From: Xiao Wang <xiao.w.wang@intel.com>
To: tiwei.bie@intel.com
Cc: dev@dpdk.org, xiaolong.ye@intel.com, zhihong.wang@intel.com,
Xiao Wang <xiao.w.wang@intel.com>
Subject: [dpdk-dev] [PATCH] net/ifc: add live migration support
Date: Mon, 10 Sep 2018 19:01:23 +0800 [thread overview]
Message-ID: <20180910110123.138273-1-xiao.w.wang@intel.com> (raw)
IFCVF can help to log dirty page in live migration stage,
each queue's index can be read and configured to support
VHOST_USER_GET_VRING_BASE and VHOST_USER_SET_VRING_BASE.
Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
---
drivers/net/ifc/base/ifcvf.c | 33 +++++++++++++++++++-
drivers/net/ifc/base/ifcvf.h | 7 +++++
drivers/net/ifc/ifcvf_vdpa.c | 71 ++++++++++++++++++++++++++++++++++++++++++--
3 files changed, 108 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ifc/base/ifcvf.c b/drivers/net/ifc/base/ifcvf.c
index 4b22d9ed1..3c0b2dff6 100644
--- a/drivers/net/ifc/base/ifcvf.c
+++ b/drivers/net/ifc/base/ifcvf.c
@@ -249,7 +249,7 @@ ifcvf_hw_disable(struct ifcvf_hw *hw)
IFCVF_WRITE_REG16(IFCVF_MSI_NO_VECTOR, &cfg->queue_msix_vector);
ring_state = *(u32 *)(hw->lm_cfg + IFCVF_LM_RING_STATE_OFFSET +
(i / 2) * IFCVF_LM_CFG_SIZE + (i % 2) * 4);
- hw->vring[i].last_avail_idx = (u16)ring_state;
+ hw->vring[i].last_avail_idx = (u16)(ring_state >> 16);
hw->vring[i].last_used_idx = (u16)(ring_state >> 16);
}
}
@@ -278,6 +278,37 @@ ifcvf_stop_hw(struct ifcvf_hw *hw)
ifcvf_reset(hw);
}
+void
+ifcvf_enable_logging(struct ifcvf_hw *hw, u64 log_base, u64 log_size)
+{
+ u8 *lm_cfg;
+
+ lm_cfg = hw->lm_cfg;
+
+ *(u32 *)(lm_cfg + IFCVF_LM_BASE_ADDR_LOW) =
+ log_base & IFCVF_32_BIT_MASK;
+
+ *(u32 *)(lm_cfg + IFCVF_LM_BASE_ADDR_HIGH) =
+ (log_base >> 32) & IFCVF_32_BIT_MASK;
+
+ *(u32 *)(lm_cfg + IFCVF_LM_END_ADDR_LOW) =
+ (log_base + log_size) & IFCVF_32_BIT_MASK;
+
+ *(u32 *)(lm_cfg + IFCVF_LM_END_ADDR_HIGH) =
+ ((log_base + log_size) >> 32) & IFCVF_32_BIT_MASK;
+
+ *(u32 *)(lm_cfg + IFCVF_LM_LOGGING_CTRL) = IFCVF_LM_ENABLE_VF;
+}
+
+void
+ifcvf_disable_logging(struct ifcvf_hw *hw)
+{
+ u8 *lm_cfg;
+
+ lm_cfg = hw->lm_cfg;
+ *(u32 *)(lm_cfg + IFCVF_LM_LOGGING_CTRL) = IFCVF_LM_DISABLE;
+}
+
void
ifcvf_notify_queue(struct ifcvf_hw *hw, u16 qid)
{
diff --git a/drivers/net/ifc/base/ifcvf.h b/drivers/net/ifc/base/ifcvf.h
index badacb615..f026c70ab 100644
--- a/drivers/net/ifc/base/ifcvf.h
+++ b/drivers/net/ifc/base/ifcvf.h
@@ -49,6 +49,7 @@
#define IFCVF_LM_DISABLE 0x0
#define IFCVF_LM_ENABLE_VF 0x1
#define IFCVF_LM_ENABLE_PF 0x3
+#define IFCVF_LOG_BASE 0x100000000000
#define IFCVF_32_BIT_MASK 0xffffffff
@@ -142,6 +143,12 @@ ifcvf_start_hw(struct ifcvf_hw *hw);
void
ifcvf_stop_hw(struct ifcvf_hw *hw);
+void
+ifcvf_enable_logging(struct ifcvf_hw *hw, u64 log_base, u64 log_size);
+
+void
+ifcvf_disable_logging(struct ifcvf_hw *hw);
+
void
ifcvf_notify_queue(struct ifcvf_hw *hw, u16 qid);
diff --git a/drivers/net/ifc/ifcvf_vdpa.c b/drivers/net/ifc/ifcvf_vdpa.c
index 88d814037..3c5430dc0 100644
--- a/drivers/net/ifc/ifcvf_vdpa.c
+++ b/drivers/net/ifc/ifcvf_vdpa.c
@@ -7,6 +7,7 @@
#include <fcntl.h>
#include <sys/ioctl.h>
#include <sys/epoll.h>
+#include <linux/virtio_net.h>
#include <rte_malloc.h>
#include <rte_memory.h>
@@ -276,12 +277,30 @@ vdpa_ifcvf_start(struct ifcvf_internal *internal)
return ifcvf_start_hw(&internal->hw);
}
+static void
+ifcvf_used_ring_log(struct ifcvf_hw *hw, uint32_t queue, uint8_t *log_buf)
+{
+ uint32_t i, size;
+ uint64_t pfn;
+
+ pfn = hw->vring[queue].used / PAGE_SIZE;
+ size = hw->vring[queue].size * sizeof(struct vring_used_elem) +
+ sizeof(__virtio16) * 3;
+
+ for (i = 0; i <= size / PAGE_SIZE; i++)
+ __sync_fetch_and_or_8(&log_buf[(pfn + i) / 8],
+ 1 << ((pfn + i) % 8));
+}
+
static void
vdpa_ifcvf_stop(struct ifcvf_internal *internal)
{
struct ifcvf_hw *hw = &internal->hw;
uint32_t i;
int vid;
+ uint64_t features;
+ uint64_t log_base, log_size;
+ uint8_t *log_buf;
vid = internal->vid;
ifcvf_stop_hw(hw);
@@ -289,6 +308,21 @@ vdpa_ifcvf_stop(struct ifcvf_internal *internal)
for (i = 0; i < hw->nr_vring; i++)
rte_vhost_set_vring_base(vid, i, hw->vring[i].last_avail_idx,
hw->vring[i].last_used_idx);
+
+ rte_vhost_get_negotiated_features(vid, &features);
+ if (RTE_VHOST_NEED_LOG(features)) {
+ ifcvf_disable_logging(hw);
+ rte_vhost_get_log_base(internal->vid, &log_base, &log_size);
+ rte_vfio_container_dma_unmap(internal->vfio_container_fd,
+ log_base, IFCVF_LOG_BASE, log_size);
+ /*
+ * IFCVF marks dirty memory pages for only packet buffer,
+ * SW helps to mark the used ring as dirty after device stops.
+ */
+ log_buf = (uint8_t *)(uintptr_t)log_base;
+ for (i = 0; i < hw->nr_vring; i++)
+ ifcvf_used_ring_log(hw, i, log_buf);
+ }
}
#define MSIX_IRQ_SET_BUF_LEN (sizeof(struct vfio_irq_set) + \
@@ -548,6 +582,35 @@ ifcvf_dev_close(int vid)
return 0;
}
+static int
+ifcvf_set_features(int vid)
+{
+ uint64_t features;
+ int did;
+ struct internal_list *list;
+ struct ifcvf_internal *internal;
+ uint64_t log_base, log_size;
+
+ did = rte_vhost_get_vdpa_device_id(vid);
+ list = find_internal_resource_by_did(did);
+ if (list == NULL) {
+ DRV_LOG(ERR, "Invalid device id: %d", did);
+ return -1;
+ }
+
+ internal = list->internal;
+ rte_vhost_get_negotiated_features(vid, &features);
+
+ if (RTE_VHOST_NEED_LOG(features)) {
+ rte_vhost_get_log_base(vid, &log_base, &log_size);
+ rte_vfio_container_dma_map(internal->vfio_container_fd,
+ log_base, IFCVF_LOG_BASE, log_size);
+ ifcvf_enable_logging(&internal->hw, IFCVF_LOG_BASE, log_size);
+ }
+
+ return 0;
+}
+
static int
ifcvf_get_vfio_group_fd(int vid)
{
@@ -664,7 +727,7 @@ struct rte_vdpa_dev_ops ifcvf_ops = {
.dev_conf = ifcvf_dev_config,
.dev_close = ifcvf_dev_close,
.set_vring_state = NULL,
- .set_features = NULL,
+ .set_features = ifcvf_set_features,
.migration_done = NULL,
.get_vfio_group_fd = ifcvf_get_vfio_group_fd,
.get_vfio_device_fd = ifcvf_get_vfio_device_fd,
@@ -699,7 +762,11 @@ ifcvf_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
features = ifcvf_get_features(&internal->hw);
internal->features = (features &
~(1ULL << VIRTIO_F_IOMMU_PLATFORM)) |
- (1ULL << VHOST_USER_F_PROTOCOL_FEATURES);
+ (1ULL << VIRTIO_NET_F_GUEST_ANNOUNCE) |
+ (1ULL << VIRTIO_NET_F_CTRL_VQ) |
+ (1ULL << VIRTIO_NET_F_STATUS) |
+ (1ULL << VHOST_USER_F_PROTOCOL_FEATURES) |
+ (1ULL << VHOST_F_LOG_ALL);
internal->dev_addr.pci_addr = pci_dev->addr;
internal->dev_addr.type = PCI_ADDR;
--
2.15.1
next reply other threads:[~2018-09-10 11:09 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-09-10 11:01 Xiao Wang [this message]
2018-09-13 20:39 ` Ye Xiaolong
2018-09-19 14:36 ` Zhang, Qi Z
2018-09-20 23:55 ` Ferruh Yigit
2018-10-02 14:45 ` Ferruh Yigit
2018-10-08 2:59 ` Wang, Xiao W
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180910110123.138273-1-xiao.w.wang@intel.com \
--to=xiao.w.wang@intel.com \
--cc=dev@dpdk.org \
--cc=tiwei.bie@intel.com \
--cc=xiaolong.ye@intel.com \
--cc=zhihong.wang@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).