Previously, file operations on a file-backed mount used the current
process' credentials to access the backing FD. Attempting to do so on
Android lead to SELinux denials, as ACL rules on the backing file (e.g.
/system/apex/foo.apex) is restricted to a small set of process.
Arguably, this error is redundant and leaking implementation details, as
access to files on a mount is already ACL'ed by path.

Instead, override to use the opener's cred when accessing the backing
file. This makes the behavior similar to a loop-backed mount, which
uses kworker cred when accessing the backing file and does not cause
SELinux denials.

Signed-off-by: Tatsuyuki Ishi <ishitatsuyuki@google.com>
---
 fs/erofs/fileio.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/fs/erofs/fileio.c b/fs/erofs/fileio.c
index 7d81f504bff08f3d5c5d44d131460df5c3e7847d..df5cc63f2c01eb5e7ec4afab9e054ea12cea7175 100644
--- a/fs/erofs/fileio.c
+++ b/fs/erofs/fileio.c
@@ -47,6 +47,7 @@ static void erofs_fileio_ki_complete(struct kiocb *iocb, long ret)
 
 static void erofs_fileio_rq_submit(struct erofs_fileio_rq *rq)
 {
+	const struct cred *old_cred;
 	struct iov_iter iter;
 	int ret;
 
@@ -60,7 +61,9 @@ static void erofs_fileio_rq_submit(struct erofs_fileio_rq *rq)
 		rq->iocb.ki_flags = IOCB_DIRECT;
 	iov_iter_bvec(&iter, ITER_DEST, rq->bvecs, rq->bio.bi_vcnt,
 		      rq->bio.bi_iter.bi_size);
+	old_cred = override_creds(rq->iocb.ki_filp->f_cred);
 	ret = vfs_iocb_iter_read(rq->iocb.ki_filp, &rq->iocb, &iter);
+	revert_creds(old_cred);
 	if (ret != -EIOCBQUEUED)
 		erofs_fileio_ki_complete(&rq->iocb, ret);
 }

---
base-commit: cd2e103d57e5615f9bb027d772f93b9efd567224
change-id: 20250612-b4-erofs-impersonate-d6c2926c56ca

Best regards,
-- 
Tatsuyuki Ishi <ishitatsuyuki@google.com>

On 2025/6/12 18:18, Tatsuyuki Ishi wrote:
> Previously, file operations on a file-backed mount used the current
> process' credentials to access the backing FD. Attempting to do so on
> Android lead to SELinux denials, as ACL rules on the backing file (e.g.
> /system/apex/foo.apex) is restricted to a small set of process.
> Arguably, this error is redundant and leaking implementation details, as
> access to files on a mount is already ACL'ed by path.
> 
> Instead, override to use the opener's cred when accessing the backing
> file. This makes the behavior similar to a loop-backed mount, which
> uses kworker cred when accessing the backing file and does not cause
> SELinux denials.
> 
> Signed-off-by: Tatsuyuki Ishi <ishitatsuyuki@google.com>
> ---
>   fs/erofs/fileio.c | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/fs/erofs/fileio.c b/fs/erofs/fileio.c
> index 7d81f504bff08f3d5c5d44d131460df5c3e7847d..df5cc63f2c01eb5e7ec4afab9e054ea12cea7175 100644
> --- a/fs/erofs/fileio.c
> +++ b/fs/erofs/fileio.c
> @@ -47,6 +47,7 @@ static void erofs_fileio_ki_complete(struct kiocb *iocb, long ret)
>   
>   static void erofs_fileio_rq_submit(struct erofs_fileio_rq *rq)
>   {
> +	const struct cred *old_cred;
>   	struct iov_iter iter;
>   	int ret;
>   
> @@ -60,7 +61,9 @@ static void erofs_fileio_rq_submit(struct erofs_fileio_rq *rq)
>   		rq->iocb.ki_flags = IOCB_DIRECT;
>   	iov_iter_bvec(&iter, ITER_DEST, rq->bvecs, rq->bio.bi_vcnt,
>   		      rq->bio.bi_iter.bi_size);
> +	old_cred = override_creds(rq->iocb.ki_filp->f_cred);Yeah, rq->iocb.ki_filp keep the opener's cred, so:

Reviewed-by: Hongbo Li <lihongbo22@huawei.com>

Thanks,
Hongbo

>   	ret = vfs_iocb_iter_read(rq->iocb.ki_filp, &rq->iocb, &iter);
> +	revert_creds(old_cred);
>   	if (ret != -EIOCBQUEUED)
>   		erofs_fileio_ki_complete(&rq->iocb, ret);
>   }
> 
> ---
> base-commit: cd2e103d57e5615f9bb027d772f93b9efd567224
> change-id: 20250612-b4-erofs-impersonate-d6c2926c56ca
> 
> Best regards,

Hi Tatsuyuki,

On 2025/6/12 18:18, Tatsuyuki Ishi wrote:
> Previously, file operations on a file-backed mount used the current
> process' credentials to access the backing FD. Attempting to do so on
> Android lead to SELinux denials, as ACL rules on the backing file (e.g.
> /system/apex/foo.apex) is restricted to a small set of process.
> Arguably, this error is redundant and leaking implementation details, as
> access to files on a mount is already ACL'ed by path.
> 
> Instead, override to use the opener's cred when accessing the backing
> file. This makes the behavior similar to a loop-backed mount, which
> uses kworker cred when accessing the backing file and does not cause
> SELinux denials.
> 
> Signed-off-by: Tatsuyuki Ishi <ishitatsuyuki@google.com>

Thanks for the patch.  I think overlayfs uses the similar policy
(mounter's cred), which is the same as the opener's cred here
(because it opens backing file in the mount context), so:

Reviewed-by: Gao Xiang <hsiangkao@linux.alibaba.com>

Thanks,
Gao Xiang