The QCS8550 RB5 Gen 2 device tree against the Nova's

Linux 7.3 adds Qualcomm's own QCS8550 board, the RB5 Gen 2 (arch/arm64/boot/dts/qcom/qcs8550-rb5gen2.dts, 1573 lines, read from mainline at 7.3-rc6 on 2026-10-06, with its qcs8550.dtsi). It is the closest thing to a canonical QCS8550 setup, so it was compared node by node with the Nova chain: qcs8550-retroidpocket-rpnova.dts includes qcs8550-retroidpocket-rp6.dts includes qcs8550-ayn-common.dtsi, all in projects/PortareOS/devices/SM8550/linux/dts/qcom/. Both boards sit on sm8550.dtsi; the Nova build also patches that file (0001 PCIe PHY on the root port, 1049 PCIe suspend OPP, 1051 lowest GPU OPP, 1071 MDSS regdma, 0120 gpu_cc, 20260424 gcc CX power domain) and the RB5 runs it stock.

Line references: "RB5 nnn" is the RB5 dts; "common nnn", "rp6 nnn", "nova nnn" are the three Nova files, at the tree state of 2026-10-06.

Result

Nothing in the RB5 board file is worth taking. Every node both boards set (UFS, USB PHYs and repeater supplies, pcie0_phy, the WCN7850 PMU LDOs and the wifi and Bluetooth supplies, SD card detect, pinctrl and supplies, the zap shader path, DSI and DP supplies, clocks, the reserved GPIO range, the volume-up wake key, pon) is identical or the Nova's is a superset. What the RB5 adds is dev-board hardware: pm8010 and pmr735d PMICs, modem and IPA, two TC9563 PCIe switches with Wi-Fi behind the second, an LT9611 HDMI bridge, an nb7vpq904m USB retimer, CAN, status LEDs.

Four things the comparison turned up are on the Nova's own side, and each was checked against the files on 2026-10-06:

• WCN7850 rail windows. The PMU's vddaon is s2g and vdd is s5g on both boards, but the software windows are crossed: RB5 allows s2g 0.50-1.05 V and s5g 0.80-1.00 V, the Nova s2g 0.80-1.00 V and s5g 0.50-1.00 V. The PMU driver only enables these rails and sets no voltage, so the window matters only as the floor RPMh may settle to. Which pair is right is not decidable from the files; the Wi-Fi works with ours.
• SD card CX vote. The Nova replaces sdhc_2 with the downstream driver's node for SDR104 (rp6 395-494, reason in its comment). That node carries no power-domains and no required-opps, so unlike the stock node nothing in the device tree raises CX for the 202 MHz clock. It runs at 85 MB/s today, so something else holds CX while a card is busy; worth knowing if SD ever misbehaves at idle.
• Sound card name. The card still reports model = "AYN-Odin2". That is not a leftover to fix: the kernel loads the topology blob by card name (AYN-Odin2-tplg.bin, see AUDIO_SAMPLE_RATES.md), and 030-suspend_mode and extra-firmware key on it too. Renaming it means renaming all of those together.
• SW_CTRL. RB5 hands the WCN7850's SW_CTRL line (gpio82) to the PMU driver as swctrl-gpios; the Nova pins it down in bt_default and does not. The 7.2 driver (pwrseq-qcom-wcn.c) never reads that property, so the two are equivalent today; it is the binding's documented place for the line if a later driver starts using it.

The thermal trips the Nova adds have no cooling-maps, which is deliberate: the fan is driven from user space (fancontrol), not by the kernel's thermal framework.

1. PMICs and regulators

PMIC includes:

RB5 (lines 11-20)	Nova (common 9-15)	Difference
qcs8550, pm8010, pm8550, pm8550b, pm8550ve (PMK8550VE_SID 5), pm8550vs, pmk8550, pmr735d_a, pmr735d_b	qcs8550, pm8550, pm8550b, pm8550ve (SID 5), pm8550vs, pmk8550	Nova does not include pm8010 (two camera-type PMICs, RB5 regulators-6 "m" 755-820 and regulators-7 "n" 822-887) nor pmr735d_a/b (RB5 declares no regulators for them in the board file).

regulators-0 (pm8550, pmic-id "b"; RB5 297-485, common 572-752). Input supplies are the same rails under different labels (RB5 vreg_s6g_1p86/vreg_s4g_1p25 = Nova vreg_s6g_1p8/vreg_s4g_1p3).

Rail	RB5 min-max uV	Nova min-max uV	Difference
bob1	3296000-3960000 (314-319)	3296000-3960000 (589-602)	Nova adds regulator-state-mem { on-in-suspend; mode LPM }.
bob2	2720000-3960000 (321-326)	2720000-3960000 (604-617)	Nova adds regulator-state-mem { off-in-suspend }.
l1b	1710000-1950000 (328-333)	not declared	RB5 only.
l2b	2900000-3544000 (335-340)	3008000 fixed (619-627)	Nova fixed + set-load.
l5b	3000000-3300000 (342-347)	3104000 fixed (629-637)	Nova fixed + set-load. Feeds eUSB2 repeater vdd3 on both.
l6b	1700000-3300000 (349-354)	1800000-3008000 (639-647)	Nova narrower + set-load.
l7b	1700000-3300000 (356-361)	1800000-3008000 (649-654)	Nova narrower, no set-load.
l8b (SD vqmmc)	1800000-3008000 (422-427)	1800000-3008000 (656-664)	Same range; Nova adds set-load.
l9b (SD vmmc)	2960000-3008000 (429-434)	2960000-3008000 (666-674)	Same range; Nova adds set-load.
l11b	1100000-1300000 (436-441)	1200000-1504000 (676-684)	Different window.
l12b	1710000-1950000 (443-448)	1800000 fixed (686-694)	Nova fixed + set-load.
l13b	2700000-3300000 (450-455)	3000000 fixed (696-704)	Nova fixed + set-load.
l14b	3200000-3300000, regulator-always-on (457-463)	3200000 fixed, set-load, not always-on (706-714)	RB5 keeps l14b on; it feeds its LT9611 HDMI bridge regulators (160, 170) and CAN (1382-1383). Nova has none of those consumers.
l15b	1760000-1800000 (465-470)	1800000 fixed, set-load, regulator-state-mem { on-in-suspend; LPM } (716-731)	Nova keeps it on in suspend (codec + panel 1.8 V per comment 725-726). RB5 uses it for WCN vddio, repeater vdd18, WSA speakers, retimer vcc.
l16b	2000000-3400000 (472-477)	2800000 fixed (733-741)	Nova fixed + set-load.
l17b (UFS vcc)	2400000-3300000 (479-484)	2504000 fixed (743-751)	Nova fixed + set-load.

Other PMICs:

Node	RB5	Nova	Difference
regulators-1 "c" (pm8550vs)	supplies l1/l2/l3 + vdd-s1/s4/s6 = vph_pwr; l3c 835000-912000 (487-504)	supplies l1/l2/l3 only; l3c 880000-912000 + set-load (754-771)	Nova omits the vdd-sN-supply lines and raises l3c min.
regulators-2 "d"	l1d 825000-958000, l2d 675000-808000, s4d 572000-988000, s5d 572000-988000 (506-542)	l1d 880000-920000 + set-load only (773-790)	Nova declares no l2d, s4d, s5d.
regulators-3 "e"	l1e 831000-904000, l2e 870000-970000, l3e 1200000, s1e 532000-852000, s3e 716000-884000, s4e 870100-1152000, s5e 1010000-1120000, s6e 528000-904000 (544-612)	s4e 904000-984000, s5e 1010000-1120000, l1e 880000-912000, l2e 870000-970000, l3e 1200000; set-load on LDOs (792-845)	Nova declares no s1e, s3e, s6e. s4e window differs (RB5 wider).
regulators-4 "f" (pm8550ve)	l1f 866000-958000, l2f 866000-880000, l3f 830000-920000, s1f 516000-904000, s3f 688000-952000, s4f 300000-500000, s5f 716000-884000, s7f 516000-812000 (614-682)	s4f 300000-700000, l1f/l2f/l3f 880000-912000 + set-load (847-892)	Nova declares no s1f, s3f, s5f, s7f; s4f max 700000 vs 500000.
regulators-5 "g"	l1g 1140000-1260000, l3g 1200000, s1g 1172000-1388000, s2g 500000-1053200, s3g 532000-1148000, s4g 1172000-1800000, s5g 800000-1002600, s6g 1800000-2192000 (684-753)	l1g 1200000, l3g 1200000 (+set-load), s1g 1200000-1300000, s2g 800000-1000000, s3g 300000-1004000, s4g 1200000-1352000, s5g 500000-1004000, s6g 1800000-2000000 (894-969)	Same rails, different windows. WCN vddaon = s2g: RB5 min 500000, Nova min 800000. WCN vdd = s5g: RB5 min 800000, Nova min 500000.
regulators-6/7 "m"/"n" (pm8010)	14 LDOs 755-887; regulator-allow-set-load + allowed-modes LPM HPM only on l1m, l2m, l1n, l2n (771-773, 781-783, 838-840, 848-850)	absent	RB5 is the only tree with pm8010.

regulator-allow-set-load/regulator-allowed-modes: RB5 sets them on four pm8010 LDOs only (grep); every pm8550 b/c/d/e/f/g LDO is HPM with no load switching. Nova sets them on every LDO it declares in b, c, d, e, f, g (common 624-965). regulator-always-on/boot-on: RB5 vph_pwr (247-248) and vreg_l14b_3p2 (462). Nova vph_pwr (480-481), vdd_mcu_3v3 fixed regulator boot-on (470); no RPMh rail is always-on. Nova alone has regulator-state-mem blocks (bob1, bob2, l15b, vdd_fan_5v0 456-458).

Verdict: nothing to take. RB5's regulator windows are the reference-board defaults with no load switching; the only always-on rail it adds (l14b) exists for peripherals the Nova does not have. Worth checking: the s2g/s5g minimums differ between the two boards for the same WCN7850 supplies (RB5 s2g 500000/s5g 800000 vs Nova 800000/500000); the files alone do not say which is right.

2. RPMh power-domain votes

Item	RB5	Nova	Difference
required-opps / power-domains / rpmhpd in board files	none (grep over the dts: zero hits)	none in common/rpnova; rp6's replacement sdhc_2 (rp6 415-494) has an opp-table with opp-hz, opp-peak-kBps, opp-avg-kBps but no required-opps, and the node has no power-domains	The stock sdhc_2 (sm8550.dtsi 3605, 3622-3640) votes rpmhpd RPMHPD_CX with min_svs/low_svs/svs/svs_l1 per OPP; RB5 inherits that. Nova's downstream node drops the CX vote and the per-OPP levels entirely.
opp-level choices	inherited stock sm8550.dtsi	stock plus repository patches: 1049 adds a PCIe opp-suspend OPP (opp-hz 1, required-opps low_svs, opp-peak-kBps 250000 1), 1051 adds GPU opp-124800000 at LOW_SVS_D2, 20260424 adds power-domains = <&rpmhpd RPMHPD_CX> to gcc	Nova's base dtsi has three vote changes RB5 does not carry.

Verdict: nothing to take from RB5 (it makes no board-level votes). Worth checking on Nova's side: the downstream sdhc_2 node has no power-domains/required-opps, so nothing in the device tree raises CX for SD at 202 MHz; whether the downstream driver votes by another path is not visible in these files.

