Good day,
On Thu, Jan 22, 2026 at 11:24:43AM +0800, Peng Fan (OSS) wrote:
> From: Peng Fan <[email protected]>
>
> When starting a firmware without a resource table after previously running
> one that had a resource table, imx_rproc_elf_find_loaded_rsc_table() may
> incorrectly return a valid device memory pointer (priv->rsc_table).
priv->rsc_table is not NULL if the DT has a "rsc-table" entry, indicating that
_if_ there is a resource table in memory, that's where it should be. Function
imx_rproc_elf_find_loaded_rsc_table() is buggy so the narrative about a
previously running FW with a valid resource table can be dropped.
>
> In this case rproc->cached_table is NULL because the current firmware does
> not contain a resource table, but the remoteproc core still interprets the
> non-NULL return value as a loaded resource table and attempts to memcpy()
> from rproc->cached_table, leading to a NULL pointer dereference and kernel
> panic.
>
> Fix this by returning NULL from imx_rproc_elf_find_loaded_rsc_table() when
> there is no cached resource table for the current firmware. This ensures
> that a loaded resource table is only reported when a valid cached_table
> exists, which matches the remoteproc core expectations.
>
> This issue can be reproduced by:
> 1) start a firmware with a resource table
> 2) stop the remote processor
> 3) start a firmware without a resource table
>
> With this change, starting a firmware without a resource table no longer
> causes kernel dump.
>
> Fixes: e954a1bd1610 ("remoteproc: imx_rproc: Use imx specific hook for
> find_loaded_rsc_table")
> Cc: [email protected]
> Signed-off-by: Peng Fan <[email protected]>
> ---
> drivers/remoteproc/imx_rproc.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/drivers/remoteproc/imx_rproc.c b/drivers/remoteproc/imx_rproc.c
> index
> 375de79168a1c8d11b87ac1bd63774a3feac106d..cf044b385b58fe1e17d0fc440c243d76ecf020ae
> 100644
> --- a/drivers/remoteproc/imx_rproc.c
> +++ b/drivers/remoteproc/imx_rproc.c
> @@ -729,6 +729,10 @@ imx_rproc_elf_find_loaded_rsc_table(struct rproc *rproc,
> const struct firmware *
> {
> struct imx_rproc *priv = rproc->priv;
>
> + /* No resource table in the firmware */
> + if (!rproc->cached_table)
> + return NULL;
> +
I think rproc->cached_table should be kept for internal remoteproc core usage
only. Please use rproc->table_ptr.
Thanks,
Mathieu
> if (priv->rsc_table)
> return (struct resource_table *)priv->rsc_table;
>
>
> ---
> base-commit: e3b32dcb9f23e3c3927ef3eec6a5842a988fb574
> change-id: 20260122-imx-rproc-fix-e206f8e6e477
>
> Best regards,
> --
> Peng Fan <[email protected]>
>
