On Oct 04 2024, SurajSonawane2415 wrote:
> Fix the uninitialized symbol 'rv' in the function ish_fw_xfer_direct_dma
> to resolve the following warning from the smatch tool:
> drivers/hid/intel-ish-hid/ishtp-fw-loader.c:714 ish_fw_xfer_direct_dma()
> error: uninitialized symbol 'rv'.
> Initialize 'rv' to 0 to prevent undefined behavior from uninitialized
> access.
Thanks for the patch!
FWIW, I tried to understand why this went unnoticced for so long: the
only case where rv might be used uninitialized is if fw->size == 0.
In that case, all that happens is that load_fw_from_host() tries to call
ish_fw_start() which will in turn overwrite the return value.
So I'm not sure if request_firmware() can return a firmware of size 0,
and if ish_fw_start() will be happy about that, but I would hope one of
the 2 would complain and not crash the kernel.
Anyway, I'll grab this one and adding Fixes and CC:stable as this might
catch an interesting bug, and fixing smatch complains is always good.
Cheers,
Benjamin
>
> Signed-off-by: SurajSonawane2415 <surajsonawane0215@xxxxxxxxx>
> ---
> drivers/hid/intel-ish-hid/ishtp-fw-loader.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/hid/intel-ish-hid/ishtp-fw-loader.c b/drivers/hid/intel-ish-hid/ishtp-fw-loader.c
> index e157863a8..b3c3cfcd9 100644
> --- a/drivers/hid/intel-ish-hid/ishtp-fw-loader.c
> +++ b/drivers/hid/intel-ish-hid/ishtp-fw-loader.c
> @@ -635,7 +635,7 @@ static int ish_fw_xfer_direct_dma(struct ishtp_cl_data *client_data,
> const struct firmware *fw,
> const struct shim_fw_info fw_info)
> {
> - int rv;
> + int rv = 0;
> void *dma_buf;
> dma_addr_t dma_buf_phy;
> u32 fragment_offset, fragment_size, payload_max_size;
> --
> 2.34.1
>