3. ADSP and Q6APM audio

Item	RB5	Nova	Difference
remoteproc_adsp firmware	qcom/sm8550/adsp.mbn, adsp_dtb.mbn (1335-1339)	rp6 381-385 qcom/sm8550/ayn/odin2/..., overridden by nova 21-24 qcom/sm8550/retroidpocket/rpnova/adsp.mbn, adsp_dtb.mbn	Stock Qualcomm images vs device-dumped images.
ADSP memory	inherited: adspslpi_mem 0x9ea00000+0x4080000, q6_adsp_dtb_mem 0x9e980000+0x80000 (sm8550.dtsi 3152, qcs8550.dtsi 147-155)	identical, neither region deleted	Same.
remoteproc_cdsp	qcom/sm8550/cdsp.mbn (1341-1345)	common 1480-1484 ayn/cdsp.mbn, nova 26-29 retroidpocket/rpnova/cdsp.mbn	Path only.
sound compatible	qcom,sm8550-sndcard, qcom,sm8450-sndcard (252)	same (343)	Same.
sound model	QCS8550-RB5Gen2 (253)	AYN-Odin2 (353); not overridden in rp6 or nova	Nova card still reports the Odin 2 name.
sound clocks	none	q6prmcc LPASS_CLK_ID_SEN_MI2S_IBIT, clock-names i2s_clk, assigned-clock-rates 1536000 (347-351), pinctrl-0 lpi_i2s3_active (344)	RB5 has no I2S path, so no card-level clock.
audio-routing	SpkrLeft IN->WSA_SPK1 OUT, SpkrRight IN->WSA_SPK2 OUT, VA DMIC0/1->vdd-micb (254-257)	IN1_HPHL/IN2_HPHR, AMIC2->MIC BIAS2, TX SWR_INPUT1->ADC2_OUTPUT (354-358), nothing for the amps	Different codecs.
Speaker link	wsa-dai-link "WSA Playback": cpu q6apmbedai WSA_CODEC_DMA_RX_0, codec left_spkr, right_spkr, swr0 0, lpass_wsamacro 0, platform q6apm (259-274)	speaker-i2s-dai-link "Primary MI2S Playback": cpu q6apmbedai PRIMARY_MI2S_RX, codec spk_amp_l 0, spk_amp_r 0, platform q6apm (360-374)	RB5: SoundWire WSA883x via wsamacro. Nova: I2S to AW88166. Nova's cpu DAI is PRIMARY_MI2S_RX, its clock SEN_MI2S_IBIT, its pins i2s3_* (lpass_tlmm 1044-1078, all output-high).
Capture link	va-dai-link: lpass_vamacro 0 / VA_CODEC_DMA_TX_0 (276-290)	wcd-capture-dai-link: wcd938x 1, swr2 0, lpass_txmacro 0 / TX_CODEC_DMA_TX_3 (392-406)	RB5 uses on-SoC DMICs via vamacro; Nova the WCD9385 headset codec.
Other links	none	wcd-playback-dai-link RX_CODEC_DMA_RX_0 (376-390), dp0-dai-link DISPLAY_PORT_RX_0 -> mdss_dp0 (408-421)	Nova only.
lpass_vamacro	pinctrl-0 dmic01_default, qcom,dmic-sample-rate 4800000, vdd-micb-supply l15b (1011-1018)	qcom,dmic-sample-rate 4800000 only (1080-1082)	Nova has no DMIC pins or micbias supply on vamacro.
lpass_wsamacro	used (264)	status = "disabled" (1084-1086)	Nova disables the unused macro.
SoundWire	swr0 okay, two sdw20217020400 speakers, reset tlmm 133, vdd-1p8/vdd-io l15b, qcom,port-mapping (1387-1417)	swr1 codec@0,4 rx-port-mapping 1 2 3 4 5, swr2 codec@0,3 tx-port-mapping 2 2 3 4 (1507-1523)	Different buses, different devices.
Amps	n/a	i2c_hub_2 0x34/0x35 awinic,aw88166, reset tlmm 103/100, awinic,audio-channel 0/1, awinic,sync-flag, firmware-name .../aw883xx_acf.bin (common 1011-1029, nova 31-37)	Nova only.

Canonical RB5 Q6APM card for reference: compatible qcom,sm8550-sndcard; each link has link-name, a cpu on q6apmbedai <BE id>, a codec list ending in the SoundWire controller and the LPASS macro, and platform = <&q6apm>; no clocks or pinctrl on the card node; routing from the codec widgets to the macro outputs.

Verdict: nothing to take for the I2S path (RB5 has none). Worth checking: the Nova card node still says model = "AYN-Odin2", and the PRIMARY_MI2S_RX / SEN_MI2S_IBIT / i2s3_* naming mix has no RB5 counterpart to confirm against.

4. SD card (sdhc_2)

Property	RB5 (1353-1367 + sm8550.dtsi 3590-3641)	Nova vendor node (common 1486-1501)	Nova effective node (rp6 412-494, downstream driver)
compatible	qcom,sm8550-sdhci, qcom,sdhci-msm-v5	same	qcom,sdhci-msm-v5-downstream
cd-gpios	pm8550_gpios 12 ACTIVE_LOW (1354)	same (1487)	same (465)
pinctrl	default sdc2_default+sdc2_card_det_n, sleep sdc2_sleep+sdc2_card_det_n (1356-1358)	same (1488-1490)	same (461-463)
sdc2_card_det_n pin state	gpio12 normal, power-source 1, bias-pull-up, input-enable (1258-1264)	same plus output-disable (1169-1176)	uses the common one
regulators	vmmc-supply l9b, vqmmc-supply l8b (1360-1361)	same (1491-1492)	vdd-supply l9b, vdd-io-supply l8b plus qcom,vdd-voltage-level, qcom,vdd-io-voltage-level, current levels (453-459)
bus-width	4 (inherited 3613)	4 (inherited)	4 (422)
no-sdio / no-mmc	both (1363-1364)	both (1494-1495)	both (423-424)
max-sd-hs-hz	37500000 (inherited 3614)	37500000 (1493)	absent
DLL config	qcom,dll-config 0x0007642c, qcom,ddr-config 0x80040868 (inherited 3603-3604)	qcom,dll-config 0x0007442c (1497), sdhci-caps-mask <0x3 0x0> (1498)	qcom,dll-hsr-list 0x0007442C 0x0 0x10 0x090106C0 0x80040868 (436-437)
power-domains / OPP	rpmhpd RPMHPD_CX, OPPs 19.2/50/100/202 MHz with required-opps (inherited 3605-3640)	inherited	own opp-table 100/202 MHz with bandwidth only, no required-opps, no power-domains (451, 479-493)
clocks	iface, core, xo (inherited 3598-3601)	inherited	iface, core only (428-430)
interconnects	ICC tags ALWAYS / ACTIVE_ONLY (inherited 3608-3612)	inherited	tags 0, plus qcom,msm-bus,* vectors (441-449)
extra	-	-	qcom,restore-after-cx-collapse, qcom,uses_level_shifter, qcom,dll_lock_bist_fail_wa, resets GCC_SDCC2_BCR, qos0/qos1 (425-427, 466-476)

Properties present on RB5's effective node that Nova's effective node lacks: power-domains, required-opps, max-sd-hs-hz, qcom,ddr-config, the xo clock, ICC tags. Nova lacks nothing RB5 sets at board level; the board-level lines are the same.

Verdict: nothing to take at board level; both set identical cd-gpios, pinctrl, supplies and no-sdio/no-mmc. The differences are in the Nova's node replacement (rp6 395-411 explains why), see section 2 for the missing CX vote.

5. USB

Item	RB5	Nova	Difference
pmic-glink	compatible qcom,sm8550-pmic-glink, no orientation-gpios (111-114)	same compatible, orientation-gpios = <&tlmm 11 GPIO_ACTIVE_HIGH> (58-62)	RB5 gets orientation from the nb7vpq904m retimer; Nova from a GPIO.
connector	usb-c-connector, power-role dual, data-role dual (116-120)	same (64-68)	Same.
port@0 (HS)	usb_1_dwc3_hs (129)	usb_1_dwc3_hs (78)	Same.
port@1 (SS)	redriver_usb_con_ss on the retimer (137)	usb_dp_qmpphy_out directly (86)	RB5 has a retimer in the SS path.
port@2 (SBU)	redriver_usb_con_sbu (145)	usb0_sbu_mux (94), a gpio-sbu-mux with enable tlmm 140 low, select tlmm 141, mode-switch, orientation-switch (424-441)	Different SBU hardware.
retimer / mux	i2c_hub_2 typec-mux@1c onnn,nb7vpq904m, vcc-supply l15b, retimer-switch, orientation-switch, three ports; redriver_phy_con_ss <- usb_dp_qmpphy_out with data-lanes 0 1 2 3 (911-949, 1567-1569)	none; usb_dp_qmpphy_out -> pmic_glink_ss_in (1794-1796)	RB5 only.
usb_1	status okay only (1543-1545)	same (1768-1770)	Both inherit usb-role-switch (sm8550.dtsi 4582), no dr_mode, no maximum-speed, PDC wake IRQs 14/15/17 (4541-4543).
usb_1_hsphy	vdd l1e, vdda12 l3e, phys pm8550b_eusb2_repeater (1551-1558)	same (1776-1783)	Same.
pm8550b_eusb2_repeater	vdd18 l15b, vdd3 l5b (1301-1304)	same plus qcom,tune-usb2-disc-thres 0x6, tune-usb2-amplitude 0xb, tune-usb2-preem 0x3 (1421-1427)	Nova carries eye-tuning values; RB5 uses defaults.
usb_dp_qmpphy	vdda-phy l3e, vdda-pll l3f (1560-1565)	same plus mode-switch (1785-1792)	mode-switch is already in sm8550.dtsi 4483; Nova's line is redundant.
wakeup on connector/typec	none	none	Same.

Verdict: nothing to take. The RB5 USB graph differs only where it has a retimer; everything the two boards share (HS endpoint, PHY supplies, repeater supplies, role switch) is identical.

6. Wi-Fi / Bluetooth

Item	RB5	Nova	Difference
wcn7850-pmu pinctrl	wlan_en, bt_default, sw_ctrl_default, pmk8550_sleep_clk (296-298)	wlan_en, bt_default, pmk8550_sleep_clk (513-514); bt_default contains both gpio81 and gpio82 (1584-1597)	Same pins; Nova folds SW_CTRL into bt_default.
GPIOs	wlan-enable tlmm 80, bt-enable tlmm 81, swctrl-gpios tlmm 82 (300-302)	wlan-enable tlmm 80, bt-enable tlmm 81; no swctrl-gpios (516-517)	Nova configures gpio82 as a pull-down pin but does not hand it to the PMU driver.
PMU supplies	vdd s5g, vddio l15b, vddaon s2g, vdddig s4e, vddrfa1p2 s4g, vddrfa1p8 s6g (304-309)	same six plus vddio1p2-supply = <&vreg_l3g_1p2> (519-525)	Nova adds vddio1p2.
PMU LDOs	ldo0-ldo9, vreg_pmu_rfa_cmn .. vreg_pmu_pcie_1p8 (311-351)	identical (527-567)	Same.
pcie0	vddpe-3v3-supply pcie_upd_3p3, wake-gpios tlmm 96 ACTIVE_HIGH, perst-gpios tlmm 94 ACTIVE_LOW, pinctrl pcie0_default_state, iommu-map for a TC9563 switch, /delete-property/ msi-map (1045-1067)	pinctrl pcie0_default_state only (1122-1127)	RB5 keeps wake/perst on the controller (old binding) and has a PCIe switch on pcie0.
root port	pcie0_port0 holds a pci1179,0623 switch (1076-1146); Wi-Fi is on pcie1 behind a second switch: pcie1_port0/pcie@0,0/pcie@2,0/wifi@0 (1178-1247)	pcieport0: wake-gpios tlmm 96 ACTIVE_LOW (comment 1130-1131), reset-gpios tlmm 94 ACTIVE_LOW, wifi@0 directly (1129-1149)	Nova uses the root-port binding and inverts WAKE# polarity relative to RB5.
wifi@0	compatible pci17cb,1107, reg <0x40000 ...>, 9 vdd*-supply from the PMU LDOs (1222-1235)	pci17cb,1107, reg <0x10000 ...>, same 9 supplies (1135-1147)	Same supplies; reg differs with bus position.
qcom,ath12k-calibration-variant	none	none	Neither.
wifi firmware-name	none	none	Neither.
pcie0_phy	vdda-phy l1e, vdda-pll l3e (1069-1074)	same (1151-1156)	Same.
uart14 bluetooth	qcom,wcn7850-bt, 7 supplies vddrfacmn .. vddrfa1p8 (1509-1523)	same 7 supplies plus max-speed = <3200000> (1713-1729)	Nova sets the UART speed.
LEDs	blue:bt-power trigger bluetooth-power, yellow:wlan trigger phy0tx (92-108)	none	RB5 only.

Verdict: nothing to take. Shared properties are identical or Nova's are a superset (vddio1p2, max-speed). Worth checking: RB5 passes swctrl-gpios (gpio82) to the PMU, Nova only pins it down; and RB5 has wake-gpios ACTIVE_HIGH where Nova changed it to ACTIVE_LOW with a stated reason.

7. Wake sources

Item	RB5	Nova	Difference
gpio-keys volume-up	linux,can-disable, wakeup-source, debounce 15, pm8550_gpios 6 (49-56)	same (48-55)	Same.
paddles	n/a	rp6 137-149, no wakeup-source	Nova only.
pon_pwrkey / pon_resin	okay; resin KEY_VOLUMEDOWN (1317-1325)	same (1462-1470)	Same.
wakeup-event-action	none	none	Neither.
interrupts-extended on &pdc in board files	none; the two uses are tlmm 40 (LT9611, 961) and tlmm 55 (CAN, 1379)	none; tlmm 15 touch (rp6 358-359), tlmm 13 fan (111-112) via interrupt-parent	Neither board file adds PDC wake interrupts; both inherit the SoC ones (adsp pdc 6 3134, usb pdc 14/15/17 4541-4543, tsens).
wakeup-source on uart14 / connector / pmic-glink	none	none	Neither.

Verdict: nothing. The only board-level wake source on either tree is the volume-up key, configured identically.

8. reserved-memory

Both include qcs8550.dtsi, which deletes the sm8550 reserved_memory and defines 22 regions (qcs8550.dtsi 59-160). RB5 makes no changes. Nova:

Deleted by Nova (common 17-24), kept by RB5:

Region	Address	Size
aop_image_mem	0x81c00000	0x60000
aop_config_mem	0x81c80000	0x20000
camera_mem	0x9b300000	0x800000
ipa_fw_mem	0x9b080000	0x10000
ipa_gsi_mem	0x9b090000	0xa000
mpss_dsm_mem	0xd4d00000	0x3300000
mpss_mem	0x8a800000	0x10800000
q6_mpss_dtb_mem	0x9b000000	0x80000

Nova also deletes remoteproc_mpss (common 26); RB5 enables it with modem.mbn/modem_dtb.mbn (1347-1351) and enables ipa with ipa_fws.mbn (1001-1005).

Added by Nova (common 118-204), absent on RB5:

Region	Address	Size
hyp_mem	0x80000000	0xa00000
cpusys_vm_mem	0x80a00000	0x400000
hyp_tags_mem	0x80e00000	0x3d0000
hyp_tags_reserved_mem	0x811d0000	0x30000
xbl_dt_log_merged_mem	0x81a00000	0x260000
aop_config_merged_mem	0x81c80000	0x74000 (replaces the deleted 0x20000 aop_config_mem)
chipinfo_mem	0x81cf4000	0x1000
global_sync_mem	0x82600000	0x100000
tz_stat_mem	0x82700000	0x100000
splash_region	0xb8000000	0x2b00000 (label = "cont_splash_region")
xbl_sc_mem	0xd8100000	0x40000
cpucp_fw_mem	0xd8140000	0x1c0000
qtee_mem	0xd8300000	0x500000
hwfence_shbuf	0xe6440000	0x2dd000
hyp_ext_tags_mem	0xfce00000	0x2900000
hyp_ext_reserved_mem	0xff700000	0x100000
llcc_lpi_mem	0xff800000	0x600000

Unchanged on both: aop_cmd_db_mem, smem_mem, adsp_mhi_mem, gpu_micro_code_mem, spss_region_mem, spu_secure_shared_memory_mem, video_mem, cvp_mem, cdsp_mem, q6_cdsp_dtb_mem, q6_adsp_dtb_mem, adspslpi_mem.

Verdict: nothing to take. RB5 relies on the qcs8550.dtsi comment (17-38) that UEFI/ESRT reserves the firmware regions at runtime and lists none itself; Nova lists them explicitly. The deletions on Nova are all modem/IPA/camera, which RB5 has and Nova does not.

9. Everything else RB5 has that Nova lacks

Item	RB5	Nova	Applies to a handheld?
Thermal zones	none added; stock sm8550.dtsi zones (6119-7173, cpu passive/critical trips, GPU #cooling-cells 2863)	adds trips to cpuss0-3, cpu7-top, gpuss-0..7 at 50-80 C (206-340) with polling-delay 200; no cooling-maps anywhere in the three files (grep); pwm_fan has #cooling-cells (115) but nothing maps to it	Nothing from RB5 to take; Nova's extra trips drive nothing in these files.
ADC channels / adc-tm	none	none	Neither declares PMIC ADC channels.
cpufreq / interconnect	none at board level	none except the downstream sdhc_2 ICC lines	Nothing.
UFS	reset-gpios tlmm 210, vcc l17b 1300000 uA, vccq l1g 1200000 uA, vdd-hba l3g; phy vdda-phy l1d, vdda-pll l3e (1525-1541)	identical (1750-1766)	Same. Nova also disables ice (1036-1038, comment cites #505); RB5 leaves it stock.
GPU zap shader	&gpu_zap_shader { firmware-name = "qcom/sm8550/a740_zap.mbn" } (902-904)	&gpu { zap-shader { firmware-name = "qcom/sm8550/a740_zap.mbn" } } (983-985)	Same path, different spelling (label vs nested node).
firmware-name paths	all stock qcom/sm8550/*.mbn (adsp, cdsp, modem, ipa, zap)	adsp/cdsp under qcom/sm8550/retroidpocket/rpnova/, amp aw883xx_acf.bin, zap stock	Nova uses device-dumped DSP images.
Display	mdss_dsi0 -> LT9611UXC HDMI bridge on bit-banged I2C (72-79, 952-991, 1024-1039), mdss_dp0 okay (1041-1043)	mdss_dsi0 panel il97680a,rpnova (nova 39-50; rp6 162-199), mdss_dp0 okay with sound-name-prefix (1092-1095)	Same DSI/DP enables; different sink.
mdss_dsi0 vdda-supply / phy vdds-supply	l3e / l1e (1025, 1036)	l3e / l1e (rp6 163, 197)	Same.
tlmm gpio-reserved-ranges	<32 8> (1422)	<32 8> (1526)	Same.
sleep_clk / xo_board	32764 / 76800000 (1369-1371, 1571-1573)	same (1503-1505, 1798-1800)	Same.
pm8550_pwm	three green status LEDs led@1-3 (1267-1299)	RGB power_led red@1/green@2/blue@3 (1187-1212); channel 3 also used by pwm_fan via pm8550_pwm 3 40000 (105)	Different use of the same PWM channels.
Peripherals absent on Nova	CAN mcp2518fd on spi11 (1373-1385), two TC9563 PCIe switches, pcie1 (1148-1176), LEDs (81-109), ipa, modem, pm8010, i2c_hub_4 (993-995)	-	Dev-board hardware, not applicable.

Verdict: nothing to take. Every shared SoC-level setting (UFS, zap shader path, DSI/DP supplies, clocks, reserved GPIO range) is already identical; the rest of RB5's additions are for hardware the Nova does not have. Worth checking on Nova's own side only: the added thermal trips have no cooling-maps in these files, and the card model string is still the Odin 2's.