The Linux Kernel Mailing List
 help / color / mirror / Atom feed
* [PATCH 0/1] gpu: nova-core: populate GSP_INIT timeout defaults
@ 2026-11-17  6:10 Alok Kumar
  2026-11-17  6:10 ` [PATCH 1/1] gpu: nova-core: send GPU timeout defaults in GSP_INIT Alok Kumar
  0 siblings, 1 reply; 2+ messages in thread
From: Alok Kumar @ 2026-11-17  6:10 UTC (permalink / raw)
  To: dakr, acourbot, kvm, rust-for-linux, nova-gpu, linux-pci
  Cc: alex, jgg, yishaih, skolothumtho, kevin.tian, airlied, simona,
	ojeda, alex.gaynor, boqun.feng, gary, bjorn3_gh, lossin,
	a.hindborg, aliceryhl, tmgross, markus.probst, jhubbard, apopple,
	joelagnelf, ecourtney, cjia, smitra, kjaju, alkumar, ankita,
	aniketa, kwankhede, targupta, linux-kernel, zhiwang, zhiw, boqun,
	daniel.almeida, tamird, work, jgg, bhelgaas, kwilczynski,
	dri-devel, peter, xinmengs

Populate the GSP_INIT request with nonzero default timeouts for vGPU.

The patch encodes the following keys as u32 values in microseconds:

  0xA0: vGPU timeout             1,800,000 us
  0xA1: graphics-mode timeout    4,000,000 us
  0xA2: compute-mode timeout    30,000,000 us

This series is based on:
    [PATCH 0/5] Transition nova-core to use TLV variables for heap sizes [1]

The complete integration tree, including the prerequisites and the
companion VFIO driver, is available at:

    https://github.com/alokk-nvidia/nova-core/tree/nova-vgpu-wip-20261113

[1] https://lore.kernel.org/nova-gpu/f1f132172061996beb4b5d64ebc68e4a6bb0a1e2.camel@nvidia.com/

Alok Kumar (1):
  gpu: nova-core: send GPU timeout defaults in GSP_INIT

 drivers/gpu/nova-core/gsp/fw/commands.rs | 11 +++++++++++
 1 file changed, 11 insertions(+)


base-commit: d1046d643c171875d6d350dcacbee89f56c7272a
-- 
2.53.0


^ permalink raw reply	[flat|nested] 2+ messages in thread

* [PATCH 1/1] gpu: nova-core: send GPU timeout defaults in GSP_INIT
  2026-11-17  6:10 [PATCH 0/1] gpu: nova-core: populate GSP_INIT timeout defaults Alok Kumar
@ 2026-11-17  6:10 ` Alok Kumar
  0 siblings, 0 replies; 2+ messages in thread
From: Alok Kumar @ 2026-11-17  6:10 UTC (permalink / raw)
  To: dakr, acourbot, kvm, rust-for-linux, nova-gpu, linux-pci
  Cc: alex, jgg, yishaih, skolothumtho, kevin.tian, airlied, simona,
	ojeda, alex.gaynor, boqun.feng, gary, bjorn3_gh, lossin,
	a.hindborg, aliceryhl, tmgross, markus.probst, jhubbard, apopple,
	joelagnelf, ecourtney, cjia, smitra, kjaju, alkumar, ankita,
	aniketa, kwankhede, targupta, linux-kernel, zhiwang, zhiw, boqun,
	daniel.almeida, tamird, work, jgg, bhelgaas, kwilczynski,
	dri-devel, peter, xinmengs

GSP firmware requires a nonzero vGPU timeout in GSP_INIT.

Populate the NVKV keys as u32 values:

  0xA0: vGPU timeout,          1,800,000 us
  0xA1: graphics-mode timeout, 4,000,000 us
  0xA2: compute-mode timeout,  30,000,000 us

Signed-off-by: Alok Kumar <alkumar@nvidia.com>
---
 drivers/gpu/nova-core/gsp/fw/commands.rs | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/drivers/gpu/nova-core/gsp/fw/commands.rs b/drivers/gpu/nova-core/gsp/fw/commands.rs
index 08798f150ad7..adb0b1f29ceb 100644
--- a/drivers/gpu/nova-core/gsp/fw/commands.rs
+++ b/drivers/gpu/nova-core/gsp/fw/commands.rs
@@ -192,6 +192,9 @@ pub(crate) struct GspInitRequest {
         pci_config_mirror_base: Key<u32, 0x0010>,
         pci_config_mirror_size: Key<u32, 0x0011>,
         host_arch: Key<HostArch, 0x0070, u32>,
+        default_timeout_us_vgpu: Key<u32, 0x00a0>,
+        default_timeout_us_graphics_mode: Key<u32, 0x00a1>,
+        default_timeout_us_compute_mode: Key<u32, 0x00a2>,
         bus_device_func: Key<u64, 0x1020>,
         regkeys: KVVec<RegKey>,
         vf_info: Option<VfInfo>,
@@ -230,6 +233,11 @@ struct DomainBusDevice(u64) {
 ];
 
 impl GspInitRequest {
+    // GPU timeout defaults in microseconds. All three values must be nonzero in GSP_INIT.
+    const DEFAULT_TIMEOUT_US_VGPU: u32 = 1_800_000;
+    const DEFAULT_TIMEOUT_US_GRAPHICS_MODE: u32 = 4_000_000;
+    const DEFAULT_TIMEOUT_US_COMPUTE_MODE: u32 = 30_000_000;
+
     fn build_vf_info(ctx: &GspBootContext<'_, '_>) -> Result<Option<VfInfo>> {
         let total_vfs = match *ctx.vgpu_state {
             VgpuState::Disabled => return Ok(None),
@@ -300,6 +308,9 @@ pub(crate) fn new(ctx: &GspBootContext<'_, '_>) -> Result<Self> {
             pci_config_mirror_base: mirror.start.into(),
             pci_config_mirror_size: (mirror.end - mirror.start).into(),
             host_arch: HostArch::host().into(),
+            default_timeout_us_vgpu: Self::DEFAULT_TIMEOUT_US_VGPU.into(),
+            default_timeout_us_graphics_mode: Self::DEFAULT_TIMEOUT_US_GRAPHICS_MODE.into(),
+            default_timeout_us_compute_mode: Self::DEFAULT_TIMEOUT_US_COMPUTE_MODE.into(),
             bus_device_func: u64::from(bus_device_func).into(),
             regkeys,
             vf_info: Self::build_vf_info(ctx)?,
-- 
2.53.0


^ permalink raw reply related	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-11-17  6:13 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-17  6:10 [PATCH 0/1] gpu: nova-core: populate GSP_INIT timeout defaults Alok Kumar
2026-11-17  6:10 ` [PATCH 1/1] gpu: nova-core: send GPU timeout defaults in GSP_INIT Alok Kumar

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox