On 7/29/20 5:29 AM, Barnabás Pőcze wrote:
2020. július 29., szerda 5:17 keltezéssel, Daniel Dadap <ddadap@xxxxxxxxxx> írta:
Thanks for your comments. A v2 patch will follow after I've had some
time to test it more; in the meantime, I've responded to your
suggestions inline:
[...]
+
+static int mxds_gmux_init(void)
+{
+ int ret = 0;
+ struct pci_dev *dev = NULL;
+ static struct vga_switcheroo_handler handler = {
+ .switchto = mxds_gmux_switchto,
+ .get_client_id = mxds_gmux_get_client_id,
+ };
+
Any reason why "handler" is inside the function and not const?
You're right, it can be const. I have it in the function (with static
storage) because we don't need to reference it anywhere else. I'd think
the static storage would allow the pointer to the struct to stay alive
even after the init function exits, but if you think it would be better
to have it out of the function's scope I can move it.
I see. I think having it out of the function better signals the intention that
this variable is supposed to live as long as the module is loaded. Furthermore,
- although I am not sure about this, but - I think having a function static variable
here prevents mxds_gmux_init() from being marked __init.
Also, I forgot to mention before, but mxds_gmux_fini() is not marked __exit.
Any reason why?
The same reason mxds_gmux_init() wasn't marked __init: an oversight.
