[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PULL v2 41/58] Refactor out libvhost-user fault generation logic
From: |
Michael S. Tsirkin |
Subject: |
[PULL v2 41/58] Refactor out libvhost-user fault generation logic |
Date: |
Fri, 12 Jun 2020 10:52:32 -0400 |
From: Raphael Norwitz <raphael.norwitz@nutanix.com>
In libvhost-user, the incoming postcopy migration path for setting the
backend's memory tables has become convolued. In particular, moving the
logic which starts generating faults, having received the final ACK from
qemu can be moved to a separate function. This simplifies the code
substantially.
This logic will also be needed by the postcopy path once the
VHOST_USER_PROTOCOL_F_CONFIGURE_MEM_SLOTS feature is supported.
Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Message-Id: <1588533678-23450-7-git-send-email-raphael.norwitz@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
contrib/libvhost-user/libvhost-user.c | 147 ++++++++++++++------------
1 file changed, 79 insertions(+), 68 deletions(-)
diff --git a/contrib/libvhost-user/libvhost-user.c
b/contrib/libvhost-user/libvhost-user.c
index 3bca996c62..cccfa22209 100644
--- a/contrib/libvhost-user/libvhost-user.c
+++ b/contrib/libvhost-user/libvhost-user.c
@@ -583,6 +583,84 @@ map_ring(VuDev *dev, VuVirtq *vq)
return !(vq->vring.desc && vq->vring.used && vq->vring.avail);
}
+static bool
+generate_faults(VuDev *dev) {
+ int i;
+ for (i = 0; i < dev->nregions; i++) {
+ VuDevRegion *dev_region = &dev->regions[i];
+ int ret;
+#ifdef UFFDIO_REGISTER
+ /*
+ * We should already have an open ufd. Mark each memory
+ * range as ufd.
+ * Discard any mapping we have here; note I can't use MADV_REMOVE
+ * or fallocate to make the hole since I don't want to lose
+ * data that's already arrived in the shared process.
+ * TODO: How to do hugepage
+ */
+ ret = madvise((void *)(uintptr_t)dev_region->mmap_addr,
+ dev_region->size + dev_region->mmap_offset,
+ MADV_DONTNEED);
+ if (ret) {
+ fprintf(stderr,
+ "%s: Failed to madvise(DONTNEED) region %d: %s\n",
+ __func__, i, strerror(errno));
+ }
+ /*
+ * Turn off transparent hugepages so we dont get lose wakeups
+ * in neighbouring pages.
+ * TODO: Turn this backon later.
+ */
+ ret = madvise((void *)(uintptr_t)dev_region->mmap_addr,
+ dev_region->size + dev_region->mmap_offset,
+ MADV_NOHUGEPAGE);
+ if (ret) {
+ /*
+ * Note: This can happen legally on kernels that are configured
+ * without madvise'able hugepages
+ */
+ fprintf(stderr,
+ "%s: Failed to madvise(NOHUGEPAGE) region %d: %s\n",
+ __func__, i, strerror(errno));
+ }
+ struct uffdio_register reg_struct;
+ reg_struct.range.start = (uintptr_t)dev_region->mmap_addr;
+ reg_struct.range.len = dev_region->size + dev_region->mmap_offset;
+ reg_struct.mode = UFFDIO_REGISTER_MODE_MISSING;
+
+ if (ioctl(dev->postcopy_ufd, UFFDIO_REGISTER, ®_struct)) {
+ vu_panic(dev, "%s: Failed to userfault region %d "
+ "@%p + size:%zx offset: %zx: (ufd=%d)%s\n",
+ __func__, i,
+ dev_region->mmap_addr,
+ dev_region->size, dev_region->mmap_offset,
+ dev->postcopy_ufd, strerror(errno));
+ return false;
+ }
+ if (!(reg_struct.ioctls & ((__u64)1 << _UFFDIO_COPY))) {
+ vu_panic(dev, "%s Region (%d) doesn't support COPY",
+ __func__, i);
+ return false;
+ }
+ DPRINT("%s: region %d: Registered userfault for %"
+ PRIx64 " + %" PRIx64 "\n", __func__, i,
+ (uint64_t)reg_struct.range.start,
+ (uint64_t)reg_struct.range.len);
+ /* Now it's registered we can let the client at it */
+ if (mprotect((void *)(uintptr_t)dev_region->mmap_addr,
+ dev_region->size + dev_region->mmap_offset,
+ PROT_READ | PROT_WRITE)) {
+ vu_panic(dev, "failed to mprotect region %d for postcopy (%s)",
+ i, strerror(errno));
+ return false;
+ }
+ /* TODO: Stash 'zero' support flags somewhere */
+#endif
+ }
+
+ return true;
+}
+
static bool
vu_set_mem_table_exec_postcopy(VuDev *dev, VhostUserMsg *vmsg)
{
@@ -655,74 +733,7 @@ vu_set_mem_table_exec_postcopy(VuDev *dev, VhostUserMsg
*vmsg)
}
/* OK, now we can go and register the memory and generate faults */
- for (i = 0; i < dev->nregions; i++) {
- VuDevRegion *dev_region = &dev->regions[i];
- int ret;
-#ifdef UFFDIO_REGISTER
- /* We should already have an open ufd. Mark each memory
- * range as ufd.
- * Discard any mapping we have here; note I can't use MADV_REMOVE
- * or fallocate to make the hole since I don't want to lose
- * data that's already arrived in the shared process.
