Repository navigation
Conversation
|
Thank you, and good find: a pstore that survives a watchdog reset is worth a lot on this phone, and it's how you found the cluster power-down hang. The node itself is right (12-byte One question about the layout. Stock zumapro's ramoops node ( Could you either use stock's two values, as the other boards do, or show that an oops record survives? A forced panic ( |
3af4c70 to
807c91e
Compare
807c91e to
b67bd83
Compare
|
This branch moved only because #105 was rebased under it. The ramoops commit itself is unchanged, so my question from this morning is still open. The layout differs from stock zumapro ( The simplest fix is stock's two values and no This matters for the order now, because #119 and #122 are stacked on this PR. While it's open, they can't land either. If you'd rather not decide on the layout yet, put #119 and #122 on #105 and move this commit to the top. |
b67bd83 to
3328438
Compare
|
Switched to stock zumapro's layout (0d76b8e): |
3328438 to
0d76b8e
Compare
|
Thanks, that settles it. The PR is merged onto today's One optional wording nit, only if you touch the commit again: "which every other Google board in this tree also uses" isn't quite right, because oriole and felix have no ramoops node. "which the other Google boards here that describe ramoops also use" would be exact. When you boot #122's head (it includes this commit), could you add As for order, nothing changes: this goes in after #96 and after #105, which waits for #100 and #104. Then #119 and #122 follow on top of it. |
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>
Add the Pixel 9a's haptics, a Cirrus Logic CS40L26 on USI9's I2C bus (0x43, interrupt on gpa6-3, reset on gph2-3), with Google's limits for its LRA (2.5 A boost peak, 10 V boost, 162 Hz, ReDC 14.1 ohm) and VP brownout prevention at 3.008 V, and its torch, a TI LM3644 on USI4's I2C bus (0x63, enable on gpp24-1) driving one white LED from both outputs, limited to 75 mA. Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
The Tensor G3/G4 PMU (google,zuma-pmu) gates the NFC controller's reference clock pad. Say what #clock-cells numbers there: clock 0 is that pad. Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
The Pixel 9a's NFC controller (an ST54L) asks the SoC for its reference clock on a CLK_REQ line, which reaches the controller only once the PMU follows that line: PMU 0x3e84 bit 16 (EXTPAD_EN), which Google's st21nfc driver sets from its DT (pmu_clk_pad = <0x3e84>, st,extpad_control_en). Without it the controller doesn't ACK its I2C address. When the zuma PMU's node has #clock-cells, register that bit as clock 0, a gate written through the SMC-backed regmap, and add the register to zuma's access table. Tested on the Pixel 9a (7.3.0-rc6-omamobile-tegu-s2mpg15-m1, with this clock and the NFC node's clocks = <&pmu_system_controller 0>): the clock is enabled by nxp-nci's probe (clk_summary: zuma_pmu_nfc_clk_pad, 25-0008), the controller answers CORE_RESET and polls, with no other code touching the pad. Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
Some boards gate the controller's reference clock, or the SoC pad it comes from, in a clock provider: the Pixel 9a's ST54L takes its clock from a pad the PMU enables. Allow one clock. Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
Take an optional clock and keep it enabled while the device is bound, for boards whose controller clock (or the pad it comes from) is gated: the Pixel 9a's ST54L needs the PMU's NFC clock pad. Tested on the Pixel 9a: the clock is enabled at probe and the controller answers and polls. Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
The zuma PMU provides the NFC controller's clock pad as clock 0; give its node #clock-cells. Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
The Pixel 9a's ST54L sits at 0x08 on hsi2c9, its interrupt on gpa7-5 (level high), its RST_N on gph1-6, and its reference clock from the PMU's clock pad. Its chip enable (gpa0-1) powers the whole chip and is held high by a hog: stock's eSE driver raises it, and nothing here drives the eSE. Tested on the Pixel 9a: the controller probes on nxp-nci, polls and reads a contactless card (ISO-DEP over NFC-A) through neard. Link: https://android.googlesource.com/kernel/devices/google/tegu/+/6276884d9940694b25a6934671918ec8ac412ee8/dts/zuma-tegu-nfc.dtsi Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
The region stock uses for ramoops was only reserved, because zuma's user-build bootloader doesn't keep it across a warm reset. zumapro's does: on the Pixel 9a, after a hang that ended in a watchdog reset, the console zone held the previous kernel's log up to the hang (after the bootloader's own log, which it writes at the start of the zone). Describe it as ramoops with stock zumapro's layout, which every other Google board in this tree also uses: console-size 0x200000 and pmsg-size 0x200000, with no record-size, so the console zone starts where the bootloader writes its log. Signed-off-by: W4HO <87742609+W4HO-ham@users.noreply.github.com>
0d76b8e to
2c4ddaf
Compare
|
Rebased onto #105's new head (one commit, unchanged). Compiles; checkpatch clean. |
|
This one was ready and doesn't depend on anything below it, so I cherry-picked its single commit onto |
Stacked on #105 (the DT series #80 → #105); only the last commit is new.
zumapro's ramoops region (stock's,
fd3ff000, 4 MiB) was only reserved, because zuma's user-build bootloader drops it on a warm reset. zumapro's keeps it: on the palmtop, after a hang that ended in a watchdog reset,/sys/fs/pstore/console-ramoops-0held the previous kernel's console up to the hang, after the bootloader's own log at the start of the zone. That's how I found what kills cluster power-down hints in cpuidle (ACPM stops answering its DVFS channel).The node gets
compatible = "ramoops"with stock's 2 MiB console zone, 1 MiB pmsg and 256 KiB oops records. Booted on the palmtop (round ps1): ramoops binds, and the region survives a watchdog reset.