[RFC PATCH v2 4/5] rust/hw/core: Provide some interfaces for the GPIO device

[RFC PATCH v2 4/5] rust/hw/core: Provide some interfaces for the GPIO device

[chenmiao]

In irq.rs, we added a new get method for the InterruptSource type to determine
whether an InterruptSource is null. This eliminates the need to repeatedly
call self.cell.get().is_null() for null checks during comparisons.
Additionally, we exposed the slice_as_ptrmethod to support external usage with
the &[InterruptSource]type.

In qdev.rs, we implemented the init_gpio_out_namedfunction, which corresponds
to the C function qdev_init_gpio_out_named. We also refactored the
init_gpio_outfunction to reuse the init_gpio_out_namedinterface.

Signed-off-by: chenmiao <chenmiao@openatom.club>
---
 rust/hw/core/src/irq.rs  |  6 +++++-
 rust/hw/core/src/qdev.rs | 12 +++++++++---
 2 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/rust/hw/core/src/irq.rs b/rust/hw/core/src/irq.rs
index e0d7784d97..dd5d0cadbc 100644
--- a/rust/hw/core/src/irq.rs
+++ b/rust/hw/core/src/irq.rs
@@ -71,6 +71,10 @@ pub fn pulse(&self) {
     pub fn raise(&self) {
         self.set(true);
     }
+
+    pub fn get(&self) -> bool {
+        !self.cell.get().is_null()
+    }
 }
 
 impl<T> InterruptSource<T>
@@ -91,7 +95,7 @@ pub(crate) const fn as_ptr(&self) -> *mut *mut bindings::IRQState {
         self.cell.as_ptr()
     }
 
-    pub(crate) const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
+    pub const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
         assert!(!slice.is_empty());
         slice[0].as_ptr()
     }
diff --git a/rust/hw/core/src/qdev.rs b/rust/hw/core/src/qdev.rs
index c3097a284d..1d7a0b7c1a 100644
--- a/rust/hw/core/src/qdev.rs
+++ b/rust/hw/core/src/qdev.rs
@@ -17,7 +17,7 @@
 
 pub use crate::bindings::{ClockEvent, DeviceClass, Property, ResetType};
 use crate::{
-    bindings::{self, qdev_init_gpio_in, qdev_init_gpio_out, ResettableClass},
+    bindings::{self, qdev_init_gpio_in, qdev_init_gpio_out_named, ResettableClass},
     irq::InterruptSource,
 };
 
@@ -399,11 +399,17 @@ fn do_init_gpio_in(
     }
 
     fn init_gpio_out(&self, pins: &[InterruptSource]) {
+        self.init_gpio_out_named(pins, "", pins.len());
+    }
+
+    fn init_gpio_out_named(&self, pins: &[InterruptSource], name: &str, n: usize) {
+        let c_name = CString::new(name).expect("gpio name new failed");
         unsafe {
-            qdev_init_gpio_out(
+            qdev_init_gpio_out_named(
                 self.upcast().as_mut_ptr(),
                 InterruptSource::slice_as_ptr(pins),
-                pins.len() as c_int,
+                c_name.as_ptr(),
+                n as c_int,
             );
         }
     }
-- 
2.43.0

Re: [RFC PATCH v2 4/5] rust/hw/core: Provide some interfaces for the GPIO device

[Paolo Bonzini]

On Tue, Oct 28, 2025 at 11:18 AM chenmiao <chenmiao@openatom.club> wrote:
>
> In irq.rs, we added a new get method for the InterruptSource type to determine
> whether an InterruptSource is null. This eliminates the need to repeatedly
> call self.cell.get().is_null() for null checks during comparisons.
> Additionally, we exposed the slice_as_ptrmethod to support external usage with
> the &[InterruptSource]type.
>
> In qdev.rs, we implemented the init_gpio_out_namedfunction, which corresponds
> to the C function qdev_init_gpio_out_named. We also refactored the
> init_gpio_outfunction to reuse the init_gpio_out_namedinterface.
>
> Signed-off-by: chenmiao <chenmiao@openatom.club>
> ---
>  rust/hw/core/src/irq.rs  |  6 +++++-
>  rust/hw/core/src/qdev.rs | 12 +++++++++---
>  2 files changed, 14 insertions(+), 4 deletions(-)
>
> diff --git a/rust/hw/core/src/irq.rs b/rust/hw/core/src/irq.rs
> index e0d7784d97..dd5d0cadbc 100644
> --- a/rust/hw/core/src/irq.rs
> +++ b/rust/hw/core/src/irq.rs
> @@ -71,6 +71,10 @@ pub fn pulse(&self) {
>      pub fn raise(&self) {
>          self.set(true);
>      }
> +
> +    pub fn get(&self) -> bool {
> +        !self.cell.get().is_null()
> +    }

This should not be get(), but "is_connected()". Also it should be
implemented for any T, therefore in the "impl<T> InterruptSource<T>"
block below.

>  }
>
>  impl<T> InterruptSource<T>
> @@ -91,7 +95,7 @@ pub(crate) const fn as_ptr(&self) -> *mut *mut bindings::IRQState {
>          self.cell.as_ptr()
>      }
>
> -    pub(crate) const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
> +    pub const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
>          assert!(!slice.is_empty());
>          slice[0].as_ptr()
>      }

Since you are not using this, you don't need to expose it outside the crate.

Paolo

Re: [RFC PATCH v2 4/5] rust/hw/core: Provide some interfaces for the GPIO device

[Chen Miao]

On 10/28/2025 6:49 PM, Paolo Bonzini wrote:
> On Tue, Oct 28, 2025 at 11:18 AM chenmiao <chenmiao@openatom.club> wrote:
>> In irq.rs, we added a new get method for the InterruptSource type to determine
>> whether an InterruptSource is null. This eliminates the need to repeatedly
>> call self.cell.get().is_null() for null checks during comparisons.
>> Additionally, we exposed the slice_as_ptrmethod to support external usage with
>> the &[InterruptSource]type.
>>
>> In qdev.rs, we implemented the init_gpio_out_namedfunction, which corresponds
>> to the C function qdev_init_gpio_out_named. We also refactored the
>> init_gpio_outfunction to reuse the init_gpio_out_namedinterface.
>>
>> Signed-off-by: chenmiao <chenmiao@openatom.club>
>> ---
>>   rust/hw/core/src/irq.rs  |  6 +++++-
>>   rust/hw/core/src/qdev.rs | 12 +++++++++---
>>   2 files changed, 14 insertions(+), 4 deletions(-)
>>
>> diff --git a/rust/hw/core/src/irq.rs b/rust/hw/core/src/irq.rs
>> index e0d7784d97..dd5d0cadbc 100644
>> --- a/rust/hw/core/src/irq.rs
>> +++ b/rust/hw/core/src/irq.rs
>> @@ -71,6 +71,10 @@ pub fn pulse(&self) {
>>       pub fn raise(&self) {
>>           self.set(true);
>>       }
>> +
>> +    pub fn get(&self) -> bool {
>> +        !self.cell.get().is_null()
>> +    }
> This should not be get(), but "is_connected()". Also it should be
> implemented for any T, therefore in the "impl<T> InterruptSource<T>"
> block below.
I'll fix it later.
>>   }
>>
>>   impl<T> InterruptSource<T>
>> @@ -91,7 +95,7 @@ pub(crate) const fn as_ptr(&self) -> *mut *mut bindings::IRQState {
>>           self.cell.as_ptr()
>>       }
>>
>> -    pub(crate) const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
>> +    pub const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
>>           assert!(!slice.is_empty());
>>           slice[0].as_ptr()
>>       }
> Since you are not using this, you don't need to expose it outside the crate.
>
> Paolo

I have used this function in the next patch.

Chen Miao

Re: [RFC PATCH v2 4/5] rust/hw/core: Provide some interfaces for the GPIO device

[Paolo Bonzini]

On Tue, Oct 28, 2025 at 12:24 PM Chen Miao <chenmiao@openatom.club> wrote:
>
> On 10/28/2025 6:49 PM, Paolo Bonzini wrote:
> > On Tue, Oct 28, 2025 at 11:18 AM chenmiao <chenmiao@openatom.club> wrote:
> >> In irq.rs, we added a new get method for the InterruptSource type to determine
> >> whether an InterruptSource is null. This eliminates the need to repeatedly
> >> call self.cell.get().is_null() for null checks during comparisons.
> >> Additionally, we exposed the slice_as_ptrmethod to support external usage with
> >> the &[InterruptSource]type.
> >>
> >> In qdev.rs, we implemented the init_gpio_out_namedfunction, which corresponds
> >> to the C function qdev_init_gpio_out_named. We also refactored the
> >> init_gpio_outfunction to reuse the init_gpio_out_namedinterface.
> >>
> >> Signed-off-by: chenmiao <chenmiao@openatom.club>
> >> ---
> >>   rust/hw/core/src/irq.rs  |  6 +++++-
> >>   rust/hw/core/src/qdev.rs | 12 +++++++++---
> >>   2 files changed, 14 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/rust/hw/core/src/irq.rs b/rust/hw/core/src/irq.rs
> >> index e0d7784d97..dd5d0cadbc 100644
> >> --- a/rust/hw/core/src/irq.rs
> >> +++ b/rust/hw/core/src/irq.rs
> >> @@ -71,6 +71,10 @@ pub fn pulse(&self) {
> >>       pub fn raise(&self) {
> >>           self.set(true);
> >>       }
> >> +
> >> +    pub fn get(&self) -> bool {
> >> +        !self.cell.get().is_null()
> >> +    }
> > This should not be get(), but "is_connected()". Also it should be
> > implemented for any T, therefore in the "impl<T> InterruptSource<T>"
> > block below.
> I'll fix it later.

You can drop it actually (see review of 5/5).

> >>   }
> >>
> >>   impl<T> InterruptSource<T>
> >> @@ -91,7 +95,7 @@ pub(crate) const fn as_ptr(&self) -> *mut *mut bindings::IRQState {
> >>           self.cell.as_ptr()
> >>       }
> >>
> >> -    pub(crate) const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
> >> +    pub const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
> >>           assert!(!slice.is_empty());
> >>           slice[0].as_ptr()
> >>       }
> > Since you are not using this, you don't need to expose it outside the crate.
>
> I have used this function in the next patch.

Isn't it commented out? The code that is used is:

+        self.init_gpio_in(self.handler_size(), PCF8574State::gpio_set);
+        self.init_gpio_out(from_ref(&self.handler[0]));
+        self.init_gpio_out_named(from_ref(&self.intrq), "nINT", 1);

Re: [RFC PATCH v2 4/5] rust/hw/core: Provide some interfaces for the GPIO device

[Chen Miao]

On 10/28/2025 7:30 PM, Paolo Bonzini wrote:
> On Tue, Oct 28, 2025 at 12:24 PM Chen Miao <chenmiao@openatom.club> wrote:
>> On 10/28/2025 6:49 PM, Paolo Bonzini wrote:
>>> On Tue, Oct 28, 2025 at 11:18 AM chenmiao <chenmiao@openatom.club> wrote:
>>>> In irq.rs, we added a new get method for the InterruptSource type to determine
>>>> whether an InterruptSource is null. This eliminates the need to repeatedly
>>>> call self.cell.get().is_null() for null checks during comparisons.
>>>> Additionally, we exposed the slice_as_ptrmethod to support external usage with
>>>> the &[InterruptSource]type.
>>>>
>>>> In qdev.rs, we implemented the init_gpio_out_namedfunction, which corresponds
>>>> to the C function qdev_init_gpio_out_named. We also refactored the
>>>> init_gpio_outfunction to reuse the init_gpio_out_namedinterface.
>>>>
>>>> Signed-off-by: chenmiao <chenmiao@openatom.club>
>>>> ---
>>>>    rust/hw/core/src/irq.rs  |  6 +++++-
>>>>    rust/hw/core/src/qdev.rs | 12 +++++++++---
>>>>    2 files changed, 14 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/rust/hw/core/src/irq.rs b/rust/hw/core/src/irq.rs
>>>> index e0d7784d97..dd5d0cadbc 100644
>>>> --- a/rust/hw/core/src/irq.rs
>>>> +++ b/rust/hw/core/src/irq.rs
>>>> @@ -71,6 +71,10 @@ pub fn pulse(&self) {
>>>>        pub fn raise(&self) {
>>>>            self.set(true);
>>>>        }
>>>> +
>>>> +    pub fn get(&self) -> bool {
>>>> +        !self.cell.get().is_null()
>>>> +    }
>>> This should not be get(), but "is_connected()". Also it should be
>>> implemented for any T, therefore in the "impl<T> InterruptSource<T>"
>>> block below.
>> I'll fix it later.
> You can drop it actually (see review of 5/5).
Ok!
>>>>    }
>>>>
>>>>    impl<T> InterruptSource<T>
>>>> @@ -91,7 +95,7 @@ pub(crate) const fn as_ptr(&self) -> *mut *mut bindings::IRQState {
>>>>            self.cell.as_ptr()
>>>>        }
>>>>
>>>> -    pub(crate) const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
>>>> +    pub const fn slice_as_ptr(slice: &[Self]) -> *mut *mut bindings::IRQState {
>>>>            assert!(!slice.is_empty());
>>>>            slice[0].as_ptr()
>>>>        }
>>> Since you are not using this, you don't need to expose it outside the crate.
>> I have used this function in the next patch.
> Isn't it commented out? The code that is used is:
>
> +        self.init_gpio_in(self.handler_size(), PCF8574State::gpio_set);
> +        self.init_gpio_out(from_ref(&self.handler[0]));
> +        self.init_gpio_out_named(from_ref(&self.intrq), "nINT", 1);

Oh, sorry, I was referring to my initial version. Initially, this function was 
used externally, but later I implemented init_gpio_out_named, and now it's 
internal again.

You're right, this function should not be exposed.

Chen Miao

>