arm64: USB_ARCH_HAS_EHCI

arm64: USB_ARCH_HAS_EHCI

[Paul Bolle]

Paul,

Your commit d035fdfa27ac ("arm64: Add Tegra132 support") is included in
today's linux-next (ie, next-20150127). I noticed because a script I use
to check linux-next spotted a minor problem with it.

It adds a select for the Kconfig symbol USB_ARCH_HAS_EHCI. That Kconfig
was obsoleted in v3.11 and removed in v3.15. See commit b797b76fb464
("usb: host: remove USB_ARCH_HAS_?HCI"). That select is now a nop.

Should I submit the trivial patch to remove that line or do you prefer
to do that yourself?

Thanks,


Paul Bolle

Re: arm64: USB_ARCH_HAS_EHCI

[Paul Walmsley]

On Tue, 27 Jan 2015, Paul Bolle wrote:

> Your commit d035fdfa27ac ("arm64: Add Tegra132 support") is included in
> today's linux-next (ie, next-20150127). I noticed because a script I use
> to check linux-next spotted a minor problem with it.
> 
> It adds a select for the Kconfig symbol USB_ARCH_HAS_EHCI. That Kconfig
> was obsoleted in v3.11 and removed in v3.15. See commit b797b76fb464
> ("usb: host: remove USB_ARCH_HAS_?HCI"). That select is now a nop.
> 
> Should I submit the trivial patch to remove that line or do you prefer
> to do that yourself?

Thanks, I'll post a new version.


- Paul

Re: arm64: USB_ARCH_HAS_EHCI

[Paul Walmsley]

On Wed, 28 Jan 2015, Paul Bolle wrote:

> On Tue, 2015-01-27 at 21:31 +0000, Paul Walmsley wrote:
> > On Tue, 27 Jan 2015, Paul Bolle wrote:
> > 
> > > Your commit d035fdfa27ac ("arm64: Add Tegra132 support") is included in
> > > today's linux-next (ie, next-20150127). I noticed because a script I use
> > > to check linux-next spotted a minor problem with it.
> > > 
> > > It adds a select for the Kconfig symbol USB_ARCH_HAS_EHCI. That Kconfig
> > > was obsoleted in v3.11 and removed in v3.15. See commit b797b76fb464
> > > ("usb: host: remove USB_ARCH_HAS_?HCI"). That select is now a nop.
> > > 
> > > Should I submit the trivial patch to remove that line or do you prefer
> > > to do that yourself?
> > 
> > Thanks, I'll post a new version.
> 
> I just noticed that your commit added another unneeded select. The entry
> for ARCH_TEGRA selects HAVE_SMP, but HAVE_SMP is only available on arm
> or xtensa. There's no point in selecting it in arm64.

OK thanks, will roll that in.  Looks like Thierry's already sent a pull 
request for the original patch, so I'll probably just send a follow-up.

Is the Kconfig checker script you mentioned on Tuesday available in the 
mainline tree?  I don't see anything obvious under the "Static Analysers" 
section of the Makefile, but perhaps I'm overlooking it.


- Paul

Re: arm64: USB_ARCH_HAS_EHCI

[Paul Bolle]

Hi Paul,

On Wed, 2015-01-28 at 16:25 +0000, Paul Walmsley wrote:
> Is the Kconfig checker script you mentioned on Tuesday available in the 
> mainline tree?  I don't see anything obvious under the "Static Analysers" 
> section of the Makefile, but perhaps I'm overlooking it.

No, it's not.

In tree you'll find scripts/checkkconfigsymbols.py. That catches over
90% of the stuff my local script catches, in less than 10% of runtime.
It would certainly have caught your use of USB_ARCH_HAS_EHCI.

My local script is my third (or fourth) attempt at a script that checks
these kind of issues. I've used it on linux-next since for about 3/4 of
a year now. While doing that I ran into people that are planning into
adding similar, but far more advanced, functionality into one of the
bots that is checking linux. So I never bothered making my script
public. (I'm not sure what the status of those plan is now.)

In the mean time I'll keep on checking linux-next with my script. That
should also help to see what problems an actual bot could run into, for
the benefit of those people. Problems like:
- should it check each commit (I think it should only check releases);
- should it be a true bot (I think it needs a wetware frontend to
  filter out non-issues).

Hope this answers your question.


Paul Bolle

Re: arm64: USB_ARCH_HAS_EHCI

[Paul Walmsley]

Hi Paul,

On Wed, 28 Jan 2015, Paul Bolle wrote:

> On Wed, 2015-01-28 at 16:25 +0000, Paul Walmsley wrote:
> > Is the Kconfig checker script you mentioned on Tuesday available in the 
> > mainline tree?  I don't see anything obvious under the "Static Analysers" 
> > section of the Makefile, but perhaps I'm overlooking it.
> 
> No, it's not.
> 
> In tree you'll find scripts/checkkconfigsymbols.py. That catches over
> 90% of the stuff my local script catches, in less than 10% of runtime.
> It would certainly have caught your use of USB_ARCH_HAS_EHCI.

OK great, I'll add that to my static analysis scripts.

> My local script is my third (or fourth) attempt at a script that checks
> these kind of issues. I've used it on linux-next since for about 3/4 of
> a year now. While doing that I ran into people that are planning into
> adding similar, but far more advanced, functionality into one of the
> bots that is checking linux. So I never bothered making my script
> public. (I'm not sure what the status of those plan is now.)
> 
> In the mean time I'll keep on checking linux-next with my script. That
> should also help to see what problems an actual bot could run into, for
> the benefit of those people. 

That sounds good; it's a very useful service you're providing.

> Problems like:
> - should it check each commit (I think it should only check releases);
> - should it be a true bot (I think it needs a wetware frontend to
>   filter out non-issues).
> 
> Hope this answers your question.

It does.  Thanks for the detailed response.


- Paul