Re: [v5.4 stable] arm: stm32: Regression observed on "no-map" reserved memory region

Hi Quentin

On 4/22/21 2:59 PM, Quentin Perret wrote:

On Wednesday 21 Apr 2021 at 07:33:52 (-0700), Florian Fainelli wrote:

quoted

It is not, otherwise I would have noticed earlier, can you try the same
thing that happens on my platform with a reserved region (without
no-map) adjacent to a reserved region with 'no-map'?

I just tried, but still no luck. FTR, I tried to reproduce your setup
with the following DT:

         memory@40000000 {
                 reg = <0x00 0x40000000 0x01 0x00>;
                 device_type = "memory";
         };

         reserved-memory {
                 #address-cells = <2>;
                 #size-cells = <2>;
                 ranges;

                 foo@fdfff000{
                         reg = <0x00 0xfdfff000 0x0 0x1000>;
                 };
                 bar@fe000000{
                         reg = <0x00 0xfe000000 0x0 0x2000000>;
                         no-map;
                 };
         };

And with 5.4.102 and 5.10.31 I get the following in /proc/iomem

<...>
40000000-fdffffff : System RAM
   40080000-412cffff : Kernel code
   412d0000-417affff : reserved
   417b0000-419f8fff : Kernel data
   48000000-48008fff : reserved
   f7c00000-fdbfffff : reserved
   fdfff000-fdffffff : reserved
fe000000-ffffffff : reserved
100000000-13fffffff : System RAM
<...>

which looks about right. I'll keep trying a few other things.

Did you get time to continue some tests on this issue ?

On my side this DT is not working:

memory@c0000000 {
         reg = <0xc0000000 0x20000000>;
};

reserved-memory {
         #address-cells = <1>;
         #size-cells = <1>;
         ranges;

         gpu_reserved: gpu@d4000000 {
                 reg = <0xd4000000 0x4000000>;
                 no-map;
         };
};

Let me know if I can help.

regards
Alex

Thanks,
Quentin