[PATCH for-5.2 08/10] block/export: port virtio-blk discard/write zeroes input validation

[Stefan Hajnoczi]

Validate discard/write zeroes the same way we do for virtio-blk. Some of
these checks are mandated by the VIRTIO specification, others are
internal to QEMU.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/export/vhost-user-blk-server.c | 115 +++++++++++++++++++++------
 1 file changed, 92 insertions(+), 23 deletions(-)

diff --git a/block/export/vhost-user-blk-server.c 
b/block/export/vhost-user-blk-server.c
index 62672d1cb9..3295d3c951 100644
--- a/block/export/vhost-user-blk-server.c
+++ b/block/export/vhost-user-blk-server.c
@@ -22,6 +22,8 @@
 
 enum {
     VHOST_USER_BLK_NUM_QUEUES_DEFAULT = 1,
+    VHOST_USER_BLK_MAX_DISCARD_SECTORS = 32768,
+    VHOST_USER_BLK_MAX_WRITE_ZEROES_SECTORS = 32768,
 };
 struct virtio_blk_inhdr {
     unsigned char status;
@@ -58,30 +60,101 @@ static void vu_blk_req_complete(VuBlkReq *req)
     free(req);
 }
 
+static bool vu_blk_sect_range_ok(VuBlkExport *vexp, uint64_t sector,
+                                 size_t size)
+{
+    uint64_t nb_sectors = size >> BDRV_SECTOR_BITS;
+    uint64_t total_sectors;
+
+    if (nb_sectors > BDRV_REQUEST_MAX_SECTORS) {
+        return false;
+    }
+    if ((sector << BDRV_SECTOR_BITS) % vexp->blk_size) {
+        return false;
+    }
+    blk_get_geometry(vexp->export.blk, &total_sectors);
+    if (sector > total_sectors || nb_sectors > total_sectors - sector) {
+        return false;
+    }
+    return true;
+}
+
 static int coroutine_fn
-vu_blk_discard_write_zeroes(BlockBackend *blk, struct iovec *iov,
+vu_blk_discard_write_zeroes(VuBlkExport *vexp, struct iovec *iov,
                             uint32_t iovcnt, uint32_t type)
 {
+    BlockBackend *blk = vexp->export.blk;
     struct virtio_blk_discard_write_zeroes desc;
-    ssize_t size = iov_to_buf(iov, iovcnt, 0, &desc, sizeof(desc));
+    ssize_t size;
+    uint64_t sector;
+    uint32_t num_sectors;
+    uint32_t max_sectors;
+    uint32_t flags;
+    int bytes;
+
+    /* Only one desc is currently supported */
+    if (unlikely(iov_size(iov, iovcnt) > sizeof(desc))) {
+        return VIRTIO_BLK_S_UNSUPP;
+    }
+
+    size = iov_to_buf(iov, iovcnt, 0, &desc, sizeof(desc));
     if (unlikely(size != sizeof(desc))) {
-        error_report("Invalid size %zd, expect %zu", size, sizeof(desc));
-        return -EINVAL;
+        error_report("Invalid size %zd, expected %zu", size, sizeof(desc));
+        return VIRTIO_BLK_S_IOERR;
     }
 
-    uint64_t range[2] = { le64_to_cpu(desc.sector) << 9,
-                          le32_to_cpu(desc.num_sectors) << 9 };
-    if (type == VIRTIO_BLK_T_DISCARD) {
-        if (blk_co_pdiscard(blk, range[0], range[1]) == 0) {
-            return 0;
+    sector = le64_to_cpu(desc.sector);
+    num_sectors = le32_to_cpu(desc.num_sectors);
+    flags = le32_to_cpu(desc.flags);
+    max_sectors = (type == VIRTIO_BLK_T_WRITE_ZEROES) ?
+                  VHOST_USER_BLK_MAX_WRITE_ZEROES_SECTORS :
+                  VHOST_USER_BLK_MAX_DISCARD_SECTORS;
+
+    /* This check ensures "num_sectors << BDRV_SECTOR_BITS" fits in an int */
+    if (unlikely(num_sectors > max_sectors)) {
+        return VIRTIO_BLK_S_IOERR;
+    }
+
+    bytes = num_sectors << BDRV_SECTOR_BITS;
+
+    if (unlikely(!vu_blk_sect_range_ok(vexp, sector, bytes))) {
+        return VIRTIO_BLK_S_IOERR;
+    }
+
+    /*
+     * The device MUST set the status byte to VIRTIO_BLK_S_UNSUPP for discard
+     * and write zeroes commands if any unknown flag is set.
+     */
+    if (unlikely(flags & ~VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP)) {
+        return VIRTIO_BLK_S_UNSUPP;
+    }
+
+    if (type == VIRTIO_BLK_T_WRITE_ZEROES) {
+        int blk_flags = 0;
+
+        if (flags & VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP) {
+            blk_flags |= BDRV_REQ_MAY_UNMAP;
+        }
+
+        if (blk_co_pwrite_zeroes(blk, sector << BDRV_SECTOR_BITS,
+                                 bytes, blk_flags) == 0) {
+            return VIRTIO_BLK_S_OK;
+        }
+    } else if (type == VIRTIO_BLK_T_DISCARD) {
+        /*
+         * The device MUST set the status byte to VIRTIO_BLK_S_UNSUPP for
+         * discard commands if the unmap flag is set.
+         */
+        if (unlikely(flags & VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP)) {
+            return VIRTIO_BLK_S_UNSUPP;
         }
-    } else if (type == VIRTIO_BLK_T_WRITE_ZEROES) {
-        if (blk_co_pwrite_zeroes(blk, range[0], range[1], 0) == 0) {
-            return 0;
+
+        if (blk_co_pdiscard(blk, sector << BDRV_SECTOR_BITS, bytes) == 0) {
+            return VIRTIO_BLK_S_OK;
         }
     }
 
-    return -EINVAL;
+    return VIRTIO_BLK_S_IOERR;
 }
 
 static void coroutine_fn vu_blk_virtio_process_req(void *opaque)
@@ -170,19 +243,13 @@ static void coroutine_fn vu_blk_virtio_process_req(void 
*opaque)
     }
     case VIRTIO_BLK_T_DISCARD:
     case VIRTIO_BLK_T_WRITE_ZEROES: {
-        int rc;
-
         if (!vexp->writable) {
             req->in->status = VIRTIO_BLK_S_IOERR;
             break;
         }
 
-        rc = vu_blk_discard_write_zeroes(blk, &elem->out_sg[1], out_num, type);
-        if (rc == 0) {
-            req->in->status = VIRTIO_BLK_S_OK;
-        } else {
-            req->in->status = VIRTIO_BLK_S_IOERR;
-        }
+        req->in->status = vu_blk_discard_write_zeroes(vexp, elem->out_sg,
+                                                      out_num, type);
         break;
     }
     default:
@@ -352,10 +419,12 @@ vu_blk_initialize_config(BlockDriverState *bs,
     config->min_io_size = cpu_to_le16(1);
     config->opt_io_size = cpu_to_le32(1);
     config->num_queues = cpu_to_le16(num_queues);
-    config->max_discard_sectors = cpu_to_le32(32768);
+    config->max_discard_sectors =
+        cpu_to_le32(VHOST_USER_BLK_MAX_DISCARD_SECTORS);
     config->max_discard_seg = cpu_to_le32(1);
     config->discard_sector_alignment = cpu_to_le32(config->blk_size >> 9);
-    config->max_write_zeroes_sectors = cpu_to_le32(32768);
+    config->max_write_zeroes_sectors
+        = cpu_to_le32(VHOST_USER_BLK_MAX_WRITE_ZEROES_SECTORS);
     config->max_write_zeroes_seg = cpu_to_le32(1);
 }
 
-- 
2.28.0