aboutsummaryrefslogtreecommitdiff
path: root/lib/dma.h
diff options
context:
space:
mode:
authorThanos Makatos <thanos.makatos@nutanix.com>2020-03-24 11:45:50 -0400
committerThanos Makatos <thanos.makatos@nutanix.com>2020-03-24 11:45:50 -0400
commita16180d98169816f62c26d0971300e4d04039c55 (patch)
tree19c0e08b2d7380bf1a0280958a82b85c828a6d02 /lib/dma.h
parent848572b45370ba6f4b3b708d9094042a71a21b9c (diff)
downloadlibvfio-user-a16180d98169816f62c26d0971300e4d04039c55.zip
libvfio-user-a16180d98169816f62c26d0971300e4d04039c55.tar.gz
libvfio-user-a16180d98169816f62c26d0971300e4d04039c55.tar.bz2
don't use fd for matching DMA region for unmap
The fd might come from another process so it doesn't make sense, the IOVA range alone is sufficient to determine the DMA region to unmap. Also, don't fail unmapping DMA regions that don't exist. Maybe we should revisit this and ignore the error at the client side? Signed-off-by: Thanos Makatos <thanos.makatos@nutanix.com>
Diffstat (limited to 'lib/dma.h')
-rw-r--r--lib/dma.h2
1 files changed, 1 insertions, 1 deletions
diff --git a/lib/dma.h b/lib/dma.h
index 8eda546..9f83ed7 100644
--- a/lib/dma.h
+++ b/lib/dma.h
@@ -107,7 +107,7 @@ dma_controller_add_region(lm_ctx_t *ctx, dma_controller_t *dma,
int
dma_controller_remove_region(lm_ctx_t *ctx, dma_controller_t *dma,
- dma_addr_t dma_addr, size_t size, int fd);
+ dma_addr_t dma_addr, size_t size);
// Helper for dma_addr_to_sg() slow path.
int