From: Ahmad Fatoum <ahmad@a3f.at>
To: barebox@lists.infradead.org
Cc: Ahmad Fatoum <ahmad@a3f.at>
Subject: [PATCH 5/9] virtio: ring: fix erroneous behavior around caches and MMU
Date: Sun, 28 Feb 2021 20:08:32 +0100 [thread overview]
Message-ID: <20210228190836.1451663-5-ahmad@a3f.at> (raw)
In-Reply-To: <20210228190836.1451663-1-ahmad@a3f.at>
The U-Boot code assumes 1:1 phys2virt mapping and either disabled
caches or a lot of luck. Use the DMA API to get appropriate addresses
for DMA and use coherent/streaming DMA mappings where appropriate.
This is required for proper operation on MIPS.
Signed-off-by: Ahmad Fatoum <ahmad@a3f.at>
---
drivers/virtio/virtio_ring.c | 98 +++++++++++++++++++++++++++++++-----
include/linux/virtio_ring.h | 2 +
2 files changed, 87 insertions(+), 13 deletions(-)
diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
index cac3362e7251..68180fe37da0 100644
--- a/drivers/virtio/virtio_ring.c
+++ b/drivers/virtio/virtio_ring.c
@@ -22,12 +22,45 @@
#define vq_info(vq, fmt, ...) \
dev_info(&vq->vdev->dev, fmt, ##__VA_ARGS__)
+static inline struct device_d *vring_dma_dev(const struct virtqueue *vq)
+{
+ return vq->vdev->dev.parent;
+}
+
+/* Map one sg entry. */
+static dma_addr_t vring_map_one_sg(struct virtqueue *vq,
+ struct virtio_sg *sg,
+ enum dma_data_direction direction)
+{
+ return dma_map_single(vring_dma_dev(vq), sg->addr, sg->length, direction);
+}
+
+static int vring_mapping_error(struct virtqueue *vq,
+ dma_addr_t addr)
+{
+ return dma_mapping_error(vring_dma_dev(vq), addr);
+}
+
+static void vring_unmap_one(struct virtqueue *vq,
+ struct vring_desc *desc)
+{
+ u16 flags;
+
+ flags = virtio16_to_cpu(vq->vdev, desc->flags);
+
+ dma_unmap_single(vring_dma_dev(vq),
+ virtio64_to_cpu(vq->vdev, desc->addr),
+ virtio32_to_cpu(vq->vdev, desc->len),
+ (flags & VRING_DESC_F_WRITE) ?
+ DMA_FROM_DEVICE : DMA_TO_DEVICE);
+}
+
int virtqueue_add(struct virtqueue *vq, struct virtio_sg *sgs[],
unsigned int out_sgs, unsigned int in_sgs)
{
struct vring_desc *desc;
unsigned int total_sg = out_sgs + in_sgs;
- unsigned int i, n, avail, descs_used, uninitialized_var(prev);
+ unsigned int i, err_idx, n, avail, descs_used, uninitialized_var(prev);
int head;
WARN_ON(total_sg == 0);
@@ -53,9 +86,13 @@ int virtqueue_add(struct virtqueue *vq, struct virtio_sg *sgs[],
for (n = 0; n < out_sgs; n++) {
struct virtio_sg *sg = sgs[n];
+ dma_addr_t addr = vring_map_one_sg(vq, sg, DMA_TO_DEVICE);
+ if (vring_mapping_error(vq, addr))
+ goto unmap_release;
+
desc[i].flags = cpu_to_virtio16(vq->vdev, VRING_DESC_F_NEXT);
- desc[i].addr = cpu_to_virtio64(vq->vdev, (u64)(size_t)sg->addr);
+ desc[i].addr = cpu_to_virtio64(vq->vdev, addr);
desc[i].len = cpu_to_virtio32(vq->vdev, sg->length);
prev = i;
@@ -63,11 +100,13 @@ int virtqueue_add(struct virtqueue *vq, struct virtio_sg *sgs[],
}
for (; n < (out_sgs + in_sgs); n++) {
struct virtio_sg *sg = sgs[n];
+ dma_addr_t addr = vring_map_one_sg(vq, sg, DMA_FROM_DEVICE);
+ if (vring_mapping_error(vq, addr))
+ goto unmap_release;
desc[i].flags = cpu_to_virtio16(vq->vdev, VRING_DESC_F_NEXT |
VRING_DESC_F_WRITE);
- desc[i].addr = cpu_to_virtio64(vq->vdev,
- (u64)(uintptr_t)sg->addr);
+ desc[i].addr = cpu_to_virtio64(vq->vdev, addr);
desc[i].len = cpu_to_virtio32(vq->vdev, sg->length);
prev = i;
@@ -106,6 +145,19 @@ int virtqueue_add(struct virtqueue *vq, struct virtio_sg *sgs[],
virtqueue_kick(vq);
return 0;
+
+unmap_release:
+ err_idx = i;
+
+ for (n = 0; n < total_sg; n++) {
+ if (i == err_idx)
+ break;
+ vring_unmap_one(vq, &desc[i]);
+ i = virtio16_to_cpu(vq->vdev, desc[i].next);
+ }
+
+ return -ENOMEM;
+
}
static bool virtqueue_kick_prepare(struct virtqueue *vq)
@@ -149,10 +201,12 @@ static void detach_buf(struct virtqueue *vq, unsigned int head)
i = head;
while (vq->vring.desc[i].flags & nextflag) {
+ vring_unmap_one(vq, &vq->vring.desc[i]);
i = virtio16_to_cpu(vq->vdev, vq->vring.desc[i].next);
vq->num_free++;
}
+ vring_unmap_one(vq, &vq->vring.desc[i]);
vq->vring.desc[i].next = cpu_to_virtio16(vq->vdev, vq->free_head);
vq->free_head = head;
@@ -225,6 +279,8 @@ static struct virtqueue *__vring_new_virtqueue(unsigned int index,
vq->avail_flags_shadow = 0;
vq->avail_idx_shadow = 0;
vq->num_added = 0;
+ vq->queue_dma_addr = 0;
+ vq->queue_size_in_bytes = 0;
list_add_tail(&vq->list, &vdev->vqs);
vq->event = virtio_has_feature(vdev, VIRTIO_RING_F_EVENT_IDX);
@@ -243,12 +299,24 @@ static struct virtqueue *__vring_new_virtqueue(unsigned int index,
return vq;
}
+static void *vring_alloc_queue(size_t size, dma_addr_t *dma_handle)
+{
+ return dma_alloc_coherent(size, dma_handle);
+}
+
+static void vring_free_queue(size_t size, void *queue, dma_addr_t dma_handle)
+{
+ dma_free_coherent(queue, dma_handle, size);
+}
+
struct virtqueue *vring_create_virtqueue(unsigned int index, unsigned int num,
unsigned int vring_align,
struct virtio_device *vdev)
{
struct virtqueue *vq;
void *queue = NULL;
+ dma_addr_t dma_addr;
+ size_t queue_size_in_bytes;
struct vring vring;
/* We assume num is a power of 2 */
@@ -259,7 +327,7 @@ struct virtqueue *vring_create_virtqueue(unsigned int index, unsigned int num,
/* TODO: allocate each queue chunk individually */
for (; num && vring_size(num, vring_align) > PAGE_SIZE; num /= 2) {
- queue = memalign(PAGE_SIZE, vring_size(num, vring_align));
+ queue = vring_alloc_queue(vring_size(num, vring_align), &dma_addr);
if (queue)
break;
}
@@ -269,27 +337,31 @@ struct virtqueue *vring_create_virtqueue(unsigned int index, unsigned int num,
if (!queue) {
/* Try to get a single page. You are my only hope! */
- queue = memalign(PAGE_SIZE, vring_size(num, vring_align));
+ queue = vring_alloc_queue(vring_size(num, vring_align), &dma_addr);
}
if (!queue)
return NULL;
- memset(queue, 0, vring_size(num, vring_align));
+ queue_size_in_bytes = vring_size(num, vring_align);
vring_init(&vring, num, queue, vring_align);
