Odd (and incorrect) code generation from SVD

I was in the process of updating a project that I am working on to new SVD's and when I generated the PAC I got some strange code. This first block is and excerpt from the relevant part of the SVD. The full thing can be found here:

<field>
    <name>CTS_ASSERT</name>
    <description>CTS assert level</description>
    <bitOffset>7</bitOffset>
    <bitWidth>1</bitWidth>
    <enumeratedValues>
        <name>ASSERT_LEVEL</name>
        <enumeratedValue>
            <name>HIGH</name>
            <description>Assert high</description>
            <value>0</value>
        </enumeratedValue>
        <enumeratedValue>
            <name>LOW</name>
            <description>Assert low</description>
            <value>1</value>
        </enumeratedValue>
    </enumeratedValues>
</field>
<field>
    <name>RTS_ASSERT</name>
    <description>RTS assert level</description>
    <bitOffset>6</bitOffset>
    <bitWidth>1</bitWidth>
    <enumeratedValues derivedFrom="UART1.CNTL.CTS_ASSERT.ASSERT_LEVEL"></enumeratedValues>
</field>

Generated code is as follows:

#[doc = "RTS assert level"]
pub use AssertLevel;
#[doc = "Field `RTS_ASSERT` reader - RTS assert level"]
pub use CtsAssertR as RtsAssertR;
#[doc = "Field `RTS_ASSERT` writer - RTS assert level"]
pub use CtsAssertW as RtsAssertW;
#[doc = "CTS assert level\n\nValue on reset: 0"]
#[cfg_attr(feature = "defmt", derive(defmt::Format))]
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
pub enum AssertLevel {
    #[doc = "0: Assert high"]
    High = 0,
    #[doc = "1: Assert low"]
    Low = 1,
}
impl From<AssertLevel> for bool {
    #[inline(always)]
    fn from(variant: AssertLevel) -> Self {
        variant as u8 != 0
    }
}
#[doc = "Field `CTS_ASSERT` reader - CTS assert level"]
pub type CtsAssertR = crate::BitReader<AssertLevel>;
impl CtsAssertR {
    #[doc = "Get enumerated values variant"]
    #[inline(always)]
    pub const fn variant(&self) -> AssertLevel {
        match self.bits {
            false => AssertLevel::High,
            true => AssertLevel::Low,
        }
    }
    #[doc = "Assert high"]
    #[inline(always)]
    pub fn is_high(&self) -> bool {
        *self == AssertLevel::High
    }
    #[doc = "Assert low"]
    #[inline(always)]
    pub fn is_low(&self) -> bool {
        *self == AssertLevel::Low
    }
}
#[doc = "Field `CTS_ASSERT` writer - CTS assert level"]
pub type CtsAssertW<'a, REG> = crate::BitWriter<'a, REG, AssertLevel>;
impl<'a, REG> CtsAssertW<'a, REG>
where
    REG: crate::Writable + crate::RegisterSpec,
{
    #[doc = "Assert high"]
    #[inline(always)]
    pub fn high(self) -> &'a mut crate::W<REG> {
        self.variant(AssertLevel::High)
    }
    #[doc = "Assert low"]
    #[inline(always)]
    pub fn low(self) -> &'a mut crate::W<REG> {
        self.variant(AssertLevel::Low)
    }
}

The error here is that for some reason a pub use AssertLevel; was generated just before pub enum AssertLevel {...}. I am using svd2rust version 0.37.1 and rust version 1.92.0. I am running Ubuntu 24.04.

As to what the actual issue here is I'm not sure. However it is unlikely to be the SVD because that specific file has been the same for years.

Let me know if anymore information is needed.

You are right. It looks like a bug and should be easy to fix.

But I understand why it was not found earlier. Usually we use relative derive path instead of absolute in such cases.

Oh. No. It looks like there is other issue.

Due to this sort:

svd2rust/src/generate/register.rs

Line 631 in b0f2715

fields.sort_by_key(|f| f.bit_offset());

RTS_ASSERT is being processed before CTS_ASSERT
This leads to this hashset

svd2rust/src/generate/register.rs

Line 762 in b0f2715

let derives;

is empty when pub use AssertLevel; is generated.

Fast fix from SVD side is to derive CTS_ASSERT from RTS_ASSERT instead of RTS_ASSERT from CTS_ASSERT.

The fix from svd2rust side is to move filling of derives hashset out of main cycle.

I was wrong. Only deleting sort by offset could fix your issue from svd2rust side. But I afraid it can lead to more bugs.

Sorry I've been away without internet for the past few days. Using the branch and using a the reverse flag works perfectly! Thank you so much!

@burrbull

@burrbull
Ok I'm getting the same issue now but in another part of the code:

#[doc = "Resistor select for 15"]
pub use BpPull;
...
#[doc = "Resistor select for 0"]
#[cfg_attr(feature = "defmt", derive(defmt::Format))]
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
#[repr(u8)]
pub enum BpPull {
    #[doc = "0: No pull"]
    None = 0,
    #[doc = "1: Pull up"]
    Up = 1,
    #[doc = "2: Pull down"]
    Down = 2,
}

SVD:

<register>
    <name>GPIO_PUP_PDN_CNTRL_REG0</name>
    <description>GPIO Pull-up / Pull-down Register 0</description>
    <addressOffset>0xe4</addressOffset>
    <access>read-write</access>
    <fields>
        <field>
            <name>GPIO_PUP_PDN_CNTRL0</name>
            <description>Resistor select for 0</description>
            <bitOffset>0</bitOffset>
            <bitWidth>2</bitWidth>
            <enumeratedValues>
                <name>BP_PULL</name>
                <usage>read-write</usage>
                <headerEnumName>BP_PULL</headerEnumName>
                <enumeratedValue>
                    <name>NONE</name>
                    <description>No pull</description>
                    <value>0</value>
                </enumeratedValue>
                <enumeratedValue>
                    <name>UP</name>
                    <description>Pull up</description>
                    <value>1</value>
                </enumeratedValue>
                <enumeratedValue>
                    <name>DOWN</name>
                    <description>Pull down</description>
                    <value>2</value>
                </enumeratedValue>
            </enumeratedValues>
        </field>
        {% for i in range(1, 16) %}
        <field>
            <name>GPIO_PUP_PDN_CNTRL{{ i }}</name>
            <description>Resistor select for {{ i }}</description>
            <bitOffset>{{ 2 * i }}</bitOffset>
            <bitWidth>2</bitWidth>
            <enumeratedValues derivedFrom="GPIO.GPIO_PUP_PDN_CNTRL_REG0.GPIO_PUP_PDN_CNTRL0.BP_PULL"/>
        </field>
        {% endfor %}
    </fields>
</register>

I've continued testing and in 0.36.0 I get the same error. Not sure about anything lower.

Navigation Menu

Repository navigation

Odd (and incorrect) code generation from SVD #966

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Odd (and incorrect) code generation from SVD #966

Description

Activity

burrbull commented on Aug 23, 2026

burrbull commented on Aug 23, 2026

burrbull commented on Aug 23, 2026

burrbull commented on Aug 23, 2026

MsFormula commented on Aug 26, 2026

MsFormula commented on Sep 1, 2026

MsFormula commented on Sep 6, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions