Skip to content

gs201-panther: the S6E3FC3 panel and the DSI command path, for brightness - #177

Merged
derekross merged 3 commits into
derekross:omamobilefrom
nesmeck:omamobile-gs201-panther-display
Nov 1, 2026
Merged

derekross merged 3 commits into
derekross:omamobilefrom
nesmeck:omamobile-gs201-panther-display

Conversation

@nesmeck

@nesmeck nesmeck commented Nov 1, 2026 •

Copy link
Copy Markdown
Contributor

Brightness and display off/on for the Pixel 7, as the Pixel 7 Pro has them: a driver for its panel, the Samsung S6E3FC3 "P10", and the DSI command path in panther's board file. DECON0 stays disabled, so simplefb keeps scanning out. gs-decon for panther's 1080x2400 panel is a later step.

  1. dt-bindings: samsung,s6e3fc3-p10. This is the compatible string the Pixel 7's own device tree uses for the panel under DSIM0. I'm listed in the binding's maintainers: and, as with the FT3658, not in MAINTAINERS. That's the reason for checkpatch's two "does MAINTAINERS need updating?" warnings.
  2. drm/panel: the S6E3FC3 driver. It follows this tree's S6E3HC4 driver, and the command sequences come from Google's GPL panel-samsung-s6e3fc3-p10.c (kernel/google-modules/display/samsung, panel/, branch android-gs-pantah-6.1-android16), limited to the EVT1 and later panels that ship in phones.
    • Takeover: if the reset line reads released at probe, the driver takes the running panel over and sends nothing.
    • Brightness: a backlight device over DCS 0x51 (4-2047, the normal range; HBM is left out).
    • Blanking: it turns the display off without putting the panel to sleep.
    • For KMS later: prepare runs Google's reset and cold init at 60 Hz, TE2, DSC on with the vendor PPS, MIC off (0xc2 0x14), the DSC decoder (0x9d 0x01) and WRCTRLD. Enable and disable are display on and off, and unprepare is sleep in and reset.
    • Left out: local HBM (it needs panel reads), the AOD modes and the panel revision reads.
  3. gs201-panther: the DSI command path. cmu_disp, cmu_dpu and dsi0, with the panel on reset gpa7-1, as in the live tree. The supplies, vci on S2MPG12 LDO27 and vddi on S2MPG13 BUCKA (Google's gs201-panther-display.dtsi), have no driver yet and stay as ABL leaves them.

Tested on a Pixel 7 (slot B, 7.3.0-rc6-omamobile-push1-g54fb87ae5295: today's omamobile with #133, #171, #170 and this branch). One change per boot: the boot before it ran the same tree without this branch.

  • D1, nothing sent (gs_dsim_cmd allow_tx=0):
    • both drivers bind, and the panel logs "taken over from the bootloader" (4 lanes, non-continuous clock);
    • /sys/class/backlight/1c2c0000.dsi.0 appears;
    • the screen stays exactly as it was on simplefb.
    • ABL's link timers aren't in gs_dsim_rates[] (UNDERRUN_CTRL 0x1cf8), so gs-dsim-cmd logs "no rate switching" and stays at 60 Hz, as intended.
  • D2, with allow_tx=1 on the running kernel:
    • brightness 1023, 400, 2047 and 4 each changed the screen as expected;
    • bl_power 4 turned the display off, and 0 turned it back on with the picture intact and touch working;
    • the shell's brightness slider then drove the panel;
    • no DSIM timeouts or errors in dmesg.

omamobile-device-panther 0.1.4 ships options gs_dsim_cmd allow_tx=1, in derekross/omarchy-mobile#258.

Checks: tools/kernel-pr-check --build OUT --merge origin/omamobile passes everything:

  • a clean merge onto 8b8235e;
  • the sign-offs;
  • checkpatch --strict with no errors (only the two MAINTAINERS warnings above);
  • the W=1 build of the driver;
  • dt_binding_check on the new binding;
  • the 8 device trees with W=1 and dtbs_check, with no new warnings.
tools/phone-smoke from the D1/D2 boot (…-g54fb87ae5295)

phone-smoke: panther

Kernel 7.3.0-rc6-omamobile-push1-g54fb87ae5295, tools at 19805f0, 2026-11-01T15:59Z.

Check Result Detail
boot (required) ok kernel 7.3.0-rc6-omamobile-push1-g54fb87ae5295, slot b, up 4054s
storage guard (required) ok 55 partitions; only userdata and slot B's boot images writable
root (required) ok ext4 on userdata, 97G free
kernel health (required) ok no oops or BUG; 0 WARN, 9 error lines in dmesg
systemd units ok none failed
USB gadget (required) ok bound to 11210000.usb
display ok card0-Unknown-1 connected, 1080x2400
GPU ok render node on panthor
session (required) ok Hyprland and the Omarchy shell running
lock ok PAM file present, the shell can lock (passwordPam true, locked false)
touch ok FocalTech FT3658
buttons ok gpio-keys
Wi-Fi skip no wireless interface
Bluetooth skip no controller
battery ok 92% Charging, 31 °C
sound cards skip none
sensors skip no sensor proxy
cameras skip no libcamera tools
thermal ok hottest zone 44 °C

Packages: hyprland 0.56.2-5.1, libcamera 0.7.2-4.21, mesa 1:26.2.4-1.4, omacam 0.3.3-1, omamobile-apps 0.2.0-1, omamobile-device-panther 0.1.4-1, omamobile-firstboot 0.1.2-1, omamobile-keyring 20261026-1, omamobile-mirror 0.1.1-1, omamobile-omarchy 0.1.12-1, omamobile-owl-widgets 0.1.1-1, omamobile-shell 0.2.28-1, omarchy 4.0.6-1, quickshell 0.3.2.r7.g2b4501a-1

🤖 Generated with Claude Code

@nesmeck

nesmeck commented Nov 1, 2026

Copy link
Copy Markdown
Contributor Author

I rebased this onto today's omamobile (fa55531), now that #133 and #171 are in.

kernel-pr-check passes on the new head:

  • a clean merge;
  • the sign-offs;
  • checkpatch with only the two MAINTAINERS warnings;
  • the W=1 build of the driver;
  • dt_binding_check;
  • the 8 device trees with no new warnings.

The next step is ready locally: gs-decon driving the Pixel 7's screen, with the panel asleep when the screen is off. It's tested on the phone and stacked on this branch. I'll send it once this PR is in, to keep your queue short.

🤖 Generated with Claude Code

@nesmeck
nesmeck force-pushed the omamobile-gs201-panther-display branch from 3a0f596 to 9f3b8c0 Compare November 1, 2026 18:47
@derekross

Copy link
Copy Markdown
Owner

Thank you, @nesmeck. This is a clean port. I checked it against Google's panel-samsung-s6e3fc3-p10.c on android-gs-pantah-6.1-android16: the 4-2047 brightness range is exactly the vendor's normal band, the PPS matches byte for byte, the mode timings and the 66 x 146 mm size match, and the init writes I compared match the vendor's tables. Taking the panel over without sending anything, and leaving DECON0 disabled, keeps the step small. Your D1/D2 split, nothing sent first and then allow_tx, is exactly how I'd like these steps run. The gate passes on today's omamobile (clean merge, sign-offs, checkpatch with only the MAINTAINERS warnings, W=1, no new device-tree warnings), CI is green, and every other board's DTB, the Pixel 7 Pro's included, is unchanged.

Two small things before I merge:

  1. The panel supplies in the comment. The new &dsi0 comment says vci (S2MPG12 LDO28M) and vddi (S2MPG13 buck-boost), and the commit message says "S2MPG12 LDO28". Google's gs201-panther-display.dtsi (kernel/devices/google/pantah, same branch) has vci-supply = <&m_ldo27_reg> and vddi-supply = <&s_bucka_reg>, and gs201-panther-pmic.dtsi labels m_ldo27_reg "Display: vci-supply voltage". So vci is LDO27M and vddi is S2MPG13 BUCKA. LDO28M is the Pixel 7 Pro's vddd. Please fix both, since the commit that wires the supplies will start from this comment.
  2. The smoke report. Please paste the tools/phone-smoke report from your test boot (or from one on this head) into the description, as the template asks. The driver and binding you tested are identical to this head, so the boot itself still counts.

One more thing that I'd like fixed before the gs-decon step, here or in that PR. The commit says the driver follows our S6E3HC4 driver, but it follows an older version of it. Today's panel-samsung-s6e3hc4.c marks a handed-over panel as prepared and enabled (ctx->prepared, panel.prepared, panel.enabled), skips DCS from the backlight while the panel is held in reset or asleep, and holds a mutex around all of it. Without that, a brightness write on the "held in reset" path goes to a panel in reset. Worse, once gs-decon drives this panel, the DRM core sees an unprepared panel and calls prepare(), which resets and cold-inits the panel the bootloader left running. With DECON0 disabled here nothing calls prepare() yet, so it doesn't block this PR, but gs-decon will need it.

I'll merge this together with omarchy-mobile#258. Thanks again, and I'm looking forward to the gs-decon step.

The S6E3FC3 is a Samsung Mobile Display DDIC; the P10 panel is its
1080x2400 command-mode AMOLED panel in the Google Pixel 7. Its
compatible string is the one the Pixel 7's own device tree uses.

Assisted-by: Claude Opus 5.5
Signed-off-by: Alejandro Garcia <skipper.pelygro@gmail.com>
Add a driver for the Samsung S6E3FC3 DDIC and its P10 1080x2400 AMOLED
panel, found in the Google Pixel 7 (panther).

