Skip to contentSkip to conversation

Enable virtio-fs on s390x

PATCH In review Marc Hartmayer 4 participants 12 messages Jul 30, 2020 – Aug 27, 2020

lore
Share & tools
lore
#0 lore raw

cover Enable virtio-fs on s390x

This patch series is about enabling virtio-fs on s390x. For that we need
+ some shim code (first patch), and we need
+ libvhost-user to deal with virtio endiannes for non-legacy virtio devices as mandated by the spec.

How to use?

For general instructions how to use virtio-fs (on x86) please have a look at https://virtio-fs.gitlab.io/howto-qemu.html. Most of the instructions can also be applied on s390x.

In short:

1. Install self-compiled QEMU with this patch series applied
2. Prepare host and guest kernel so they support virtio-fs

Start virtiofsd on the host

 $ virtiofsd -f --socket-path=/tmp/vhostqemu -o source=/tmp/shared

Now you can start QEMU in a separate shell on the host:

 $ qemu-system-s390x -machine type=s390-ccw-virtio,accel=kvm,memory-backend=mem \
   -object memory-backend-file,id=mem,size=2G,mem-path=/dev/shm/virtiofs,share=on,prealloc=on,prealloc-threads=1 \
   -chardev socket,id=char0,path=/tmp/vhostqemu -device vhost-user-fs-ccw,queue-size=1024,chardev=char0,tag=myfs \
   -drive if=virtio,file=disk.qcow2 \
   -m 2G -smp 2 -nographic

Log into the guest and mount it

 $ mount -t virtiofs myfs /mnt

Changelog:
RFC v2 -> v1:
- patch 1:
+ Added force_revision_1 = true (Conny)
+ I didn't add the r-b from Stefan Hajnoczi as I've added the changes suggested by Conny
- squashed patches 2 and 3:
+ removed assertion in performance critical code path (Stefan)
+ dropped all dead code (Stefan)
+ removed libvhost-access.h

RFC v1 -> RFC v2:
+ rebased
+ drop patch "libvhost-user: print invalid address on vu_panic" as it's not related to this series
+ drop patch "[RFC 4/4] HACK: Hard-code the libvhost-user.o-cflags for s390x"
+ patch "virtio: add vhost-user-fs-ccw device": replace qdev_set_parent_bus and object_property_set_bool by qdev_realize
+ patch "libvhost-user: handle endianness as mandated by the spec":
Drop support for legacy virtio devices
+ add patch to fence legacy virtio devices
*** BLURB HERE ***

Halil Pasic (1):
virtio: add vhost-user-fs-ccw device

Marc Hartmayer (1):
libvhost-user: handle endianness as mandated by the spec

 contrib/libvhost-user/libvhost-user.c | 77 +++++++++++++++------------
 hw/s390x/Makefile.objs                |  1 +
 hw/s390x/vhost-user-fs-ccw.c          | 75 ++++++++++++++++++++++++++
 3 files changed, 119 insertions(+), 34 deletions(-)
 create mode 100644 hw/s390x/vhost-user-fs-ccw.c
Signature · 1 line
2.25.4
#1 lore raw

PATCH 1/2 virtio: add vhost-user-fs-ccw device

From: Halil Pasic <[email protected]>

Wire up the CCW device for vhost-user-fs.

---
 hw/s390x/Makefile.objs       |  1 +
 hw/s390x/vhost-user-fs-ccw.c | 75 ++++++++++++++++++++++++++++++++++++
 2 files changed, 76 insertions(+)
 create mode 100644 hw/s390x/vhost-user-fs-ccw.c
hw/s390x/Makefile.objs +1 -0
index a46a1c7894e0..c4086ec3171e 100644
@@ -20,6 +20,7 @@ obj-$(CONFIG_VIRTIO_NET) += virtio-ccw-net.o
 obj-$(CONFIG_VIRTIO_BLK) += virtio-ccw-blk.o
 obj-$(call land,$(CONFIG_VIRTIO_9P),$(CONFIG_VIRTFS)) += virtio-ccw-9p.o
 obj-$(CONFIG_VHOST_VSOCK) += vhost-vsock-ccw.o
+obj-$(CONFIG_VHOST_USER_FS) += vhost-user-fs-ccw.o
 endif
 obj-y += css-bridge.o
 obj-y += ccw-device.o
hw/s390x/vhost-user-fs-ccw.c new file +75 -0

75 lines not on this page · raw message

Signature · 1 line
2.25.4
Reviewed-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] virtio: add vhost-user-fs-ccw device

Quoted 10 lines · On Thu, 30 Jul 2020 16:07:30 +0200 Marc Hartmayer <[email protected]>…

From: Halil Pasic <[email protected]>

Wire up the CCW device for vhost-user-fs.

---
 hw/s390x/Makefile.objs       |  1 +
 hw/s390x/vhost-user-fs-ccw.c | 75 ++++++++++++++++++++++++++++++++++++
 2 files changed, 76 insertions(+)
 create mode 100644 hw/s390x/vhost-user-fs-ccw.c
#3 lore raw

PATCH 2/2 libvhost-user: handle endianness as mandated by the spec

In reply to Marc Hartmayer

Since virtio existed even before it got standardized, the virtio standard defines the following types of virtio devices:

+ legacy device (pre-virtio 1.0)
+ non-legacy or VIRTIO 1.0 device
+ transitional device (which can act both as legacy and non-legacy)

Virtio 1.0 defines the fields of the virtqueues as little endian, while legacy uses guest's native endian [1]. Currently libvhost-user does not handle virtio endianness at all, i.e. it works only if the native endianness matches with whatever is actually needed. That means things break spectacularly on big-endian targets. Let us handle virtio endianness for non-legacy as required by the virtio specification [1]. The fencing of legacy virtio devices is done in vu_set_features_exec.

[1] https://docs.oasis-open.org/virtio/virtio/v1.1/cs01/virtio-v1.1-cs01.html#x1-210003

---
 contrib/libvhost-user/libvhost-user.c | 77 +++++++++++++++------------
 1 file changed, 43 insertions(+), 34 deletions(-)
contrib/libvhost-user/libvhost-user.c +43 -34
index 53f16bdf082c..e2238a0400c9 100644
@@ -42,6 +42,7 @@
 
 #include "qemu/atomic.h"
 #include "qemu/osdep.h"
+#include "qemu/bswap.h"
 #include "qemu/memfd.h"
 
 #include "libvhost-user.h"
@@ -539,6 +540,14 @@ vu_set_features_exec(VuDev *dev, VhostUserMsg *vmsg)
     DPRINT("u64: 0x%016"PRIx64"\n", vmsg->payload.u64);
 
     dev->features = vmsg->payload.u64;
+    if (!vu_has_feature(dev, VIRTIO_F_VERSION_1)) {
+        /*
+         * We only support devices conforming to VIRTIO 1.0 or
+         * later
+         */
+        vu_panic(dev, "virtio legacy devices aren't supported by libvhost-user");
+        return false;
+    }
 
     if (!(dev->features & VHOST_USER_F_PROTOCOL_FEATURES)) {
         vu_set_enable_all_rings(dev, true);
@@ -1074,7 +1083,7 @@ vu_set_vring_addr_exec(VuDev *dev, VhostUserMsg *vmsg)
         return false;
     }
 
-    vq->used_idx = vq->vring.used->idx;
+    vq->used_idx = lduw_le_p(&vq->vring.used->idx);
 
     if (vq->last_avail_idx != vq->used_idx) {
         bool resume = dev->iface->queue_is_processed_in_order &&
@@ -1191,7 +1200,7 @@ vu_check_queue_inflights(VuDev *dev, VuVirtq *vq)
         return 0;
     }
 
-    vq->used_idx = vq->vring.used->idx;
+    vq->used_idx = lduw_le_p(&vq->vring.used->idx);
     vq->resubmit_num = 0;
     vq->resubmit_list = NULL;
     vq->counter = 0;
@@ -2021,13 +2030,13 @@ vu_queue_started(const VuDev *dev, const VuVirtq *vq)
 static inline uint16_t
 vring_avail_flags(VuVirtq *vq)
 {
-    return vq->vring.avail->flags;
+    return lduw_le_p(&vq->vring.avail->flags);
 }
 
 static inline uint16_t
 vring_avail_idx(VuVirtq *vq)
 {
-    vq->shadow_avail_idx = vq->vring.avail->idx;
+    vq->shadow_avail_idx = lduw_le_p(&vq->vring.avail->idx);
 
     return vq->shadow_avail_idx;
 }
@@ -2035,7 +2044,7 @@ vring_avail_idx(VuVirtq *vq)
 static inline uint16_t
 vring_avail_ring(VuVirtq *vq, int i)
 {
-    return vq->vring.avail->ring[i];
+    return lduw_le_p(&vq->vring.avail->ring[i]);
 }
 
 static inline uint16_t
@@ -2123,12 +2132,12 @@ virtqueue_read_next_desc(VuDev *dev, struct vring_desc *desc,
                          int i, unsigned int max, unsigned int *next)
 {
     /* If this descriptor says it doesn't chain, we're done. */
-    if (!(desc[i].flags & VRING_DESC_F_NEXT)) {
+    if (!(lduw_le_p(&desc[i].flags) & VRING_DESC_F_NEXT)) {
         return VIRTQUEUE_READ_DESC_DONE;
     }
 
     /* Check they're not leading us off end of descriptors. */
-    *next = desc[i].next;
+    *next = lduw_le_p(&desc[i].next);
     /* Make sure compiler knows to grab that: we don't want it changing! */
     smp_wmb();
 
@@ -2171,8 +2180,8 @@ vu_queue_get_avail_bytes(VuDev *dev, VuVirtq *vq, unsigned int *in_bytes,
         }
         desc = vq->vring.desc;
 
-        if (desc[i].flags & VRING_DESC_F_INDIRECT) {
-            if (desc[i].len % sizeof(struct vring_desc)) {
+        if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_INDIRECT) {
+            if (ldl_le_p(&desc[i].len) % sizeof(struct vring_desc)) {
                 vu_panic(dev, "Invalid size for indirect buffer table");
                 goto err;
             }
@@ -2185,8 +2194,8 @@ vu_queue_get_avail_bytes(VuDev *dev, VuVirtq *vq, unsigned int *in_bytes,
 
             /* loop over the indirect descriptor table */
             indirect = 1;
-            desc_addr = desc[i].addr;
-            desc_len = desc[i].len;
+            desc_addr = ldq_le_p(&desc[i].addr);
+            desc_len = ldl_le_p(&desc[i].len);
             max = desc_len / sizeof(struct vring_desc);
             read_len = desc_len;
             desc = vu_gpa_to_va(dev, &read_len, desc_addr);
@@ -2213,10 +2222,10 @@ vu_queue_get_avail_bytes(VuDev *dev, VuVirtq *vq, unsigned int *in_bytes,
                 goto err;
             }
 
-            if (desc[i].flags & VRING_DESC_F_WRITE) {
-                in_total += desc[i].len;
+            if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_WRITE) {
+                in_total += ldl_le_p(&desc[i].len);
             } else {
-                out_total += desc[i].len;
+                out_total += ldl_le_p(&desc[i].len);
             }
             if (in_total >= max_in_bytes && out_total >= max_out_bytes) {
                 goto done;
@@ -2367,7 +2376,7 @@ vring_used_flags_set_bit(VuVirtq *vq, int mask)
 
     flags = (uint16_t *)((char*)vq->vring.used +
                          offsetof(struct vring_used, flags));
-    *flags |= mask;
+    stw_le_p(flags, lduw_le_p(flags) | mask);
 }
 
 static inline void
@@ -2377,7 +2386,7 @@ vring_used_flags_unset_bit(VuVirtq *vq, int mask)
 
     flags = (uint16_t *)((char*)vq->vring.used +
                          offsetof(struct vring_used, flags));
-    *flags &= ~mask;
+    stw_le_p(flags, lduw_le_p(flags) & ~mask);
 }
 
 static inline void
@@ -2387,7 +2396,7 @@ vring_set_avail_event(VuVirtq *vq, uint16_t val)
         return;
     }
 
-    *((uint16_t *) &vq->vring.used->ring[vq->vring.num]) = val;
+    stw_le_p(&vq->vring.used->ring[vq->vring.num], val);
 }
 
 void
@@ -2476,14 +2485,14 @@ vu_queue_map_desc(VuDev *dev, VuVirtq *vq, unsigned int idx, size_t sz)
     struct vring_desc desc_buf[VIRTQUEUE_MAX_SIZE];
     int rc;
 
-    if (desc[i].flags & VRING_DESC_F_INDIRECT) {
-        if (desc[i].len % sizeof(struct vring_desc)) {
+    if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_INDIRECT) {
+        if (ldl_le_p(&desc[i].len) % sizeof(struct vring_desc)) {
             vu_panic(dev, "Invalid size for indirect buffer table");
         }
 
         /* loop over the indirect descriptor table */
-        desc_addr = desc[i].addr;
-        desc_len = desc[i].len;
+        desc_addr = ldq_le_p(&desc[i].addr);
+        desc_len = ldl_le_p(&desc[i].len);
         max = desc_len / sizeof(struct vring_desc);
         read_len = desc_len;
         desc = vu_gpa_to_va(dev, &read_len, desc_addr);
@@ -2505,10 +2514,10 @@ vu_queue_map_desc(VuDev *dev, VuVirtq *vq, unsigned int idx, size_t sz)
 
     /* Collect all the descriptors */
     do {
-        if (desc[i].flags & VRING_DESC_F_WRITE) {
+        if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_WRITE) {
             virtqueue_map_desc(dev, &in_num, iov + out_num,
                                VIRTQUEUE_MAX_SIZE - out_num, true,
-                               desc[i].addr, desc[i].len);
+                               ldq_le_p(&desc[i].addr), ldl_le_p(&desc[i].len));
         } else {
             if (in_num) {
                 vu_panic(dev, "Incorrect order for descriptors");
@@ -2516,7 +2525,7 @@ vu_queue_map_desc(VuDev *dev, VuVirtq *vq, unsigned int idx, size_t sz)
             }
             virtqueue_map_desc(dev, &out_num, iov,
                                VIRTQUEUE_MAX_SIZE, false,
-                               desc[i].addr, desc[i].len);
+                               ldq_le_p(&desc[i].addr), ldl_le_p(&desc[i].len));
         }
 
         /* If we've got too many, that implies a descriptor loop. */
@@ -2712,14 +2721,14 @@ vu_log_queue_fill(VuDev *dev, VuVirtq *vq,
     max = vq->vring.num;
     i = elem->index;
 
-    if (desc[i].flags & VRING_DESC_F_INDIRECT) {
-        if (desc[i].len % sizeof(struct vring_desc)) {
+    if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_INDIRECT) {
+        if (ldl_le_p(&desc[i].len) % sizeof(struct vring_desc)) {
             vu_panic(dev, "Invalid size for indirect buffer table");
         }
 
         /* loop over the indirect descriptor table */
-        desc_addr = desc[i].addr;
-        desc_len = desc[i].len;
+        desc_addr = ldq_le_p(&desc[i].addr);
+        desc_len = ldl_le_p(&desc[i].len);
         max = desc_len / sizeof(struct vring_desc);
         read_len = desc_len;
         desc = vu_gpa_to_va(dev, &read_len, desc_addr);
@@ -2745,9 +2754,9 @@ vu_log_queue_fill(VuDev *dev, VuVirtq *vq,
             return;
         }
 
-        if (desc[i].flags & VRING_DESC_F_WRITE) {
-            min = MIN(desc[i].len, len);
-            vu_log_write(dev, desc[i].addr, min);
+        if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_WRITE) {
+            min = MIN(ldl_le_p(&desc[i].len), len);
+            vu_log_write(dev, ldq_le_p(&desc[i].addr), min);
             len -= min;
         }
 
@@ -2772,15 +2781,15 @@ vu_queue_fill(VuDev *dev, VuVirtq *vq,
 
     idx = (idx + vq->used_idx) % vq->vring.num;
 
-    uelem.id = elem->index;
-    uelem.len = len;
+    stl_le_p(&uelem.id, elem->index);
+    stl_le_p(&uelem.len, len);
     vring_used_write(dev, vq, &uelem, idx);
 }
 
 static inline
 void vring_used_idx_set(VuDev *dev, VuVirtq *vq, uint16_t val)
 {
-    vq->vring.used->idx = val;
+    stw_le_p(&vq->vring.used->idx, val);
     vu_log_write(dev,
                  vq->vring.log_guest_addr + offsetof(struct vring_used, idx),
                  sizeof(vq->vring.used->idx));
Signature · 1 line
2.25.4
Reviewed-by on v1 2/2
#4 lore raw

Re: [PATCH 2/2] libvhost-user: handle endianness as mandated by the spec

Quoted 16 lines · Marc Hartmayer

Since virtio existed even before it got standardized, the virtio standard defines the following types of virtio devices:

+ legacy device (pre-virtio 1.0)
+ non-legacy or VIRTIO 1.0 device
+ transitional device (which can act both as legacy and non-legacy)

Virtio 1.0 defines the fields of the virtqueues as little endian, while legacy uses guest's native endian [1]. Currently libvhost-user does not handle virtio endianness at all, i.e. it works only if the native endianness matches with whatever is actually needed. That means things break spectacularly on big-endian targets. Let us handle virtio endianness for non-legacy as required by the virtio specification [1]. The fencing of legacy virtio devices is done in vu_set_features_exec.

Quoted 3 lines

[1] https://docs.oasis-open.org/virtio/virtio/v1.1/cs01/virtio-v1.1-cs01.html#x1-210003

Quoted 246 lines
---
 contrib/libvhost-user/libvhost-user.c | 77 +++++++++++++++------------
 1 file changed, 43 insertions(+), 34 deletions(-)
contrib/libvhost-user/libvhost-user.c +43 -34
index 53f16bdf082c..e2238a0400c9 100644
@@ -42,6 +42,7 @@
 
 #include "qemu/atomic.h"
 #include "qemu/osdep.h"
+#include "qemu/bswap.h"
 #include "qemu/memfd.h"
 
 #include "libvhost-user.h"
@@ -539,6 +540,14 @@ vu_set_features_exec(VuDev *dev, VhostUserMsg *vmsg)
     DPRINT("u64: 0x%016"PRIx64"\n", vmsg->payload.u64);
 
     dev->features = vmsg->payload.u64;
+    if (!vu_has_feature(dev, VIRTIO_F_VERSION_1)) {
+        /*
+         * We only support devices conforming to VIRTIO 1.0 or
+         * later
+         */
+        vu_panic(dev, "virtio legacy devices aren't supported by libvhost-user");
+        return false;
+    }
 
     if (!(dev->features & VHOST_USER_F_PROTOCOL_FEATURES)) {
         vu_set_enable_all_rings(dev, true);
@@ -1074,7 +1083,7 @@ vu_set_vring_addr_exec(VuDev *dev, VhostUserMsg *vmsg)
         return false;
     }
 
-    vq->used_idx = vq->vring.used->idx;
+    vq->used_idx = lduw_le_p(&vq->vring.used->idx);
 
     if (vq->last_avail_idx != vq->used_idx) {
         bool resume = dev->iface->queue_is_processed_in_order &&
@@ -1191,7 +1200,7 @@ vu_check_queue_inflights(VuDev *dev, VuVirtq *vq)
         return 0;
     }
 
-    vq->used_idx = vq->vring.used->idx;
+    vq->used_idx = lduw_le_p(&vq->vring.used->idx);
     vq->resubmit_num = 0;
     vq->resubmit_list = NULL;
     vq->counter = 0;
@@ -2021,13 +2030,13 @@ vu_queue_started(const VuDev *dev, const VuVirtq *vq)
 static inline uint16_t
 vring_avail_flags(VuVirtq *vq)
 {
-    return vq->vring.avail->flags;
+    return lduw_le_p(&vq->vring.avail->flags);
 }
 
 static inline uint16_t
 vring_avail_idx(VuVirtq *vq)
 {
-    vq->shadow_avail_idx = vq->vring.avail->idx;
+    vq->shadow_avail_idx = lduw_le_p(&vq->vring.avail->idx);
 
     return vq->shadow_avail_idx;
 }
@@ -2035,7 +2044,7 @@ vring_avail_idx(VuVirtq *vq)
 static inline uint16_t
 vring_avail_ring(VuVirtq *vq, int i)
 {
-    return vq->vring.avail->ring[i];
+    return lduw_le_p(&vq->vring.avail->ring[i]);
 }
 
 static inline uint16_t
@@ -2123,12 +2132,12 @@ virtqueue_read_next_desc(VuDev *dev, struct vring_desc *desc,
                          int i, unsigned int max, unsigned int *next)
 {
     /* If this descriptor says it doesn't chain, we're done. */
-    if (!(desc[i].flags & VRING_DESC_F_NEXT)) {
+    if (!(lduw_le_p(&desc[i].flags) & VRING_DESC_F_NEXT)) {
         return VIRTQUEUE_READ_DESC_DONE;
     }
 
     /* Check they're not leading us off end of descriptors. */
-    *next = desc[i].next;
+    *next = lduw_le_p(&desc[i].next);
     /* Make sure compiler knows to grab that: we don't want it changing! */
     smp_wmb();
 
@@ -2171,8 +2180,8 @@ vu_queue_get_avail_bytes(VuDev *dev, VuVirtq *vq, unsigned int *in_bytes,
         }
         desc = vq->vring.desc;
 
-        if (desc[i].flags & VRING_DESC_F_INDIRECT) {
-            if (desc[i].len % sizeof(struct vring_desc)) {
+        if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_INDIRECT) {
+            if (ldl_le_p(&desc[i].len) % sizeof(struct vring_desc)) {
                 vu_panic(dev, "Invalid size for indirect buffer table");
                 goto err;
             }
@@ -2185,8 +2194,8 @@ vu_queue_get_avail_bytes(VuDev *dev, VuVirtq *vq, unsigned int *in_bytes,
 
             /* loop over the indirect descriptor table */
             indirect = 1;
-            desc_addr = desc[i].addr;
-            desc_len = desc[i].len;
+            desc_addr = ldq_le_p(&desc[i].addr);
+            desc_len = ldl_le_p(&desc[i].len);
             max = desc_len / sizeof(struct vring_desc);
             read_len = desc_len;
             desc = vu_gpa_to_va(dev, &read_len, desc_addr);
@@ -2213,10 +2222,10 @@ vu_queue_get_avail_bytes(VuDev *dev, VuVirtq *vq, unsigned int *in_bytes,
                 goto err;
             }
 
-            if (desc[i].flags & VRING_DESC_F_WRITE) {
-                in_total += desc[i].len;
+            if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_WRITE) {
+                in_total += ldl_le_p(&desc[i].len);
             } else {
-                out_total += desc[i].len;
+                out_total += ldl_le_p(&desc[i].len);
             }
             if (in_total >= max_in_bytes && out_total >= max_out_bytes) {
                 goto done;
@@ -2367,7 +2376,7 @@ vring_used_flags_set_bit(VuVirtq *vq, int mask)
 
     flags = (uint16_t *)((char*)vq->vring.used +
                          offsetof(struct vring_used, flags));
-    *flags |= mask;
+    stw_le_p(flags, lduw_le_p(flags) | mask);
 }
 
 static inline void
@@ -2377,7 +2386,7 @@ vring_used_flags_unset_bit(VuVirtq *vq, int mask)
 
     flags = (uint16_t *)((char*)vq->vring.used +
                          offsetof(struct vring_used, flags));
-    *flags &= ~mask;
+    stw_le_p(flags, lduw_le_p(flags) & ~mask);
 }
 
 static inline void
@@ -2387,7 +2396,7 @@ vring_set_avail_event(VuVirtq *vq, uint16_t val)
         return;
     }
 
-    *((uint16_t *) &vq->vring.used->ring[vq->vring.num]) = val;
+    stw_le_p(&vq->vring.used->ring[vq->vring.num], val);
 }
 
 void
@@ -2476,14 +2485,14 @@ vu_queue_map_desc(VuDev *dev, VuVirtq *vq, unsigned int idx, size_t sz)
     struct vring_desc desc_buf[VIRTQUEUE_MAX_SIZE];
     int rc;
 
-    if (desc[i].flags & VRING_DESC_F_INDIRECT) {
-        if (desc[i].len % sizeof(struct vring_desc)) {
+    if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_INDIRECT) {
+        if (ldl_le_p(&desc[i].len) % sizeof(struct vring_desc)) {
             vu_panic(dev, "Invalid size for indirect buffer table");
         }
 
         /* loop over the indirect descriptor table */
-        desc_addr = desc[i].addr;
-        desc_len = desc[i].len;
+        desc_addr = ldq_le_p(&desc[i].addr);
+        desc_len = ldl_le_p(&desc[i].len);
         max = desc_len / sizeof(struct vring_desc);
         read_len = desc_len;
         desc = vu_gpa_to_va(dev, &read_len, desc_addr);
@@ -2505,10 +2514,10 @@ vu_queue_map_desc(VuDev *dev, VuVirtq *vq, unsigned int idx, size_t sz)
 
     /* Collect all the descriptors */
     do {
-        if (desc[i].flags & VRING_DESC_F_WRITE) {
+        if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_WRITE) {
             virtqueue_map_desc(dev, &in_num, iov + out_num,
                                VIRTQUEUE_MAX_SIZE - out_num, true,
-                               desc[i].addr, desc[i].len);
+                               ldq_le_p(&desc[i].addr), ldl_le_p(&desc[i].len));
         } else {
             if (in_num) {
                 vu_panic(dev, "Incorrect order for descriptors");
@@ -2516,7 +2525,7 @@ vu_queue_map_desc(VuDev *dev, VuVirtq *vq, unsigned int idx, size_t sz)
             }
             virtqueue_map_desc(dev, &out_num, iov,
                                VIRTQUEUE_MAX_SIZE, false,
-                               desc[i].addr, desc[i].len);
+                               ldq_le_p(&desc[i].addr), ldl_le_p(&desc[i].len));
         }
 
         /* If we've got too many, that implies a descriptor loop. */
@@ -2712,14 +2721,14 @@ vu_log_queue_fill(VuDev *dev, VuVirtq *vq,
     max = vq->vring.num;
     i = elem->index;
 
-    if (desc[i].flags & VRING_DESC_F_INDIRECT) {
-        if (desc[i].len % sizeof(struct vring_desc)) {
+    if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_INDIRECT) {
+        if (ldl_le_p(&desc[i].len) % sizeof(struct vring_desc)) {
             vu_panic(dev, "Invalid size for indirect buffer table");
         }
 
         /* loop over the indirect descriptor table */
-        desc_addr = desc[i].addr;
-        desc_len = desc[i].len;
+        desc_addr = ldq_le_p(&desc[i].addr);
+        desc_len = ldl_le_p(&desc[i].len);
         max = desc_len / sizeof(struct vring_desc);
         read_len = desc_len;
         desc = vu_gpa_to_va(dev, &read_len, desc_addr);
@@ -2745,9 +2754,9 @@ vu_log_queue_fill(VuDev *dev, VuVirtq *vq,
             return;
         }
 
-        if (desc[i].flags & VRING_DESC_F_WRITE) {
-            min = MIN(desc[i].len, len);
-            vu_log_write(dev, desc[i].addr, min);
+        if (lduw_le_p(&desc[i].flags) & VRING_DESC_F_WRITE) {
+            min = MIN(ldl_le_p(&desc[i].len), len);
+            vu_log_write(dev, ldq_le_p(&desc[i].addr), min);
             len -= min;
         }
 
@@ -2772,15 +2781,15 @@ vu_queue_fill(VuDev *dev, VuVirtq *vq,
 
     idx = (idx + vq->used_idx) % vq->vring.num;
 
-    uelem.id = elem->index;
-    uelem.len = len;
+    stl_le_p(&uelem.id, elem->index);
+    stl_le_p(&uelem.len, len);
     vring_used_write(dev, vq, &uelem, idx);
 }
 
 static inline
 void vring_used_idx_set(VuDev *dev, VuVirtq *vq, uint16_t val)
 {
-    vq->vring.used->idx = val;
+    stw_le_p(&vq->vring.used->idx, val);
     vu_log_write(dev,
                  vq->vring.log_guest_addr + offsetof(struct vring_used, idx),
                  sizeof(vq->vring.used->idx));
Signature · 1 line
2.25.4
#5 lore raw

Re: [PATCH 2/2] libvhost-user: handle endianness as mandated by the spec

Quoted 17 lines · Michael S. Tsirkin
Quoted 16 lines · Marc Hartmayer

Since virtio existed even before it got standardized, the virtio standard defines the following types of virtio devices:

+ legacy device (pre-virtio 1.0)
+ non-legacy or VIRTIO 1.0 device
+ transitional device (which can act both as legacy and non-legacy)

Virtio 1.0 defines the fields of the virtqueues as little endian, while legacy uses guest's native endian [1]. Currently libvhost-user does not handle virtio endianness at all, i.e. it works only if the native endianness matches with whatever is actually needed. That means things break spectacularly on big-endian targets. Let us handle virtio endianness for non-legacy as required by the virtio specification [1]. The fencing of legacy virtio devices is done in vu_set_features_exec.

Quoted 6 lines
Quoted 3 lines

[1] https://docs.oasis-open.org/virtio/virtio/v1.1/cs01/virtio-v1.1-cs01.html#x1-210003

Thanks.

[…snip]

Signature · 8 lines
Kind regards / Beste Grüße
   Marc Hartmayer

IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Gregor Pillen 
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294
#6 lore raw

Re: [PATCH 2/2] libvhost-user: handle endianness as mandated by the spec

In reply to Marc Hartmayer
Quoted 7 lines · On Thu, 30 Jul 2020 16:07:31 +0200 Marc Hartmayer <[email protected]>…

Since virtio existed even before it got standardized, the virtio standard defines the following types of virtio devices:

+ legacy device (pre-virtio 1.0)
+ non-legacy or VIRTIO 1.0 device
+ transitional device (which can act both as legacy and non-legacy)

Quoted 7 lines

Virtio 1.0 defines the fields of the virtqueues as little endian, while legacy uses guest's native endian [1]. Currently libvhost-user does not handle virtio endianness at all, i.e. it works only if the native endianness matches with whatever is actually needed. That means things break spectacularly on big-endian targets. Let us handle virtio endianness for non-legacy as required by the virtio specification [1].

Maybe add

"and fence legacy virtio, as there is no safe way to figure out the needed endianness conversions for all cases."

Quoted 2 lines

The fencing of legacy virtio devices is done in
vu_set_features_exec.

Not that I disagree with fencing legacy virtio, but looking at some vhost-user* drivers, I'm not sure everything will work as desired for those (I might be missing something, though.)

- vhost-user-blk lists VERSION_1 in the supported features, but vhost-user-scsi doesn't... is there some inheritance going on that I'm missing?
- vhost-user-gpu-pci inherits from virtio-gpu-pci, so I guess it's fine
- vhost-user-input should also always have been virtio-1

So, has anybody been using vhost-user-scsi and can confirm that it still works, or at least can be made to work?

Quoted 7 lines

[1] https://docs.oasis-open.org/virtio/virtio/v1.1/cs01/virtio-v1.1-cs01.html#x1-210003

---
 contrib/libvhost-user/libvhost-user.c | 77 +++++++++++++++------------
 1 file changed, 43 insertions(+), 34 deletions(-)

The code change per se LGTM.

#7 lore raw

Re: [PATCH 2/2] libvhost-user: handle endianness as mandated by the spec

Quoted 18 lines · Cornelia Huck
Quoted 7 lines · On Thu, 30 Jul 2020 16:07:31 +0200 Marc Hartmayer <[email protected]>…

Since virtio existed even before it got standardized, the virtio standard defines the following types of virtio devices:

+ legacy device (pre-virtio 1.0)
+ non-legacy or VIRTIO 1.0 device
+ transitional device (which can act both as legacy and non-legacy)

Quoted 7 lines

Virtio 1.0 defines the fields of the virtqueues as little endian, while legacy uses guest's native endian [1]. Currently libvhost-user does not handle virtio endianness at all, i.e. it works only if the native endianness matches with whatever is actually needed. That means things break spectacularly on big-endian targets. Let us handle virtio endianness for non-legacy as required by the virtio specification [1].

Quoted 4 lines

Maybe add

"and fence legacy virtio, as there is no safe way to figure out the needed endianness conversions for all cases."

Okay.

Quoted 8 lines
Quoted 3 lines

The fencing of legacy virtio devices is done in
vu_set_features_exec.

Not that I disagree with fencing legacy virtio, but looking at some vhost-user* drivers, I'm not sure everything will work as desired for those (I might be missing something, though.)

Quoted 8 lines

- vhost-user-blk lists VERSION_1 in the supported features, but vhost-user-scsi doesn't... is there some inheritance going on that I'm missing?
- vhost-user-gpu-pci inherits from virtio-gpu-pci, so I guess it's fine
- vhost-user-input should also always have been virtio-1

So, has anybody been using vhost-user-scsi and can confirm that it still works, or at least can be made to work?

Unfortunately, I don’t have the required hardware :/ Can please anybody verify this?

Quoted 10 lines
Quoted 8 lines

[1] https://docs.oasis-open.org/virtio/virtio/v1.1/cs01/virtio-v1.1-cs01.html#x1-210003

---
 contrib/libvhost-user/libvhost-user.c | 77 +++++++++++++++------------
 1 file changed, 43 insertions(+), 34 deletions(-)

The code change per se LGTM.

Thanks for the feedback!

Quoted 1 line
Signature · 8 lines
Kind regards / Beste Grüße
   Marc Hartmayer

IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Gregor Pillen 
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294
#8 lore raw
In reply to Marc Hartmayer
Quoted 28 lines · Marc Hartmayer

This patch series is about enabling virtio-fs on s390x. For that we need
+ some shim code (first patch), and we need
+ libvhost-user to deal with virtio endiannes for non-legacy virtio devices as mandated by the spec.

How to use?

For general instructions how to use virtio-fs (on x86) please have a look at https://virtio-fs.gitlab.io/howto-qemu.html. Most of the instructions can also be applied on s390x.

In short:

1. Install self-compiled QEMU with this patch series applied
2. Prepare host and guest kernel so they support virtio-fs

Start virtiofsd on the host

 $ virtiofsd -f --socket-path=/tmp/vhostqemu -o source=/tmp/shared

Now you can start QEMU in a separate shell on the host:

 $ qemu-system-s390x -machine type=s390-ccw-virtio,accel=kvm,memory-backend=mem \
   -object memory-backend-file,id=mem,size=2G,mem-path=/dev/shm/virtiofs,share=on,prealloc=on,prealloc-threads=1 \
   -chardev socket,id=char0,path=/tmp/vhostqemu -device vhost-user-fs-ccw,queue-size=1024,chardev=char0,tag=myfs \
   -drive if=virtio,file=disk.qcow2 \
   -m 2G -smp 2 -nographic
Quoted 3 lines

Log into the guest and mount it

 $ mount -t virtiofs myfs /mnt

Who's merging this? My tree?

Quoted 35 lines

Changelog:
RFC v2 -> v1:
- patch 1:
+ Added force_revision_1 = true (Conny)
+ I didn't add the r-b from Stefan Hajnoczi as I've added the changes suggested by Conny
- squashed patches 2 and 3:
+ removed assertion in performance critical code path (Stefan)
+ dropped all dead code (Stefan)
+ removed libvhost-access.h

RFC v1 -> RFC v2:
+ rebased
+ drop patch "libvhost-user: print invalid address on vu_panic" as it's not related to this series
+ drop patch "[RFC 4/4] HACK: Hard-code the libvhost-user.o-cflags for s390x"
+ patch "virtio: add vhost-user-fs-ccw device": replace qdev_set_parent_bus and object_property_set_bool by qdev_realize
+ patch "libvhost-user: handle endianness as mandated by the spec":
Drop support for legacy virtio devices
+ add patch to fence legacy virtio devices
*** BLURB HERE ***

Halil Pasic (1):
virtio: add vhost-user-fs-ccw device

Marc Hartmayer (1):
libvhost-user: handle endianness as mandated by the spec

 contrib/libvhost-user/libvhost-user.c | 77 +++++++++++++++------------
 hw/s390x/Makefile.objs                |  1 +
 hw/s390x/vhost-user-fs-ccw.c          | 75 ++++++++++++++++++++++++++
 3 files changed, 119 insertions(+), 34 deletions(-)
 create mode 100644 hw/s390x/vhost-user-fs-ccw.c
Signature · 1 line
2.25.4
#9 lore raw
Quoted 29 lines · * Michael S. Tsirkin ([email protected]) wrote:
Quoted 28 lines · Marc Hartmayer

This patch series is about enabling virtio-fs on s390x. For that we need
+ some shim code (first patch), and we need
+ libvhost-user to deal with virtio endiannes for non-legacy virtio devices as mandated by the spec.

How to use?

For general instructions how to use virtio-fs (on x86) please have a look at https://virtio-fs.gitlab.io/howto-qemu.html. Most of the instructions can also be applied on s390x.

In short:

1. Install self-compiled QEMU with this patch series applied
2. Prepare host and guest kernel so they support virtio-fs

Start virtiofsd on the host

 $ virtiofsd -f --socket-path=/tmp/vhostqemu -o source=/tmp/shared

Now you can start QEMU in a separate shell on the host:

 $ qemu-system-s390x -machine type=s390-ccw-virtio,accel=kvm,memory-backend=mem \
   -object memory-backend-file,id=mem,size=2G,mem-path=/dev/shm/virtiofs,share=on,prealloc=on,prealloc-threads=1 \
   -chardev socket,id=char0,path=/tmp/vhostqemu -device vhost-user-fs-ccw,queue-size=1024,chardev=char0,tag=myfs \
   -drive if=virtio,file=disk.qcow2 \
   -m 2G -smp 2 -nographic
Quoted 5 lines
Quoted 3 lines

Log into the guest and mount it

 $ mount -t virtiofs myfs /mnt

Who's merging this? My tree?

I think so; it seems either generic virtio or s390 more than actually virtiofs specific in most of it.

Dave

Quoted 27 lines
Quoted 27 lines

Changelog:
RFC v2 -> v1:
- patch 1:
+ Added force_revision_1 = true (Conny)
+ I didn't add the r-b from Stefan Hajnoczi as I've added the changes suggested by Conny
- squashed patches 2 and 3:
+ removed assertion in performance critical code path (Stefan)
+ dropped all dead code (Stefan)
+ removed libvhost-access.h

RFC v1 -> RFC v2:
+ rebased
+ drop patch "libvhost-user: print invalid address on vu_panic" as it's not related to this series
+ drop patch "[RFC 4/4] HACK: Hard-code the libvhost-user.o-cflags for s390x"
+ patch "virtio: add vhost-user-fs-ccw device": replace qdev_set_parent_bus and object_property_set_bool by qdev_realize
+ patch "libvhost-user: handle endianness as mandated by the spec":
Drop support for legacy virtio devices
+ add patch to fence legacy virtio devices
*** BLURB HERE ***

Halil Pasic (1):
virtio: add vhost-user-fs-ccw device

Marc Hartmayer (1):
libvhost-user: handle endianness as mandated by the spec

Quoted 9 lines
Quoted 8 lines
 contrib/libvhost-user/libvhost-user.c | 77 +++++++++++++++------------
 hw/s390x/Makefile.objs                |  1 +
 hw/s390x/vhost-user-fs-ccw.c          | 75 ++++++++++++++++++++++++++
 3 files changed, 119 insertions(+), 34 deletions(-)
 create mode 100644 hw/s390x/vhost-user-fs-ccw.c
Signature · 1 line
2.25.4

--
Dr. David Alan Gilbert / [email protected] / Manchester, UK

#10 lore raw
In reply to Marc Hartmayer
Quoted 4 lines · Marc Hartmayer

This patch series is about enabling virtio-fs on s390x. For that we need
+ some shim code (first patch), and we need
+ libvhost-user to deal with virtio endiannes for non-legacy virtio devices as mandated by the spec.

Please rebase, address Cornelia's minor comments and repost. Thanks!

Quoted 62 lines

How to use?

For general instructions how to use virtio-fs (on x86) please have a look at https://virtio-fs.gitlab.io/howto-qemu.html. Most of the instructions can also be applied on s390x.

In short:

1. Install self-compiled QEMU with this patch series applied
2. Prepare host and guest kernel so they support virtio-fs

Start virtiofsd on the host

 $ virtiofsd -f --socket-path=/tmp/vhostqemu -o source=/tmp/shared

Now you can start QEMU in a separate shell on the host:

 $ qemu-system-s390x -machine type=s390-ccw-virtio,accel=kvm,memory-backend=mem \
   -object memory-backend-file,id=mem,size=2G,mem-path=/dev/shm/virtiofs,share=on,prealloc=on,prealloc-threads=1 \
   -chardev socket,id=char0,path=/tmp/vhostqemu -device vhost-user-fs-ccw,queue-size=1024,chardev=char0,tag=myfs \
   -drive if=virtio,file=disk.qcow2 \
   -m 2G -smp 2 -nographic

Log into the guest and mount it

 $ mount -t virtiofs myfs /mnt

Changelog:
RFC v2 -> v1:
- patch 1:
+ Added force_revision_1 = true (Conny)
+ I didn't add the r-b from Stefan Hajnoczi as I've added the changes suggested by Conny
- squashed patches 2 and 3:
+ removed assertion in performance critical code path (Stefan)
+ dropped all dead code (Stefan)
+ removed libvhost-access.h

RFC v1 -> RFC v2:
+ rebased
+ drop patch "libvhost-user: print invalid address on vu_panic" as it's not related to this series
+ drop patch "[RFC 4/4] HACK: Hard-code the libvhost-user.o-cflags for s390x"
+ patch "virtio: add vhost-user-fs-ccw device": replace qdev_set_parent_bus and object_property_set_bool by qdev_realize
+ patch "libvhost-user: handle endianness as mandated by the spec":
Drop support for legacy virtio devices
+ add patch to fence legacy virtio devices
*** BLURB HERE ***

Halil Pasic (1):
virtio: add vhost-user-fs-ccw device

Marc Hartmayer (1):
libvhost-user: handle endianness as mandated by the spec

 contrib/libvhost-user/libvhost-user.c | 77 +++++++++++++++------------
 hw/s390x/Makefile.objs                |  1 +
 hw/s390x/vhost-user-fs-ccw.c          | 75 ++++++++++++++++++++++++++
 3 files changed, 119 insertions(+), 34 deletions(-)
 create mode 100644 hw/s390x/vhost-user-fs-ccw.c
Signature · 1 line
2.25.4
#11 lore raw
Quoted 8 lines · On Thu, 27 Aug 2020 08:19:44 -0400 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 4 lines · Marc Hartmayer

This patch series is about enabling virtio-fs on s390x. For that we need
+ some shim code (first patch), and we need
+ libvhost-user to deal with virtio endiannes for non-legacy virtio devices as mandated by the spec.

Please rebase, address Cornelia's minor comments and repost.

I think we're still waiting for someone to confirm the status of vhost-user-scsi?