vq = __vring_new_virtqueue(index, vring, vdev);
if (!vq) {
- free(queue);
+ vring_free_queue(queue_size_in_bytes, queue, dma_addr);
return NULL;
}
- vq_debug(vq, "created vring @ %p for vq with num %u\n", queue, num);
+ vq_debug(vq, "created vring @ (virt=%p, phys=%pad) for vq with num %u\n",
+ queue, &dma_addr, num);
+
+ vq->queue_dma_addr = dma_addr;
+ vq->queue_size_in_bytes = queue_size_in_bytes;
return vq;
}
void vring_del_virtqueue(struct virtqueue *vq)
{
- free(vq->vring.desc);
+ vring_free_queue(vq->queue_size_in_bytes, vq->vring.desc, vq->queue_dma_addr);
list_del(&vq->list);
free(vq);
}
@@ -301,18 +373,18 @@ unsigned int virtqueue_get_vring_size(struct virtqueue *vq)
dma_addr_t virtqueue_get_desc_addr(struct virtqueue *vq)
{
- return (dma_addr_t)vq->vring.desc;
+ return vq->queue_dma_addr;
}
dma_addr_t virtqueue_get_avail_addr(struct virtqueue *vq)
{
- return (dma_addr_t)vq->vring.desc +
+ return vq->queue_dma_addr +
((char *)vq->vring.avail - (char *)vq->vring.desc);
}
dma_addr_t virtqueue_get_used_addr(struct virtqueue *vq)
{
- return (dma_addr_t)vq->vring.desc +
+ return vq->queue_dma_addr +
((char *)vq->vring.used - (char *)vq->vring.desc);
}
diff --git a/include/linux/virtio_ring.h b/include/linux/virtio_ring.h
index 3c11592b09e4..c349af90ce50 100644
--- a/include/linux/virtio_ring.h
+++ b/include/linux/virtio_ring.h
@@ -108,6 +108,8 @@ struct virtqueue {
u16 last_used_idx;
u16 avail_flags_shadow;
u16 avail_idx_shadow;
+ dma_addr_t queue_dma_addr;
+ size_t queue_size_in_bytes;
};
/*
--
2.30.0
_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox
next prev parent reply other threads:[~2021-02-28 19:10 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-28 19:08 [PATCH 1/9] virtio: align virtio_config_ops::generation return type with Linux Ahmad Fatoum
2021-02-28 19:08 ` [PATCH 2/9] virtio: remove unused, left-over, virtio_config_ops::set_features Ahmad Fatoum
2021-02-28 19:08 ` [PATCH 3/9] hw_random: virtio: simplify code Ahmad Fatoum
2021-02-28 19:08 ` [PATCH 4/9] dma: move dma_map/unmap_single from ARM to common code Ahmad Fatoum
2021-03-03 12:08 ` Sascha Hauer
2021-03-03 16:12 ` [PATCH] " Ahmad Fatoum
2021-02-28 19:08 ` Ahmad Fatoum [this message]
2021-02-28 19:08 ` [PATCH 6/9] virtio: fix support for big-endian clients Ahmad Fatoum
2021-02-28 19:08 ` [PATCH 7/9] PCI: port Linux pci_find_capability Ahmad Fatoum
2021-02-28 19:08 ` [PATCH 8/9] virtio: support virtio-based device drivers over PCI Ahmad Fatoum
2021-02-28 19:08 ` [PATCH 9/9] virtio: virtio-pci: restrict MIPS support to MMU configuration Ahmad Fatoum
2021-03-02 14:28 ` Antony Pavlov
2021-03-03 17:10 ` Ahmad Fatoum
2021-03-01 16:29 ` [PATCH 1/9] virtio: align virtio_config_ops::generation return type with Linux Sascha Hauer
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=20210228190836.1451663-5-ahmad@a3f.at \
--to=ahmad@a3f.at \
--cc=barebox@lists.infradead.org \
/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