drm_gpusvm_get_pages() only stored the local dpagemap into
svm_pages->dpagemap on the success path. If a later page failed (e.g.
-EOPNOTSUPP when ctx->allow_mixed is false) and jumped to err_unmap,
svm_pages->dpagemap was still NULL, so __drm_gpusvm_unmap_pages() skipped
device_unmap() and leaked the device mappings already created.
Assign svm_pages->dpagemap when the first device page is mapped so the
err_unmap path can device_unmap() those mappings.
This issue was found by Sashiko AI review.
Fixes: f70da6f99d4f ("drm/gpusvm: pull out drm_gpusvm_pages substructure")
Cc: [email protected]
Reviewed-by: Matthew Brost <[email protected]>
Signed-off-by: Honglei Huang <[email protected]>
---
drivers/gpu/drm/drm_gpusvm.c | 16 +++++++++++-----
1 file changed, 11 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/drm_gpusvm.c b/drivers/gpu/drm/drm_gpusvm.c
index 44bb19658dd..9a06ff7d260 100644
--- a/drivers/gpu/drm/drm_gpusvm.c
+++ b/drivers/gpu/drm/drm_gpusvm.c
@@ -1544,6 +1544,16 @@ int drm_gpusvm_get_pages(struct drm_gpusvm *gpusvm,
err = -EAGAIN;
goto err_unmap;
}
+
+ /*
+ * Set the dpagemap as soon as the first
+ * device page is mapped so the err_unmap path
+ * can device_unmap() the device mappings that
+ * have already been created.
+ */
+ drm_pagemap_get(dpagemap);
+ drm_pagemap_put(svm_pages->dpagemap);
+ svm_pages->dpagemap = dpagemap;
}
svm_pages->dma_addr[j] =
dpagemap->ops->device_map(dpagemap,
@@ -1611,12 +1621,8 @@ int drm_gpusvm_get_pages(struct drm_gpusvm *gpusvm,
goto err_unmap;
}
- if (pagemap) {
+ if (pagemap)
flags.has_devmem_pages = true;
- drm_pagemap_get(dpagemap);
- drm_pagemap_put(svm_pages->dpagemap);
- svm_pages->dpagemap = dpagemap;
- }
/* WRITE_ONCE pairs with READ_ONCE for opportunistic checks */
WRITE_ONCE(svm_pages->flags.__flags, flags.__flags);
--
2.34.1
