[PATCH] hw/arm/raspi4b: NOP all DTB nodes when removing unimplemented devices

[PATCH] hw/arm/raspi4b: NOP all DTB nodes when removing unimplemented devices

[Osama Abdelkader]

fdt_node_offset_by_compatible(fdt, -1, compat) only finds the first match.
If the blob has more than one node with the same compatible string, extra
nodes will remain active. using the same loop as imx8mp-evk.c

Signed-off-by: Osama Abdelkader <osama.abdelkader@gmail.com>
---
 hw/arm/raspi4b.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/hw/arm/raspi4b.c b/hw/arm/raspi4b.c
index 3eeb8f447e..06aeb8db01 100644
--- a/hw/arm/raspi4b.c
+++ b/hw/arm/raspi4b.c
@@ -72,12 +72,14 @@ static void raspi4_modify_dtb(const struct arm_boot_info *info, void *fdt)
 
     for (int i = 0; i < ARRAY_SIZE(nodes_to_remove); i++) {
         const char *dev_str = nodes_to_remove[i];
+        int offset;
 
-        int offset = fdt_node_offset_by_compatible(fdt, -1, dev_str);
-        if (offset >= 0) {
-            if (!fdt_nop_node(fdt, offset)) {
-                warn_report("bcm2711 dtc: %s has been disabled!", dev_str);
+        offset = fdt_node_offset_by_compatible(fdt, -1, dev_str);
+        while (offset >= 0) {
+            if (fdt_nop_node(fdt, offset) == 0) {
+                warn_report("bcm2711 dtb: %s has been disabled!", dev_str);
             }
+            offset = fdt_node_offset_by_compatible(fdt, offset, dev_str);
         }
     }
 
-- 
2.43.0

Re: [PATCH] hw/arm/raspi4b: NOP all DTB nodes when removing unimplemented devices

[Peter Maydell]

On Mon, 20 Apr 2026 at 17:21, Osama Abdelkader
<osama.abdelkader@gmail.com> wrote:
>
> fdt_node_offset_by_compatible(fdt, -1, compat) only finds the first match.
> If the blob has more than one node with the same compatible string, extra
> nodes will remain active. using the same loop as imx8mp-evk.c
>
> Signed-off-by: Osama Abdelkader <osama.abdelkader@gmail.com>
> ---

I wondered about whether it would be worth abstracting this
out into a helper function, but it looks like at the moment
only this raspi code and the imx8mp-evk do this. If we get
more places we want this we might do that, but not worth
it for two.

Applied to target-arm.next, thanks.

-- PMM

Re: [PATCH] hw/arm/raspi4b: NOP all DTB nodes when removing unimplemented devices

[Philippe Mathieu-Daudé]

On 27/4/26 11:53, Peter Maydell wrote:
> On Mon, 20 Apr 2026 at 17:21, Osama Abdelkader
> <osama.abdelkader@gmail.com> wrote:
>>
>> fdt_node_offset_by_compatible(fdt, -1, compat) only finds the first match.
>> If the blob has more than one node with the same compatible string, extra
>> nodes will remain active. using the same loop as imx8mp-evk.c
>>
>> Signed-off-by: Osama Abdelkader <osama.abdelkader@gmail.com>
>> ---
> 
> I wondered about whether it would be worth abstracting this
> out into a helper function, but it looks like at the moment
> only this raspi code and the imx8mp-evk do this. If we get
> more places we want this we might do that, but not worth
> it for two.

IIRC you were first against the approach of "use a real world
dtb and patch it", recommending to "generate a dtb with what
we emulate"; and we merged that as 'temporary kludge until we
properly emulate the missing device'.

Re: [PATCH] hw/arm/raspi4b: NOP all DTB nodes when removing unimplemented devices

[Osama Abdelkader]

On Mon, Apr 27, 2026 at 01:07:00PM +0200, Philippe Mathieu-Daudé wrote:
> On 27/4/26 11:53, Peter Maydell wrote:
> > On Mon, 20 Apr 2026 at 17:21, Osama Abdelkader
> > <osama.abdelkader@gmail.com> wrote:
> > > 
> > > fdt_node_offset_by_compatible(fdt, -1, compat) only finds the first match.
> > > If the blob has more than one node with the same compatible string, extra
> > > nodes will remain active. using the same loop as imx8mp-evk.c
> > > 
> > > Signed-off-by: Osama Abdelkader <osama.abdelkader@gmail.com>
> > > ---
> > 
> > I wondered about whether it would be worth abstracting this
> > out into a helper function, but it looks like at the moment
> > only this raspi code and the imx8mp-evk do this. If we get
> > more places we want this we might do that, but not worth
> > it for two.
> 
> IIRC you were first against the approach of "use a real world
> dtb and patch it", recommending to "generate a dtb with what
> we emulate"; and we merged that as 'temporary kludge until we
> properly emulate the missing device'.

Thanks Peter and Philippe for the clarification, I would be happy to participate
in that solution too.

Best regards,
Osama