Skip to content

[26.04_linux-nvidia-bos] NVIDIA: SAUCE: i2c: mt65xx: use the ACPI declared bus speed when no clock-frequency is given - #719

Open
dcemin-nv wants to merge 1 commit into
NVIDIA:26.04_linux-nvidia-bosfrom
dcemin-nv:i2c-mt65xx-acpi-bus-speed-bos
Open

dcemin-nv wants to merge 1 commit into
NVIDIA:26.04_linux-nvidia-bosfrom
dcemin-nv:i2c-mt65xx-acpi-bus-speed-bos

Conversation

@dcemin-nv

@dcemin-nv dcemin-nv commented Nov 19, 2026 •

Copy link
Copy Markdown

On ACPI platforms the controller node carries no "clock-frequency" property; the bus speed is declared by the peripherals in their I2cSerialBus resources. The driver only looked at the controller property and fell back to standard mode, so every bus ran at 100 kHz regardless of what the peripherals asked for (1 MHz for the touchscreen digitizer, 400 kHz for the keyboard and touchpad on the reference laptops).

At 100 kHz the host cannot keep up with a high report rate HID device such as a pen digitizer; the device buffers its reports and the user sees a constant input lag of about a second. Keyboard and touchpad latency is affected the same way to a smaller degree.

When the controller property is absent and the device has an ACPI companion, take the slowest speed declared on the bus through i2c_acpi_find_bus_speed(), and keep the standard mode fallback only when nothing is declared.

Tested on an arm64 ACPI laptop: the digitizer bus comes up at 1 MHz, the keyboard and touchpad buses at 400 kHz, and the stylus follows the hand with no visible delay.

Upstream status: not upstreamable on its own. The ACPI probe path of i2c-mt65xx is itself a SAUCE change in this tree (c85b3bf), so this fix depends on it and will go upstream together with the ACPI support.

BugLink: https://bugs.launchpad.net/ubuntu/+source/linux-nvidia/+bug/2174977

…lock-frequency is given

On ACPI platforms the controller node carries no "clock-frequency"
property; the bus speed is declared by the peripherals in their
I2cSerialBus resources. The driver only looked at the controller
property and fell back to standard mode, so every bus ran at 100 kHz
regardless of what the peripherals asked for (1 MHz for the touchscreen
digitizer, 400 kHz for the keyboard and touchpad on the reference
laptops).

At 100 kHz the host cannot keep up with a high report rate HID device
such as a pen digitizer; the device buffers its reports and the user
sees a constant input lag of about a second. Keyboard and touchpad
latency is affected the same way to a smaller degree.

When the controller property is absent and the device has an ACPI
companion, take the slowest speed declared on the bus through
i2c_acpi_find_bus_speed(), and keep the standard mode fallback only when
nothing is declared.

Tested on an arm64 ACPI laptop: the digitizer bus comes up at 1 MHz,
the keyboard and touchpad buses at 400 kHz, and the stylus follows the
hand with no visible delay.

Signed-off-by: David Cemin <[email protected]>
(cherry picked from commit fd8059c22bf109c0cad693252f2e0189baf149e8)
@dcemin-nv

Copy link
Copy Markdown
Author

Could someone from BaseOS file the Launchpad bug for this one as well? I will add the BugLink to both PRs (719 and 720) once it exists. The 7.3-next counterpart is #720.

@nirmoy nirmoy added the help wanted Extra attention is needed label Nov 19, 2026
@nirmoy

nirmoy commented Nov 19, 2026 •

Copy link
Copy Markdown
Collaborator

BaseOS Kernel Review

Warning

⚠️ Review needs attention

The mt65xx bus-speed comment is misleading: forced ACPI speed overrides take precedence over the slowest declared speed.

Findings: Critical 0 · High 0 · Medium 0 · Low 1

🔍 Review artifacts

📦 Build checks — 🔴 2/4 passed

Note

Build reports and debs are retained for 10 days after the PR closes.

  • 🟢 PR Illustrated: ready
Review metadata
  • Reviewed head: 3a141b4f9c49
  • Overall status: attention needed
  • Build checks: 2/4 passed

This comment is maintained by BaseOS Reviewer and updated when the GitHub watcher publishes a newer review.

@github-actions

github-actions Bot commented Nov 19, 2026 •

Copy link
Copy Markdown
Contributor

PR Validation Report

Patchscan ✅ No Missing Fixes

All cherry-picked commits checked — no missing upstream fixes found.

PR Lint ❌ Errors found

Details
Checking 1 commits...

Cherry-pick digest:
E: 3a141b4f9c49 ("NVIDIA: SAUCE: i2c: mt65xx: use the ACPI"): cannot resolve upstream SHA fd8059c22bf1
┌──────────────┬──────────────────────────────────────────────────────────────────┬────────────┬─────────┬───────────────────────────┐
│ Local        │ Referenced upstream / Patch subject                              │ Patch-ID   │ Subject │ SoB chain                 │
├──────────────┼──────────────────────────────────────────────────────────────────┼────────────┼─────────┼───────────────────────────┤
│ 3a141b4f9c49 │ fd8059c22bf1                                                     │ ERROR      │ ERROR   │ ERROR                     │
└──────────────┴──────────────────────────────────────────────────────────────────┴────────────┴─────────┴───────────────────────────┘

Lint: all checks passed.

@jamieNguyenNVIDIA

Copy link
Copy Markdown
Collaborator

Can you drop the (cherry picked from commit fd8059c22bf1...) line from both PRs?

This snippet also looks like it can be simplified to:

	if (ret < 0) {
		/* ACPI declares the speed in the peripherals' I2cSerialBus resources */
		i2c->speed_hz = i2c_acpi_find_bus_speed(i2c->dev);
		if (!i2c->speed_hz)
			i2c->speed_hz = I2C_MAX_STANDARD_MODE_FREQ;
	}

i2c_acpi_find_bus_speed() already returns 0 without an ACPI companion, so the extra has_acpi_companion() check and speed_hz = 0 reset aren't needed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

help wanted Extra attention is needed pending_review_comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants