Re: [PATCH v3 15/15] Drivers: hv: Add modules to expose /dev/mshv to VMMs running on Hyper-V

[Saurabh Singh Sengar]

On Sat, Sep 23, 2023 at 08:58:00PM +0000, Wei Liu wrote:
> Greg, thank you for looking at the code.
> 
> On Sat, Sep 23, 2023 at 09:56:13AM +0200, Greg KH wrote:
> > On Fri, Sep 22, 2023 at 11:38:35AM -0700, Nuno Das Neves wrote:
> > > +static int __init mshv_vtl_init(void)
> > > +{
> > > +	int ret;
> > > +
> > > +	tasklet_init(&msg_dpc, mshv_vtl_sint_on_msg_dpc, 0);
> > > +	init_waitqueue_head(&fd_wait_queue);
> > > +
> > > +	if (mshv_vtl_get_vsm_regs()) {
> > > +		pr_emerg("%s: Unable to get VSM capabilities !!\n", __func__);
> > > +		BUG();
> > > +	}
> > 
> > 
> > So you crash the whole kernel if someone loads this module on a non-mshv
> > system?
> > 
> > That seems quite excessive and hostile :(
> > 
> 
> I agree this can be more lenient. It should be safe to just return an
> error code (ENODEV) and let user space decide the next action.
> 
> Saurabh, let me know what you think.

Thanks for reporting this. I agree, returning an error makes more sense here.
We should do this for both BUG() in this init.

- Saurabh

> 
> Thanks,
> Wei.
> 
> > Or am I somehow reading this incorrectly?
> > 
> > thanks,
> > 
> > greg k-h
> >