| Message ID | 20191202084924.29893-1-tiny.windzz@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | xen/arm: Basic support for sunxi/sun50i h6 platform. | expand |
(+Andre) Hi, @Andre, IIRC you originally added the support for sunxi in Xen. Could you have a look at this patch? Cheers, On 02/12/2019 08:49, Yangtao Li wrote: > adding compatible strings for h6 SoCs, Specifically orangepi3. > > Signed-off-by: Yangtao Li <tiny.windzz@gmail.com> > --- > xen/arch/arm/platforms/sunxi.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/xen/arch/arm/platforms/sunxi.c b/xen/arch/arm/platforms/sunxi.c > index 55705b15b2..e8e4d88bef 100644 > --- a/xen/arch/arm/platforms/sunxi.c > +++ b/xen/arch/arm/platforms/sunxi.c > @@ -119,6 +119,7 @@ static const char * const sunxi_v8_dt_compat[] __initconst = > { > "allwinner,sun50i-a64", > "allwinner,sun50i-h5", > + "allwinner,sun50i-h6", > NULL > }; > >
On Tue, 3 Dec 2019 11:39:58 +0000 Julien Grall <julien@xen.org> wrote: Hi, > (+Andre) > > Hi, > > @Andre, IIRC you originally added the support for sunxi in Xen. Could > you have a look at this patch? Looks alright, and indeed the H6 needs it. Even though Allwinner totally re-arranged the memory map, they missed the opportunity to put each device at least in their own 4K page. Reviewed-by: Andre Przywara <andre.przywara@arm.com> If you can wait till this evening, I can even test it. It's actually a shame that we need this enumeration, when all we are after is an answer to the question: Does a device used by Xen share a 4K page with a device handed off to Dom0? It sounds like a nice rainy afternoon exercise to scan the DT to find those devices automatically and mask them (on the A64 for instance UART4 is on a different page). Cheers, Andre > On 02/12/2019 08:49, Yangtao Li wrote: > > adding compatible strings for h6 SoCs, Specifically orangepi3. > > > > Signed-off-by: Yangtao Li <tiny.windzz@gmail.com> > > --- > xen/arch/arm/platforms/sunxi.c | 1 + > > 1 file changed, 1 insertion(+) > > > > diff --git a/xen/arch/arm/platforms/sunxi.c b/xen/arch/arm/platforms/sunxi.c > > index 55705b15b2..e8e4d88bef 100644 > > --- a/xen/arch/arm/platforms/sunxi.c > > +++ b/xen/arch/arm/platforms/sunxi.c > > @@ -119,6 +119,7 @@ static const char * const sunxi_v8_dt_compat[] __initconst = > > { > > "allwinner,sun50i-a64", > > "allwinner,sun50i-h5", > > + "allwinner,sun50i-h6", > > NULL > > }; > > > > >
On 03/12/2019 14:38, Andre Przywara wrote: > On Tue, 3 Dec 2019 11:39:58 +0000 > Julien Grall <julien@xen.org> wrote: > > Hi, > >> (+Andre) >> >> Hi, >> >> @Andre, IIRC you originally added the support for sunxi in Xen. Could >> you have a look at this patch? > > Looks alright, and indeed the H6 needs it. Even though Allwinner totally re-arranged the memory map, they missed the opportunity to put each device at least in their own 4K page. > > Reviewed-by: Andre Przywara <andre.przywara@arm.com> Thank you for the review! > > If you can wait till this evening, I can even test it. I can wait until tomorrow before comitting the patch. > > It's actually a shame that we need this enumeration, when all we are after is an answer to the question: Does a device used by Xen share a 4K page with a device handed off to Dom0? It sounds like a nice rainy afternoon exercise to scan the DT to find those devices automatically and mask them (on the A64 for instance UART4 is on a different page). I agree and I think we discussed about it before :). I would welcome such improvement in Xen, this would actually allow us to drop sunxi.c completely. Cheers,
