Re: [PATCH v19 2/9] usb: dwc3: core: Access XHCI address space temporarily to read port info

[Greg Kroah-Hartman]

On Thu, Apr 04, 2024 at 06:25:48PM -0700, Bjorn Andersson wrote:
> On Thu, Apr 04, 2024 at 02:58:29PM +0200, Greg Kroah-Hartman wrote:
> > On Thu, Apr 04, 2024 at 10:07:27AM +0200, Krzysztof Kozlowski wrote:
> > > On 04/04/2024 09:21, Johan Hovold wrote:
> > > > On Thu, Apr 04, 2024 at 10:42:22AM +0530, Krishna Kurapati wrote:
> > > >  
> > > >> +static int dwc3_get_num_ports(struct dwc3 *dwc)
> > > >> +{
> > > >> +	void __iomem *base;
> > > >> +	u8 major_revision;
> > > >> +	u32 offset;
> > > >> +	u32 val;
> > > >> +
> > > >> +	/*
> > > >> +	 * Remap xHCI address space to access XHCI ext cap regs since it is
> > > >> +	 * needed to get information on number of ports present.
> > > >> +	 */
> > > >> +	base = ioremap(dwc->xhci_resources[0].start,
> > > >> +		       resource_size(&dwc->xhci_resources[0]));
> > > >> +	if (!base)
> > > >> +		return PTR_ERR(base);
> > > > 
> > > > This is obviously still broken. You need to update the return value as
> > > > well.
> > > > 
> > > > Fix in v20.
> > > 
> > > If one patchset reaches 20 versions, I think it is time to stop and
> > > really think from the beginning, why issues keep appearing and reviewers
> > > are still not happy.
> > > 
> > > Maybe you did not perform extensive internal review, which you are
> > > encouraged to by your own internal policies, AFAIR. Before posting next
> > > version, please really get some internal review first.
> > 
> > Also get those internal reviewers to sign-off on the commits and have
> > that show up when you post them next.  That way they are also
> > responsible for this patchset, it's not fair that they are making you do
> > all the work here :)
> > 
> 
> I like this idea and I'm open to us changing our way of handling this.
> 
> But unless such internal review brings significant input to the
> development I'd say a s-o-b would take the credit from the actual
> author.

It does not do that at all.  It provides proof that someone else has
reviewed it and agrees with it.  Think of it as a "path of blame" for
when things go bad (i.e. there is a bug in the submission.)  Putting
your name on it makes you take responsibility if that happens.

> We've discussed a few times about carrying Reviewed-by et al from the
> internal reviews, but as maintainer I dislike this because I'd have no
> way to know if a r-b on vN means the patch was reviewed, or if it was
> just "accidentally" carried from v(N-1).
> But it might be worth this risk, is this something you think would be
> appropriate?

For some companies we REQUIRE this to happen due to low-quality
submissions and waste of reviewer's time.  Based on the track record
here for some of these patchsets, hopefully it doesn't become a
requirement for this company as well :)

thanks,

greg k-h