[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Minios-devel] [PATCH 00/40] MINI-OS: enable the arm64 support



On Wed, Nov 08, 2017 at 09:40:38AM +0000, Julien Grall wrote:
> 
> 
> On 08/11/17 06:01, Huang Shijie wrote:
> > On Tue, Nov 07, 2017 at 12:31:30PM +0000, Julien Grall wrote:
> > > Hi Wei,
> > > 
> > > On 07/11/17 12:09, Wei Liu wrote:
> > > > On Fri, Nov 03, 2017 at 11:49:16AM +0000, Julien Grall wrote:
> > > > > I can see two solutions going forward:
> > > > >       1) The arm directory is first reshaped to welcome arm64. This 
> > > > > means:
> > > > >               * moving out arm32 specific code
> > > > >               * switch to LPAE page-table
> > > > >               * introducing helpers for common code to call 
> > > > > arch-specific code
> > > > >          On the code is reshaped, the arm64 series is added on top.
> > > > > 
> > > > >       2) Start the arm64 port from a clean slate and then port arm32 
> > > > > over.
> > > > > 
> > > > > Knowing the state of the arm32 port, I would lean towards 2). This 
> > > > > would
> > > > > allow more flexibility and make easier to review. At the moment, I 
> > > > > have to
> > > > > hunt down the code to see what is missing.
> > > > > 
> > > > > I would be interested to hear the opinion of the maintainers here.
> > > > 
> > > > If you're sure arm32 can't work, #2 is probably easier.  Please stick a
> > > > patch at the beginning to rip out the old port. That can easily be
> > > > applied.
> > > 
> > > arch.mk is inexistent for the arm32 port. Also looking at the series 
> > > here, I
> > > noticed quite a few patches that were meant to be Arm64 only were fixing
> > > Arm32 port to.
> > > 
> > > But aside the compilation issues, I don't think the Arm32 port is in good
> > > shape. Looking back to the Mini-OS ML archive, then port was indeed never
> > > finished (a couple of patches where still present).
> > > 
> > > It looks like MirageOS has a Arm32 port based on the series on the ML. But
> > > they never upstreamed changes. I briefly looked at it and I think my 
> > > points
> > > at in my previous e-mail stands and the code is more in a hacking state.
> > > 
> > > So I agree with rip out the old port. I think it would be easy to add 
> > > Arm32
> > > if the Arm64 port has been correctly written. But I am not sure Shijie is
> > > planning to do the Arm32 port?
> > Hi Julien,
> >      I am not sure I have enough time to do the arm32 port..
> > > 
> > > > 
> > > > Also may I suggest the new port follow $arch/$subarch hierarchy? Just
> > > > like what we do in Xen.
> > > +1 here.
> > Do you mean we should place all the arm64 code in arch/arm64, not in
> > arch/arm/arm64?
> No, by $arch/$subarch Wei meant that any arm64 code should be under
> arch/arm/arm64. You can substitute $arch with arm and $subarch with arm64.
I think it is okay to write it in clean slate.
So can I put the new gic.c event.c(may partially be copied from arch/arm/
folder ) ... at the arch/arm/arm64/ folder?

Another solution is like the linux kernel, use the arch/arm64 folder to
store all the arm64 code.

Thanks
Huang Shijie

_______________________________________________
Minios-devel mailing list
Minios-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/cgi-bin/mailman/listinfo/minios-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.