On Tue, 3 Dec 2019 16:52:45 +0000 Julien Grall <julien@xen.org> wrote: Hi, > On 03/12/2019 14:38, Andre Przywara wrote: > > On Tue, 3 Dec 2019 11:39:58 +0000 > > Julien Grall <julien@xen.org> wrote: > > > > Hi, > > > >> (+Andre) > >> > >> Hi, > >> > >> @Andre, IIRC you originally added the support for sunxi in Xen. Could > >> you have a look at this patch? > > > > Looks alright, and indeed the H6 needs it. Even though Allwinner totally re-arranged the memory map, they missed the opportunity to put each device at least in their own 4K page. > > > > Reviewed-by: Andre Przywara <andre.przywara@arm.com> > > Thank you for the review! > > > > > If you can wait till this evening, I can even test it. > > I can wait until tomorrow before comitting the patch. I booted Xen 4.12.1 to the Dom0 prompt on the Pine H64, and saw all the serial devices in /sys/firmware/devicetree/base/soc. Then I applied the patch, and the serial devices were gone. And yes, all the four main serial ports share one 4K page on the H6. Tested-by: Andre Przywara <andre.przywara@arm.com> Cheers, Andre. > > It's actually a shame that we need this enumeration, when all we are after is an answer to the question: Does a device used by Xen share a 4K page with a device handed off to Dom0? It sounds > like a nice rainy afternoon exercise to scan the DT to find those > devices automatically and mask them (on the A64 for instance UART4 is on > a different page). > > I agree and I think we discussed about it before :). I would welcome > such improvement in Xen, this would actually allow us to drop sunxi.c > completely. > > Cheers, >
Hi, On 04/12/2019 09:27, Andre Przywara wrote: > On Tue, 3 Dec 2019 16:52:45 +0000 > Julien Grall <julien@xen.org> wrote: > > Hi, > >> On 03/12/2019 14:38, Andre Przywara wrote: >>> On Tue, 3 Dec 2019 11:39:58 +0000 >>> Julien Grall <julien@xen.org> wrote: >>> >>> Hi, >>> >>>> (+Andre) >>>> >>>> Hi, >>>> >>>> @Andre, IIRC you originally added the support for sunxi in Xen. Could >>>> you have a look at this patch? >>> >>> Looks alright, and indeed the H6 needs it. Even though Allwinner totally re-arranged the memory map, they missed the opportunity to put each device at least in their own 4K page. >>> >>> Reviewed-by: Andre Przywara <andre.przywara@arm.com> >> >> Thank you for the review! >> >>> >>> If you can wait till this evening, I can even test it. >> >> I can wait until tomorrow before comitting the patch. > > I booted Xen 4.12.1 to the Dom0 prompt on the Pine H64, and saw all the serial devices in /sys/firmware/devicetree/base/soc. Then I applied the patch, and the serial devices were gone. And yes, all the four main serial ports share one 4K page on the H6. > > Tested-by: Andre Przywara <andre.przywara@arm.com> Thank you for the testing! Acked-by: Julien Grall <julien@xen.org> I have also committed it. Cheers,
diff --git a/xen/arch/arm/platforms/sunxi.c b/xen/arch/arm/platforms/sunxi.c index 55705b15b2..e8e4d88bef 100644 --- a/xen/arch/arm/platforms/sunxi.c +++ b/xen/arch/arm/platforms/sunxi.c @@ -119,6 +119,7 @@ static const char * const sunxi_v8_dt_compat[] __initconst = { "allwinner,sun50i-a64", "allwinner,sun50i-h5", + "allwinner,sun50i-h6", NULL };
adding compatible strings for h6 SoCs, Specifically orangepi3. Signed-off-by: Yangtao Li <tiny.windzz@gmail.com> --- xen/arch/arm/platforms/sunxi.c | 1 + 1 file changed, 1 insertion(+)