The command sequences come from Google's GPL panel driver
(panel-samsung-s6e3fc3-p10.c), limited to the EVT1 and later panels that
ship in phones: the cold init table at 60 Hz, TE2, DSC enable with the
vendor PPS sent as is, MIC off, and display on/off. Local HBM, the
low-power (AOD) modes and the panel revision reads are left out.

The bootloader powers the panel and leaves it scanning out. When the
reset line reads released at probe the driver takes the panel over
without sending anything, holding the supplies and keeping the reset
line where it is; only brightness and power changes reach the panel.
Brightness is a backlight device over DCS 0x51 (11 bits, the normal
range; HBM is left for later), with levels below the vendor driver's
lowest (4) raised to it, and blanking the backlight turns the display
off while keeping the panel out of sleep.

The driver follows the Pixel 7 Pro's S6E3HC4 driver in this tree.

Tested on a Pixel 7 from slot B (7.3.0-rc6), with the next patch: the
driver takes the panel over from the bootloader without changing the
picture. With the DSIM command path sending, brightness 1023, 400, 2047
and 4 each changed the screen as expected, blanking the backlight turned
the display off and unblanking turned it back on with the picture
intact, and the desktop's brightness slider drove the panel. No DSIM
errors.

Assisted-by: Claude Opus 5.5
Signed-off-by: Alejandro Garcia <skipper.pelygro@gmail.com>
…o the panel

Enable the DPU and DISP clock controllers and DSIM0, and describe the
panel on it: the Samsung S6E3FC3 P10 with its reset on gpa7-1, as in the
Pixel 7's live device tree. The bootloader leaves DECON0, DSIM0 and the
panel running in command mode, and simplefb keeps scanning out its
buffer; the DSIM command driver and the panel driver take the link and
the panel over without sending anything, so that brightness and display
on/off can go over DCS. DECON0 stays disabled.

The panel's supplies, vci on S2MPG12 LDO27 and vddi on S2MPG13 BUCKA
(Google's gs201-panther-display.dtsi and the live tree), have no driver
yet and are left as the bootloader set them.

Tested on a Pixel 7 from slot B (7.3.0-rc6): DSIM0 and the panel bind,
the panel is taken over from the bootloader, the screen stays as it was
on simplefb, and the backlight device appears. Brightness and display
off/on over DCS then work (see the panel driver's commit).

Assisted-by: Claude Opus 5.5
Signed-off-by: Alejandro Garcia <skipper.pelygro@gmail.com>
@nesmeck
nesmeck force-pushed the omamobile-gs201-panther-display branch from 9f3b8c0 to 6248800 Compare November 1, 2026 20:40
@nesmeck

nesmeck commented Nov 1, 2026

Copy link
Copy Markdown
Contributor Author

Thanks for the careful check against Google's driver.

  1. The supplies: you're right, and I've fixed both places. In D1 I read the regulator name from the node after the one the vci-supply phandle points to. Google's gs201-panther-display.dtsi and the live tree agree:

    • vci is S2MPG12 LDO27;
    • vddi is S2MPG13 BUCKA.

    The DT comment and the commit message now say so, and so does DISPLAY-PLAN in omarchy-mobile#258.

  2. The smoke report: it's in the description now, from the D1/D2 boot (…-g54fb87ae5295, the same driver and binding as this head).

  3. The handed-over panel's state: it's the first commit of the gs-decon step, drm/panel: samsung-s6e3fc3: support sleep and a cold restart from KMS, which follows today's S6E3HC4 driver:

    • a handed-over panel is marked prepared and enabled, so the DRM core never calls prepare() on it;
    • the backlight writes nothing while the panel is in reset or asleep;
    • one mutex covers all of it.

    With that, gs-decon's DPMS cycles and screen_off=2 cold restarts ran clean on the phone, and that PR will carry the results. If you'd rather have that commit here instead, I can move it.

The new head (6248800f5) is rebased onto today's omamobile (8b8235e). Only the DT commit's comment and message changed (git range-diff), and it's re-signed. kernel-pr-check passes.

🤖 Generated with Claude Code

@derekross
derekross merged commit f101812 into derekross:omamobile Nov 1, 2026
1 check passed
@derekross

Copy link
Copy Markdown
Owner

Thank you, @nesmeck. Both points are in: the comment and the commit message now name LDO27M and BUCKA, as Google's tree does, and the smoke report is in the description. The range-diff shows the binding and driver unchanged. I've re-run our gate on today's omamobile, which has moved again since your rebase: a clean merge, the sign-offs, checkpatch with only the MAINTAINERS warnings, the W=1 build and no new device-tree warnings, and CI is green.

Putting the handed-over panel's state at the start of the gs-decon series is fine with me. That's the PR where prepare() starts to matter, and I'll look for it there.

Merged; omarchy-mobile#258 goes in right after it. Thanks again. I'm looking forward to seeing the Pixel 7's screen driven by gs-decon.

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