- * TODO: How to do hugepage
- */
- ret = madvise((void *)(uintptr_t)dev_region->mmap_addr,
- dev_region->size + dev_region->mmap_offset,
- MADV_DONTNEED);
- if (ret) {
- fprintf(stderr,
- "%s: Failed to madvise(DONTNEED) region %d: %s\n",
- __func__, i, strerror(errno));
- }
- /* Turn off transparent hugepages so we dont get lose wakeups
- * in neighbouring pages.
- * TODO: Turn this backon later.
- */
- ret = madvise((void *)(uintptr_t)dev_region->mmap_addr,
- dev_region->size + dev_region->mmap_offset,
- MADV_NOHUGEPAGE);
- if (ret) {
- /* Note: This can happen legally on kernels that are configured
- * without madvise'able hugepages
- */
- fprintf(stderr,
- "%s: Failed to madvise(NOHUGEPAGE) region %d: %s\n",
- __func__, i, strerror(errno));
- }
- struct uffdio_register reg_struct;
- reg_struct.range.start = (uintptr_t)dev_region->mmap_addr;
- reg_struct.range.len = dev_region->size + dev_region->mmap_offset;
- reg_struct.mode = UFFDIO_REGISTER_MODE_MISSING;
-
- if (ioctl(dev->postcopy_ufd, UFFDIO_REGISTER, ®_struct)) {
- vu_panic(dev, "%s: Failed to userfault region %d "
- "@%p + size:%zx offset: %zx: (ufd=%d)%s\n",
- __func__, i,
- dev_region->mmap_addr,
- dev_region->size, dev_region->mmap_offset,
- dev->postcopy_ufd, strerror(errno));
- return false;
- }
- if (!(reg_struct.ioctls & ((__u64)1 << _UFFDIO_COPY))) {
- vu_panic(dev, "%s Region (%d) doesn't support COPY",
- __func__, i);
- return false;
- }
- DPRINT("%s: region %d: Registered userfault for %"
- PRIx64 " + %" PRIx64 "\n", __func__, i,
- (uint64_t)reg_struct.range.start,
- (uint64_t)reg_struct.range.len);
- /* Now it's registered we can let the client at it */
- if (mprotect((void *)(uintptr_t)dev_region->mmap_addr,
- dev_region->size + dev_region->mmap_offset,
- PROT_READ | PROT_WRITE)) {
- vu_panic(dev, "failed to mprotect region %d for postcopy (%s)",
- i, strerror(errno));
- return false;
- }
- /* TODO: Stash 'zero' support flags somewhere */
-#endif
- }
+ (void)generate_faults(dev);
return false;
}
--
MST
- [PULL v2 29/58] pci: assert configuration access is within bounds, (continued)
- [PULL v2 29/58] pci: assert configuration access is within bounds, Michael S. Tsirkin, 2020/06/12
- [PULL v2 26/58] virtio-balloon: Provide an interface for free page reporting, Michael S. Tsirkin, 2020/06/12
- [PULL v2 33/58] hw/pci-host: Use the IEC binary prefix definitions, Michael S. Tsirkin, 2020/06/12
- [PULL v2 34/58] char-socket: return -1 in case of disconnect during tcp_chr_write, Michael S. Tsirkin, 2020/06/12
- [PULL v2 35/58] vhost-user-blk: delay vhost_user_blk_disconnect, Michael S. Tsirkin, 2020/06/12
- [PULL v2 37/58] Add vhost-user helper to get MemoryRegion data, Michael S. Tsirkin, 2020/06/12
- [PULL v2 36/58] Add helper to populate vhost-user message regions, Michael S. Tsirkin, 2020/06/12
- [PULL v2 38/58] Add VHOST_USER_PROTOCOL_F_CONFIGURE_MEM_SLOTS, Michael S. Tsirkin, 2020/06/12
- [PULL v2 39/58] Transmit vhost-user memory regions individually, Michael S. Tsirkin, 2020/06/12
- [PULL v2 40/58] Lift max memory slots limit imposed by vhost-user, Michael S. Tsirkin, 2020/06/12
- [PULL v2 41/58] Refactor out libvhost-user fault generation logic,
Michael S. Tsirkin <=
- [PULL v2 42/58] Support ram slot configuration in libvhost-user, Michael S. Tsirkin, 2020/06/12
- [PULL v2 43/58] Support adding individual regions in libvhost-user, Michael S. Tsirkin, 2020/06/12
- [PULL v2 44/58] Support individual region unmap in libvhost-user, Michael S. Tsirkin, 2020/06/12
- [PULL v2 45/58] Lift max ram slots limit in libvhost-user, Michael S. Tsirkin, 2020/06/12
- [PULL v2 46/58] libvhost-user: advertise vring features, Michael S. Tsirkin, 2020/06/12
- [PULL v2 47/58] hw/pci: Fix crash when running QEMU with "-nic model=rocker", Michael S. Tsirkin, 2020/06/12
- [PULL v2 51/58] acpi: make build_madt() more generic., Michael S. Tsirkin, 2020/06/12
- [PULL v2 55/58] acpi: ged: rename event memory region, Michael S. Tsirkin, 2020/06/12
- [PULL v2 52/58] acpi: create acpi-common.c and move madt code, Michael S. Tsirkin, 2020/06/12
- [PULL v2 53/58] acpi: madt: skip pci override on pci-less systems., Michael S. Tsirkin, 2020/06/12