Skip to content

arm64: dts: exynos: google: zumapro-tegu: Enable the display - #90

Closed
W4HO-ham wants to merge 6 commits into
derekross:omamobilefrom
W4HO-ham:tegu-dt-display
Closed

W4HO-ham wants to merge 6 commits into
derekross:omamobilefrom
W4HO-ham:tegu-dt-display

Conversation

@W4HO-ham

@W4HO-ham W4HO-ham commented Oct 27, 2026 •

Copy link
Copy Markdown
Contributor

Stacked on #87, #85, #81 and #80. It uses #89's google,tg4c binding, and at run time #88 and #89. Only the last three commits are new.

  • google,gs201-dsim: add google,zumapro-dsim, with google,gs201-dsim as the fallback. The Tensor G4 has the same DSIM and register layout, and the 9a has run gs-dsim-cmd on the gs201 compatible all along.

  • zumapro.dtsi: DSIM0, with its link and D-PHY windows and GIC SPI 252 as in Google's zumapro tree. It's disabled.

  • zumapro-tegu.dts:

    • DECON0 with a 64 MiB reusable scanout pool below 4 GiB and the bootloader's splash as its boot framebuffer, scanning out to DSIM0;
    • the TG4C panel on DSIM0.

    With DECON's output on DSIM0, gs-decon switches between 60 and 120 Hz.

Against the palmtop: the DSIM0, panel, DECON port and pool nodes match the DTB it boots now (round pn1, the panel as google,tg4c driven by #89's driver), apart from DSIM0's new compatible. That boot runs the desktop at 120 Hz with DPMS sleep and wake.

CHECK_DTBS=y W=1 for zumapro-tegu.dtb is clean, dt_binding_check passes for the DSIM schema, and checkpatch --strict is clean on all three commits.

@derekross

Copy link
Copy Markdown
Owner

Thank you, the display nodes are compact and they check out against Google's tree.

What I checked:

The clock series under this branch has landed on omamobile, so this one will need a rebase once #80, #81, #85 and #87 are in; it merges last, after #86 and #89.

One question before it merges. Google's tegu device tree lists three panels under the connector (TG4A, TG4B and TG4C), and as on the Pixel 7 Pro I'd expect the bootloader to name the fitted one on the kernel command line (panel_name=). This file names TG4C for every Pixel 9a. TG4B's rate switch sends more than 2F (Google's tg4b_change_frequency adds 26 00 at 120 Hz, and F0 55 AA 52 08 00, 6F 03, C0 10 at 60 Hz), and its link probably matches TG4C's timer set, so a phone with a TG4B would get half a rate switch.

  • Could you check what the bootloader reports on your 9a (tr ' ' '\n' < /proc/cmdline | grep -i panel; it's only a panel name), and whether you know if TG4B phones are out there? It would help to know the other 9a owner's too.
  • Since gs-dsim-cmd can't read from the panel yet, the driver can't check the panel's ID itself. I'd suggest the guard goes in the tegu device package: set gs_dsim_cmd.allow_tx=1 only when the bootloader names tg4c. Another panel then gets no commands at all and keeps the bootloader's picture.

Two smaller notes: gs201 keeps the DECON0 to DSIM0 ports in the SoC dtsi, and zumapro could do the same (optional); and the "created CMA memory pool" line from your 9a's dmesg would be nice to have in the PR. For the omarchy-mobile side, tegu's kernel config will need CONFIG_DRM_PANEL_GOOGLE_TG4C=m, and the device package the allow_tx option.

@W4HO-ham
W4HO-ham force-pushed the tegu-dt-display branch 2 times, most recently from fa04098 to 39d8bb3 Compare October 27, 2026 20:19
@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Thanks for checking it against Google's tree.

The bootloader on my 9a reports exynos_drm.panel_name=google-tg4c.2d008004, so TG4C with build code 0x80, the production one. I don't know of TG4B phones in the field: Google's driver lists TG4A and TG4B before TG4C, which reads like earlier builds, but I can't prove none shipped, and I haven't heard from another 9a owner yet.

I agree on the guard. The tegu device package isn't published yet; when it is, it will set gs_dsim_cmd.allow_tx=1 only when the bootloader names google-tg4c, so any other panel gets no commands and keeps the bootloader's picture. Its kernel config will carry CONFIG_DRM_PANEL_GOOGLE_TG4C=m.

The pool from the 9a's boot log:

Reserved memory: created CMA memory pool at 0x000000009a000000, size 64 MiB
OF: reserved mem: initialized node display-pool, compatible id shared-dma-pool
OF: reserved mem: 0x000000009a000000..0x000000009dffffff (65536 KiB) map reusable display-pool
gs-decon 19470000.display-controller: assigned reserved memory node display-pool

I'll take the binding wording and moving the DECON0 to DSIM0 ports into the SoC dtsi at the rebase after #80, #81, #85 and #87.

@derekross

Copy link
Copy Markdown
Owner

Thank you for the answers and the pool lines.

  • google-tg4c.2d008004 settles it for your phone, and the pool sits where it should: 0x9a000000 plus 64 MiB ends exactly at the top of alloc-ranges.
  • The guard in the tegu device package sounds right. Since allow_tx is off by default, the kernel side sends nothing to any panel until the package turns it on, so this branch doesn't need to wait for it. When the package comes, I'll check that it only sets allow_tx for google-tg4c and carries CONFIG_DRM_PANEL_GOOGLE_TG4C=m. If you ever hear from another 9a owner, their panel name would be good to have.

Your three commits are unchanged from last time. I built every Google board's device tree with this branch: the six existing ones are byte-identical and validate exactly as before. With #89 and #88 merged in as well, the 9a's tree validates with no messages, so CI's one remaining failure (google,tg4c) goes away once #89 lands.

So the plan stays the same. #89 can merge now, then the board PRs under this one. After that, rebase this branch onto omamobile with the binding wording and the DECON0-to-DSIM0 ports moved into the SoC dtsi, as you offered, and it merges last.

@W4HO-ham

Copy link
Copy Markdown
Contributor Author

CI's dtbs_check was red here only because google,tg4c is documented in #89. This branch now carries #89's two commits under its own, on top of the current #87, so the check has the binding; it merges after #89 as before.

@derekross

Copy link
Copy Markdown
Owner

Thanks for rebasing. With #89 merged, CI passes here and the 9a's device tree validates with no messages.

Your three commits are unchanged, so the binding wording ("compatible with the gs201 DSIM for the registers its driver uses") and moving the DECON0-to-DSIM0 ports into the SoC dtsi aren't in yet. Both are optional. You'll be pushing the stack once more anyway for the "Tested on" line (see #80), so that push would be the easiest time to add them. If you'd rather leave them for later, that's fine too.

The stack will land through #96, its top, and I'll close this PR as landed there.

@derekross

Copy link
Copy Markdown
Owner

Thanks. All three of your commits are unchanged since my last look (the tree at this commit is the same, 85352566a3a0), so the two optional points, the binding wording and moving the DECON0-to-DSIM0 ports into zumapro.dtsi, aren't in. That's fine: they were optional, and it's fine to leave them for a later patch. If you do the Makefile rebase from #80 anyway, that would be an easy time to add them.

With the rebase resolved, the display commits build at each step without warnings, dt_binding_check passes for google,gs201-dsim.yaml, and the 9a's DTB validates with no messages. The DTB is byte-identical to the one from your previous head.

One thing outside this PR. On 10-27 you said the tegu device package would set gs_dsim_cmd.allow_tx=1 only when the bootloader names google-tg4c. The tegu device package now in review on omarchy-mobile (#74, from the other 9a owner) sets options gs_dsim_cmd allow_tx=1 for every panel. I'll raise it there, so that whichever package lands keeps your check. Nothing changes here: the kernel sends no panel commands until something sets allow_tx. If you hear the other owner's exynos_drm.panel_name, I'd still like to have it.

The display will be covered by the boot of #96's head (the smoke report's display row, 1080x2424 on DSI-1). This lands with #96.

Turn on CMU_G3D and the Mali-G715 for the Pixel 9a, with its core clock
on ACPM's paired "gpu" clock (index 17), which sets the shader (G3D) and
L2 (G3DL2) domains together, and coregroup and stacks on CMU_G3D's gates.

The operating points are the shader rates of the 9a's GPU DVFS table up
to 649 MHz. The levels above it, to 940 MHz, wait for a GPU thermal
zone, since nothing throttles the GPU yet.

Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
Give the three CPU clusters their ACPM DVFS clocks (cpucl0-2, by the
zumapro ACPM clock ids) and operating points for cpufreq-dt, with
capacities and power coefficients for the energy model from Google's
zumapro-cpu.dtsi, and add the ACPM TMU sensor and the big, mid, little
and GPU thermal zones.

The CPU levels are the APM firmware's own (its DVFS table, read on the
Pixel 9a), up to the 9a's thermal headroom: 1950 MHz little (Google's
max-freq for that cluster), 2367 MHz mid and 2687 MHz big. The little
cluster's 324 and 610 MHz lie below stock's 820 MHz floor. Each CPU zone
throttles its cluster from 80 C, forces the cluster's lowest levels
from 95 C and shuts down at 105 C, all below stock's 112 C (CPU) and
115 C (GPU) hot trips. gpu-thermal throttles panthor's devfreq from
85 C.

Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
… MHz

With gpu-thermal throttling the GPU, add the 9a's GPU levels above 649
MHz: 723, 807, 850 and 890 MHz. The table's top, 940 MHz, stays out
until it has run on the 9a.

Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
The Tensor G4 (zumapro) display pipeline has the gs201 DSIM, at the same
register layout and with the same three register windows. Add
google,zumapro-dsim with google,gs201-dsim as its fallback.

Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
Add DSIM0, the DSI master between DECON0 and the panel: its link and
D-PHY register windows and its interrupt (GIC SPI 252), as in Google's
GPL zumapro device tree, disabled until a board names its panel.

Link: https://android.googlesource.com/kernel/devices/google/zumapro/+/8831630eee2ad095b85ed452944e6353940966a5/dts/zumapro.dtsi
Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
Turn on the Pixel 9a's display pipeline: DECON0 with a 64 MiB reusable
scanout pool below 4 GiB and the bootloader's splash as its boot
framebuffer, scanning out to DSIM0, and the TG4C panel on DSIM0's
virtual channel 0. With DECON's output on DSIM0, gs-decon switches the
link and the panel between 60 and 120 Hz.

Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Rebased onto today's omamobile. The Tensor G4 SoC, the Pixel 9a board and its touchscreen are in omamobile now, so those commits dropped out; six remain (GPU, CPU DVFS and thermal zones, the GPU's levels, the DSIM binding, DSIM0, the display). zumapro-tegu.dts compiles with dtc; checkpatch --strict clean on each commit.

@derekross

Copy link
Copy Markdown
Owner

Thanks for the rebase. Same patches and trees as before, and the gate passes on today's omamobile. This still lands through #96 once its display and GPU boot is in.

@derekross

Copy link
Copy Markdown
Owner

Its patches landed unchanged through #96, so I'm closing this one. Thanks!

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

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants