The Linux Kernel Mailing List
 help / color / mirror / Atom feed
From: John Hubbard <jhubbard@nvidia.com>
To: Danilo Krummrich <dakr@kernel.org>,
	Joel Fernandes <joel@joelfernandes.org>,
	Alexandre Courbot <acourbot@nvidia.com>
Cc: "Timur Tabi" <ttabi@nvidia.com>,
	"Alistair Popple" <apopple@nvidia.com>,
	"Eliot Courtney" <ecourtney@nvidia.com>,
	"Shashank Sharma" <shashanks@nvidia.com>,
	"Zhi Wang" <zhiw@nvidia.com>, "David Airlie" <airlied@gmail.com>,
	"Simona Vetter" <simona@ffwll.ch>,
	"Bjorn Helgaas" <bhelgaas@google.com>,
	"Miguel Ojeda" <ojeda@kernel.org>,
	"Alex Gaynor" <alex.gaynor@gmail.com>,
	"Boqun Feng" <boqun.feng@gmail.com>,
	"Gary Guo" <gary@garyguo.net>,
	"Björn Roy Baron" <bjorn3_gh@protonmail.com>,
	"Benno Lossin" <lossin@kernel.org>,
	"Andreas Hindborg" <a.hindborg@kernel.org>,
	"Alice Ryhl" <aliceryhl@google.com>,
	"Trevor Gross" <tmgross@umich.edu>,
	nova-gpu@lists.linux.dev, LKML <linux-kernel@vger.kernel.org>,
	"John Hubbard" <jhubbard@nvidia.com>,
	"Will Pierce" <wpierce@nvidia.com>
Subject: [PATCH 16/17] gpu: nova-core: add KUnit tests for the interrupt tree and HALs
Date: Fri,  7 Aug 2026 20:11:18 -0700	[thread overview]
Message-ID: <20260808031120.363869-17-jhubbard@nvidia.com> (raw)
In-Reply-To: <20260808031120.363869-1-jhubbard@nvidia.com>

Neither the per-architecture interrupt policy nor the vector
arithmetic touches hardware, so KUnit can cover both without a GPU.

Add three suites:

    * nova_core_gin_tree covers the leaf index bounds, the
    subtree-to-leaf mapping and its out-of-range filtering, the vector
    encoding, the masking of subtrees an architecture does not
    implement, and that every supported chipset implements the subtree
    carrying the GSP notification.

    * nova_core_gin_hal covers the tree size on each family, and the
    rearm method for each combination of family and interrupt type.

    * nova_core_falcon_hal covers the falcon retrigger gate. It is keyed
    on the architecture rather than the HAL, because GA100 shares the
    Turing HAL but does have the register.

Assisted-by: Cursor:claude-opus-5
Reviewed-by: Will Pierce <wpierce@nvidia.com>
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
---
 drivers/gpu/nova-core/falcon/hal.rs         |  24 ++++
 drivers/gpu/nova-core/irq/hal.rs            | 106 ++++++++++++++++-
 drivers/gpu/nova-core/irq/interrupt_tree.rs | 121 ++++++++++++++++++++
 3 files changed, 250 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/nova-core/falcon/hal.rs b/drivers/gpu/nova-core/falcon/hal.rs
index f0828b32aebb..6bff9fea1a79 100644
--- a/drivers/gpu/nova-core/falcon/hal.rs
+++ b/drivers/gpu/nova-core/falcon/hal.rs
@@ -107,3 +107,27 @@ pub(super) fn falcon_hal<E: FalconEngine + 'static>(
 
     Ok(hal)
 }
+
+#[kunit_tests(nova_core_falcon_hal)]
+mod tests {
+    use super::*;
+
+    /// Only Turing falcons lack the interrupt retrigger register. GA100 has it even though
+    /// [`falcon_hal`] gives GA100 the Turing HAL, which is why the gate is keyed on the
+    /// architecture instead.
+    #[test]
+    fn intr_retrigger_gate_per_arch() {
+        assert!(!has_intr_retrigger(Chipset::TU102));
+
+        for chipset in [
+            Chipset::GA100,
+            Chipset::GA102,
+            Chipset::AD102,
+            Chipset::GH100,
+            Chipset::GB100,
+            Chipset::GB202,
+        ] {
+            assert!(has_intr_retrigger(chipset));
+        }
+    }
+}
diff --git a/drivers/gpu/nova-core/irq/hal.rs b/drivers/gpu/nova-core/irq/hal.rs
index cf2d1aa080fa..1993e2ef5143 100644
--- a/drivers/gpu/nova-core/irq/hal.rs
+++ b/drivers/gpu/nova-core/irq/hal.rs
@@ -8,7 +8,8 @@
 
 use kernel::{
     io::Io,
-    pci::IrqType, //
+    pci::IrqType,
+    prelude::*, //
 };
 
 use crate::{
@@ -109,3 +110,106 @@ pub(super) fn cpu_interrupt_hal(chipset: Chipset) -> &'static dyn CpuInterruptHa
         }
     }
 }
+
+#[kunit_tests(nova_core_gin_hal)]
+mod tests {
+    use super::*;
+
+    use crate::gpu::Chipset;
+
+    /// Pre-Hopper parts have an 8-leaf tree, so 4 subtrees and `0x0f`.
+    #[test]
+    fn pre_hopper_tree_size() {
+        for chipset in [Chipset::TU102, Chipset::GA102, Chipset::AD102] {
+            let hal = cpu_interrupt_hal(chipset);
+            assert_eq!(hal.num_leaves(), 8);
+            assert_eq!(hal.implemented_subtrees(), 0x0f);
+        }
+    }
+
+    /// Hopper and later implement a 16-leaf tree, so 8 subtrees and `0xff`.
+    #[test]
+    fn hopper_plus_tree_size() {
+        for chipset in [Chipset::GH100, Chipset::GB100, Chipset::GB202] {
+            let hal = cpu_interrupt_hal(chipset);
+            assert_eq!(hal.num_leaves(), 16);
+            assert_eq!(hal.implemented_subtrees(), 0xff);
+        }
+    }
+
+    /// The implemented subtrees always number exactly `num_leaves / 2`, one per subtree.
+    #[test]
+    fn implemented_subtrees_matches_leaf_count() {
+        for chipset in [
+            Chipset::TU102,
+            Chipset::GA102,
+            Chipset::AD102,
+            Chipset::GH100,
+            Chipset::GB100,
+            Chipset::GB202,
+        ] {
+            let hal = cpu_interrupt_hal(chipset);
+            assert_eq!(
+                hal.implemented_subtrees().count_ones() as usize,
+                hal.num_leaves() / 2
+            );
+        }
+    }
+
+    /// Only pre-Hopper MSI rearms through the configuration-space mirror. MSI on Hopper and later
+    /// cycles the `TOP` enables of every serviced subtree.
+    #[test]
+    fn msi_rearm_method_per_arch() {
+        for chipset in [Chipset::TU102, Chipset::GA102, Chipset::AD102] {
+            let hal = cpu_interrupt_hal(chipset);
+            assert_eq!(
+                hal.pci_irq_rearm_method(IrqType::Msi),
+                Some(PciIrqRearmMethod::ConfigMirrorEoi)
+            );
+        }
+
+        for chipset in [Chipset::GH100, Chipset::GB100, Chipset::GB202] {
+            let hal = cpu_interrupt_hal(chipset);
+            assert_eq!(
+                hal.pci_irq_rearm_method(IrqType::Msi),
+                Some(PciIrqRearmMethod::TopEnableCycleServiced)
+            );
+        }
+    }
+
+    /// MSI-X gives each subtree its own table entry, so on every architecture its rearm cycles
+    /// only the subtree the handler serves.
+    #[test]
+    fn msix_rearms_one_subtree_on_every_arch() {
+        for chipset in [
+            Chipset::TU102,
+            Chipset::GA102,
+            Chipset::AD102,
+            Chipset::GH100,
+            Chipset::GB100,
+            Chipset::GB202,
+        ] {
+            let hal = cpu_interrupt_hal(chipset);
+            assert_eq!(
+                hal.pci_irq_rearm_method(IrqType::MsiX),
+                Some(PciIrqRearmMethod::TopEnableCycleSubtree)
+            );
+        }
+    }
+
+    /// `INTx` is level-triggered and needs no rearm write on any architecture.
+    #[test]
+    fn intx_needs_no_rearm() {
+        for chipset in [
+            Chipset::TU102,
+            Chipset::GA102,
+            Chipset::AD102,
+            Chipset::GH100,
+            Chipset::GB100,
+            Chipset::GB202,
+        ] {
+            let hal = cpu_interrupt_hal(chipset);
+            assert_eq!(hal.pci_irq_rearm_method(IrqType::Intx), None);
+        }
+    }
+}
diff --git a/drivers/gpu/nova-core/irq/interrupt_tree.rs b/drivers/gpu/nova-core/irq/interrupt_tree.rs
index f4f1494cddba..42e72fa8089e 100644
--- a/drivers/gpu/nova-core/irq/interrupt_tree.rs
+++ b/drivers/gpu/nova-core/irq/interrupt_tree.rs
@@ -302,3 +302,124 @@ pub(super) fn clear_vectors(&self, bar: Bar0<'_>, vectors: u32) {
         }
     }
 }
+
+#[kunit_tests(nova_core_gin_tree)]
+mod tests {
+    use super::*;
+
+    /// A leaf index is a `Bounded<usize, 4>`, so it accepts 0..=15 and rejects 16.
+    #[test]
+    fn leaf_index_bounds() {
+        assert!(LeafIndex::try_new(0).is_some());
+        assert!(LeafIndex::try_new(15).is_some());
+        assert!(LeafIndex::try_new(16).is_none());
+    }
+
+    /// Subtree `N` covers the two adjacent leaves `2N` and `2N + 1`.
+    #[test]
+    fn subtree_covers_two_adjacent_leaves() {
+        let tree = Tree {
+            num_leaves: 16,
+            serviced_subtrees: 0xff,
+            rearm_method: None,
+        };
+
+        for index in 0..8usize {
+            let mut leaves = Subtree { index }.iter_leaves(&tree);
+            assert_eq!(leaves.next().map(|leaf| leaf.index.get()), Some(index * 2));
+            assert_eq!(
+                leaves.next().map(|leaf| leaf.index.get()),
+                Some(index * 2 + 1)
+            );
+            assert!(leaves.next().is_none());
+        }
+    }
+
+    /// Leaves that fall outside the addressable range are filtered out, never panicking. The
+    /// filter is the [`LeafIndex`] bound, not the tree's leaf count, so this holds even on the
+    /// widest tree.
+    #[test]
+    fn subtree_leaves_out_of_range_are_filtered() {
+        let tree = Tree {
+            num_leaves: 16,
+            serviced_subtrees: 0xff,
+            rearm_method: None,
+        };
+
+        // Subtree 8 would cover leaves 16 and 17, both beyond the leaf index range.
+        assert!(Subtree { index: 8 }.iter_leaves(&tree).next().is_none());
+    }
+
+    /// The production [`vector_leaf_bit`] maps every vector to a `(leaf, bit)` pair, valid leaves
+    /// stay within [`LeafIndex`], and the fixed doorbell (129) and GSP (155) vectors land where
+    /// the handlers expect.
+    #[test]
+    fn vector_maps_to_leaf_and_bit() {
+        // Every vector of a 16-leaf tree maps to an addressable leaf and a bit in 0..32.
+        for vector in 0u32..(16 * 32) {
+            let (leaf, bit) = vector_leaf_bit(vector);
+
+            assert!(LeafIndex::try_new(leaf).is_some());
+            assert!(bit < 32);
+            assert_eq!(leaf as u32 * 32 + bit, vector);
+        }
+
+        // The fixed vectors the handlers rely on: CPU doorbell 129 and GSP notification 155, both
+        // in leaf 4, which is present on both the 8-leaf (pre-Hopper) and 16-leaf trees.
+        assert_eq!(vector_leaf_bit(129), (4, 1));
+        assert_eq!(vector_leaf_bit(155), (4, 27));
+        assert!(LeafIndex::try_new(vector_leaf_bit(155).0).is_some());
+
+        // The first vector beyond the 16-leaf tree lands in leaf 16, which is out of range.
+        assert!(LeafIndex::try_new(vector_leaf_bit(16 * 32).0).is_none());
+    }
+
+    /// [`vector_subtree_mask`] agrees with [`vector_leaf_bit`] on which subtree holds a vector,
+    /// and the doorbell (129) and GSP (155) vectors share one, so a single allocation and a single
+    /// enabled subtree serve both.
+    #[test]
+    fn vector_maps_to_subtree() {
+        for vector in 0u32..(16 * 32) {
+            let (leaf, _) = vector_leaf_bit(vector);
+
+            assert_eq!(vector_subtree_mask(vector), 1u32 << (leaf / 2));
+        }
+
+        assert_eq!(vector_subtree_mask(155), 1 << 2);
+        assert_eq!(vector_subtree_mask(129), vector_subtree_mask(155));
+    }
+
+    /// [`Tree::new`] drops subtrees the architecture does not implement, so a caller cannot enable
+    /// a `TOP` bit with no leaves behind it.
+    #[test]
+    fn tree_new_masks_unimplemented_subtrees() {
+        assert_eq!(
+            Tree::new(Chipset::TU102, IrqType::Msi, 0xff).serviced_subtrees,
+            0x0f
+        );
+        assert_eq!(
+            Tree::new(Chipset::GH100, IrqType::Msi, 0xff).serviced_subtrees,
+            0xff
+        );
+    }
+
+    /// Every supported chipset implements the subtree that carries the GSP notification.
+    #[test]
+    fn serviced_subtree_is_implemented_everywhere() {
+        let serviced = crate::irq::gsp::GSP_SUBTREE;
+
+        for chipset in [
+            Chipset::TU102,
+            Chipset::GA102,
+            Chipset::AD102,
+            Chipset::GH100,
+            Chipset::GB100,
+            Chipset::GB202,
+        ] {
+            assert_eq!(
+                serviced & !cpu_interrupt_hal(chipset).implemented_subtrees(),
+                0
+            );
+        }
+    }
+}
-- 
2.55.0


  parent reply	other threads:[~2026-08-08  3:11 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-08-08  3:11 [PATCH 00/17] nova-core: GPU interrupt support and GSP event delivery John Hubbard
2026-08-08  3:11 ` [PATCH 01/17] rust: sync: completion: add wait_for_completion_timeout() John Hubbard
     [not found]   ` <DKK2DM3VK6TF.3KBBWP7S4A8T1@nvidia.com>
2026-08-09 21:43     ` John Hubbard
2026-08-08  3:11 ` [PATCH 02/17] rust: pci: expose the whole interrupt vector allocation John Hubbard
2026-08-09 13:27   ` Danilo Krummrich
2026-08-08  3:11 ` [PATCH 03/17] rust: pci: expose the allocated interrupt type John Hubbard
2026-08-09 13:24   ` Danilo Krummrich
2026-08-09 21:42     ` John Hubbard
2026-08-08  3:11 ` [PATCH 04/17] gpu: nova-core: allocate PCI MSI vector during probe John Hubbard
2026-08-08  3:11 ` [PATCH 05/17] gpu: nova-core: add the GIN CPU interrupt tree and MSI EOI registers John Hubbard
2026-08-08  3:11 ` [PATCH 06/17] gpu: nova-core: add the GIN interrupt tree API John Hubbard
2026-08-08  3:11 ` [PATCH 07/17] gpu: nova-core: add the per-architecture GIN CPU interrupt HAL John Hubbard
2026-08-08  3:11 ` [PATCH 08/17] gpu: nova-core: allocate interrupt vectors for the serviced subtrees John Hubbard
2026-08-08  3:11 ` [PATCH 09/17] gpu: nova-core: add an interrupt delivery self-test John Hubbard
2026-08-08  3:11 ` [PATCH 10/17] gpu: nova-core: dispatch GSP events instead of discarding them John Hubbard
2026-08-08  3:11 ` [PATCH 11/17] gpu: nova-core: match GSP RPC replies by sequence, not just function John Hubbard
2026-08-08  3:11 ` [PATCH 12/17] gpu: nova-core: recover the GSP receive path from corrupt framing John Hubbard
2026-08-08  3:11 ` [PATCH 13/17] gpu: nova-core: bound a GSP wait by a single deadline John Hubbard
2026-08-08  3:11 ` [PATCH 14/17] gpu: nova-core: drive GSP events with the SWGEN0 interrupt John Hubbard
2026-08-08  3:11 ` [PATCH 15/17] gpu: nova-core: retrigger the GSP falcon and clear every latched cause John Hubbard
2026-08-08  3:11 ` John Hubbard [this message]
2026-08-08  3:11 ` [PATCH 17/17] gpu: nova-core: document the GIN interrupt controller and GSP events John Hubbard

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260808031120.363869-17-jhubbard@nvidia.com \
    --to=jhubbard@nvidia.com \
    --cc=a.hindborg@kernel.org \
    --cc=acourbot@nvidia.com \
    --cc=airlied@gmail.com \
    --cc=alex.gaynor@gmail.com \
    --cc=aliceryhl@google.com \
    --cc=apopple@nvidia.com \
    --cc=bhelgaas@google.com \
    --cc=bjorn3_gh@protonmail.com \
    --cc=boqun.feng@gmail.com \
    --cc=dakr@kernel.org \
    --cc=ecourtney@nvidia.com \
    --cc=gary@garyguo.net \
    --cc=joel@joelfernandes.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lossin@kernel.org \
    --cc=nova-gpu@lists.linux.dev \
    --cc=ojeda@kernel.org \
    --cc=shashanks@nvidia.com \
    --cc=simona@ffwll.ch \
    --cc=tmgross@umich.edu \
    --cc=ttabi@nvidia.com \
    --cc=wpierce@nvidia.com \
    --cc=zhiw@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox