On Tue, Nov 19, 2024 at 11:33:51PM +0100, Marco Felsch wrote:
> It's no error if a driver indicates that the firmware is already
> up-to-date and the update can be skipped.
>
> Signed-off-by: Marco Felsch <m.felsch@xxxxxxxxxxxxxx>
> ---
> drivers/base/firmware_loader/sysfs_upload.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/drivers/base/firmware_loader/sysfs_upload.c b/drivers/base/firmware_loader/sysfs_upload.c
> index b3cbe5b156e3..44f3d8fa5e64 100644
> --- a/drivers/base/firmware_loader/sysfs_upload.c
> +++ b/drivers/base/firmware_loader/sysfs_upload.c
> @@ -174,6 +174,10 @@ static void fw_upload_main(struct work_struct *work)
> fw_upload_update_progress(fwlp, FW_UPLOAD_PROG_PREPARING);
> ret = fwlp->ops->prepare(fwl, fwlp->data, fwlp->remaining_size);
> if (ret != FW_UPLOAD_ERR_NONE) {
> + if (ret == FW_UPLOAD_ERR_SKIP) {
> + dev_info(fw_dev, "firmware already up-to-date, skip update\n");
> + ret = FW_UPLOAD_ERR_NONE;
> + }
If you change the error-code from FW_UPLOAD_ERR_SKIP to
FW_UPLOAD_ERR_NONE, then the "skip" string provided in the previous
patch will never be seen. There are currently no other instances where
an error code requires special-case modifications to the fw_upload
code and I don't think it is necessary to add it here.
The dev_info() message above can be provided by the device driver
that is using this API.
I think you can either:
(1) allow "skip" to be treated as an error. The update didn't happen...
-or-
(2) The prepare function could detect the situation and set
a flag in the same device driver. Your write function could
set *written to the full data size and return without writing
anything. Your poll_complete handler could also return
FW_UPLOAD_ERR_NONE. Then you don't need to add FW_UPLOAD_ERR_SKIP
at all. You would get the info message from the device driver
and fw_upload would exit without an error.
Thanks,
- Russ
> fw_upload_set_error(fwlp, ret);
> goto putdev_exit;
> }
>
> --
> 2.39.5
>
