to declare vectors software is going to + * actually use, and checking this on the notification path. Devices that + * don't want to follow the spec suggestion can declare all vectors as used. */ + +/* Mark vector as used. */ +int msix_vector_use(PCIDevice *dev, unsigned vector) +{ + if (vector >= dev->msix_entries_nr) + return -EINVAL; + dev->msix_entry_used[vector]++; + return 0; +} + +/* Mark vector as unused. */ +void msix_vector_unuse(PCIDevice *dev, unsigned vector) +{ + if (vector < dev->msix_entries_nr && dev->msix_entry_used[vector]) + --dev->msix_entry_used[vector]; +} diff --git a/hw/msix.h b/hw/msix.h new file mode 100644 index 0000000..79e84a3 --- /dev/null +++ b/hw/msix.h @@ -0,0 +1,35 @@ +#ifndef QEMU_MSIX_H +#define QEMU_MSIX_H + +#include "qemu-common.h" + +int msix_init(PCIDevice *pdev, unsigned short nentries, + unsigned bar_nr, unsigned bar_size); + +void msix_write_config(PCIDevice *pci_dev, uint32_t address, + uint32_t val, int len); + +void msix_mmio_map(PCIDevice *pci_dev, int region_num, + uint32_t addr, uint32_t size, int type); + +int msix_uninit(PCIDevice *d); + +void msix_save(PCIDevice *dev, QEMUFile *f); +int msix_load(PCIDevice *dev, QEMUFile *f); + +int msix_enabled(PCIDevice *dev); +int msix_present(PCIDevice *dev); + +uint32_t msix_bar_size(PCIDevice *dev); + +int msix_vector_use(PCIDevice *dev, unsigned vector); +void msix_vector_unuse(PCIDevice *dev, unsigned vector); + +void msix_notify(PCIDevice *dev, unsigned vector); + +void msix_reset(PCIDevice *dev); + +extern int msix_disable; +extern int msix_supported; + +#endif diff --git a/hw/pci.h b/hw/pci.h index 4e112a3..6da626b 100644 --- a/hw/pci.h +++ b/hw/pci.h @@ -156,6 +156,11 @@ typedef struct PCIIORegion { /* Size of the standard PCI config space */ #define PCI_CONFIG_SPACE_SIZE 0x100 +/* Bits in cap_supported/cap_present fields. */ +enum { + QEMU_PCI_CAP_MSIX = 0x1, +}; + struct PCIDevice { DeviceState qdev; /* PCI config space */ @@ -189,6 +194,21 @@ struct PCIDevice { /* Capability bits for save/load */ uint32_t cap_supported; uint32_t cap_present; + + /* Offset of MSI-X capability in config space */ + uint8_t msix_cap; + + /* MSI-X entries */ + int msix_entries_nr; + + /* Space to store MSIX table */ + uint8_t *msix_table_page; + /* MMIO index used to map MSIX table and pending bit entries. */ + int msix_mmio_index; + /* Reference-count for entries actually in use by driver. */ + unsigned *msix_entry_used; + /* Region including the MSI-X table */ + uint32_t msix_bar_size; }; PCIDevice *pci_register_device(PCIBus *bus, const char *name,