rpmsg: use less buffers when vrings are small

Adjust the number of rpmsg buffers to rely on the size of the
vring, instead of using the hard coded value of 512 (256 per
direction).

This is needed when small vrings are being used, where 256
buffers are too much to fit in a vring.

While considering the vring size, keep using the 512 hard coded
value as an upper limit to avoid wacky resource tables consuming
unreasonable amount of memory.

NOTE: The number of buffers is already assumed to be symmetrical
in each direction, and that logic is unchanged.

Signed-off-by: Suman Anna <s-anna@ti.com>
[edit commit message, small code and comment simplification]
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>

1 files changed, 30 insertions, 14 deletions

diff --git a/drivers/rpmsg/virtio_rpmsg_bus.c b/drivers/rpmsg/virtio_rpmsg_bus.c
index b6135d4d54eb..92f6af6da699 100644
--- a/drivers/rpmsg/virtio_rpmsg_bus.c
+++ b/drivers/rpmsg/virtio_rpmsg_bus.c
@@ -41,6 +41,7 @@
  * @svq:        tx virtqueue
  * @rbufs:      kernel address of rx buffers
  * @sbufs:      kernel address of tx buffers
+ * @num_bufs:   total number of buffers for rx and tx
  * @last_sbuf:  index of last tx buffer used
  * @bufs_dma:   dma base addr of the buffers
  * @tx_lock:    protects svq, sbufs and sleepers, to allow concurrent senders.
@@ -60,6 +61,7 @@ struct virtproc_info {
         struct virtio_device *vdev;
         struct virtqueue *rvq, *svq;
         void *rbufs, *sbufs;
+        unsigned int num_bufs;
         int last_sbuf;
         dma_addr_t bufs_dma;
         struct mutex tx_lock;
@@ -86,13 +88,14 @@ struct rpmsg_channel_info {
 #define to_rpmsg_driver(d) container_of(d, struct rpmsg_driver, drv)
 
 /*
- * We're allocating 512 buffers of 512 bytes for communications, and then
- * using the first 256 buffers for RX, and the last 256 buffers for TX.
+ * We're allocating buffers of 512 bytes each for communications. The
+ * number of buffers will be computed from the number of buffers supported
+ * by the vring, upto a maximum of 512 buffers (256 in each direction).
  *
  * Each buffer will have 16 bytes for the msg header and 496 bytes for
  * the payload.
  *
- * This will require a total space of 256KB for the buffers.
+ * This will utilize a maximum total space of 256KB for the buffers.
  *
  * We might also want to add support for user-provided buffers in time.
  * This will allow bigger buffer size flexibility, and can also be used
@@ -102,9 +105,8 @@ struct rpmsg_channel_info {
  * can change this without changing anything in the firmware of the remote
  * processor.
  */
-#define RPMSG_NUM_BUFS          (512)
+#define MAX_RPMSG_NUM_BUFS      (512)
 #define RPMSG_BUF_SIZE          (512)
-#define RPMSG_TOTAL_BUF_SPACE   (RPMSG_NUM_BUFS * RPMSG_BUF_SIZE)
 
 /*
  * Local addresses are dynamically allocated on-demand.
@@ -579,7 +581,7 @@ static void *get_a_tx_buf(struct virtproc_info *vrp)
          * either pick the next unused tx buffer
          * (half of our buffers are used for sending messages)
          */
-        if (vrp->last_sbuf < RPMSG_NUM_BUFS / 2)
+        if (vrp->last_sbuf < vrp->num_bufs / 2)
                 ret = vrp->sbufs + RPMSG_BUF_SIZE * vrp->last_sbuf++;
         /* or recycle a used one */
         else
@@ -948,6 +950,7 @@ static int rpmsg_probe(struct virtio_device *vdev)
         struct virtproc_info *vrp;
         void *bufs_va;
         int err = 0, i;
+        size_t total_buf_space;
 
         vrp = kzalloc(sizeof(*vrp), GFP_KERNEL);
         if (!vrp)
@@ -968,10 +971,22 @@ static int rpmsg_probe(struct virtio_device *vdev)
         vrp->rvq = vqs[0];
         vrp->svq = vqs[1];
 
+        /* we expect symmetric tx/rx vrings */
+        WARN_ON(virtqueue_get_vring_size(vrp->rvq) !=
+                virtqueue_get_vring_size(vrp->svq));
+
+        /* we need less buffers if vrings are small */
+        if (virtqueue_get_vring_size(vrp->rvq) < MAX_RPMSG_NUM_BUFS / 2)
+                vrp->num_bufs = virtqueue_get_vring_size(vrp->rvq) * 2;
+        else
+                vrp->num_bufs = MAX_RPMSG_NUM_BUFS;
+
+        total_buf_space = vrp->num_bufs * RPMSG_BUF_SIZE;
+
         /* allocate coherent memory for the buffers */
         bufs_va = dma_alloc_coherent(vdev->dev.parent->parent,
-                                RPMSG_TOTAL_BUF_SPACE,
-                                &vrp->bufs_dma, GFP_KERNEL);
+                                     total_buf_space, &vrp->bufs_dma,
+                                     GFP_KERNEL);
         if (!bufs_va) {
                 err = -ENOMEM;
                 goto vqs_del;
@@ -984,10 +999,10 @@ static int rpmsg_probe(struct virtio_device *vdev)
         vrp->rbufs = bufs_va;
 
         /* and half is dedicated for TX */
-        vrp->sbufs = bufs_va + RPMSG_TOTAL_BUF_SPACE / 2;
+        vrp->sbufs = bufs_va + total_buf_space / 2;
 
         /* set up the receive buffers */
-        for (i = 0; i < RPMSG_NUM_BUFS / 2; i++) {
+        for (i = 0; i < vrp->num_bufs / 2; i++) {
                 struct scatterlist sg;
                 void *cpu_addr = vrp->rbufs + i * RPMSG_BUF_SIZE;
 
@@ -1023,8 +1038,8 @@ static int rpmsg_probe(struct virtio_device *vdev)
         return 0;
 
 free_coherent:
-        dma_free_coherent(vdev->dev.parent->parent, RPMSG_TOTAL_BUF_SPACE,
-                                        bufs_va, vrp->bufs_dma);
+        dma_free_coherent(vdev->dev.parent->parent, total_buf_space,
+                          bufs_va, vrp->bufs_dma);
 vqs_del:
         vdev->config->del_vqs(vrp->vdev);
 free_vrp:
@@ -1042,6 +1057,7 @@ static int rpmsg_remove_device(struct device *dev, void *data)
 static void rpmsg_remove(struct virtio_device *vdev)
 {
         struct virtproc_info *vrp = vdev->priv;
+        size_t total_buf_space = vrp->num_bufs * RPMSG_BUF_SIZE;
         int ret;
 
         vdev->config->reset(vdev);
@@ -1057,8 +1073,8 @@ static void rpmsg_remove(struct virtio_device *vdev)
 
         vdev->config->del_vqs(vrp->vdev);
 
-        dma_free_coherent(vdev->dev.parent->parent, RPMSG_TOTAL_BUF_SPACE,
-                                        vrp->rbufs, vrp->bufs_dma);
+        dma_free_coherent(vdev->dev.parent->parent, total_buf_space,
+                          vrp->rbufs, vrp->bufs_dma);
 
         kfree(vrp);
 }