kernel/linux.git/drivers/pmdomain/mediatek, branch v6.18.22

pmdomain: mtk-pm-domains: Fix spinlock recursion fix in probe

2026-01-08T09:17:09+00:00

commit 305f254727bd379bbed0385afa0162f5bde1f51c upstream. Remove scpsys_get_legacy_regmap(), replacing its usage with of_find_node_with_property(). Explicitly call of_node_get(np) before each of_find_node_with_property() to maintain correct node reference counting. The of_find_node_with_property() function "consumes" its input by calling of_node_put() internally, whether or not it finds a match. Currently, dev->of_node (np) is passed multiple times in sequence without incrementing its reference count, causing it to be decremented multiple times and risking early memory release. Adding of_node_get(np) before each call balances the reference count, preventing premature node release. Fixes: c1bac49fe91f ("pmdomains: mtk-pm-domains: Fix spinlock recursion in probe") Cc: stable@vger.kernel.org Signed-off-by: Macpaul Lin Tested-by: Louis-Alexis Eyraud Signed-off-by: Ulf Hansson Signed-off-by: Greg Kroah-Hartman

pmdomains: mtk-pm-domains: Fix spinlock recursion in probe

2025-11-24T10:04:17+00:00

For some reason, of_find_node_with_property() is creating a spinlock recursion issue along with fwnode_count_parents(), and this issue is making all MediaTek boards unbootable. As of kernel v6.18-rc6, there are only three users of this function, one of which is this driver. Migrate away from of_find_node_with_property() by adding a local scpsys_get_legacy_regmap_node() function, which acts similarly to of_find_node_with_property(), and calling the former in place of the latter. This resolves the following spinlock recursion issue: [ 1.773979] BUG: spinlock recursion on CPU#2, kworker/u24:1/60 [ 1.790485] lock: devtree_lock+0x0/0x40, .magic: dead4ead, .owner: kworker/u24:1/60, .owner_cpu: 2 [ 1.791644] CPU: 2 UID: 0 PID: 60 Comm: kworker/u24:1 Tainted: G W 6.18.0-rc6 #3 PREEMPT [ 1.791649] Tainted: [W]=WARN [ 1.791650] Hardware name: MediaTek Genio-510 EVK (DT) [ 1.791653] Workqueue: events_unbound deferred_probe_work_func [ 1.791658] Call trace: [ 1.791659] show_stack+0x18/0x30 (C) [ 1.791664] dump_stack_lvl+0x68/0x94 [ 1.791668] dump_stack+0x18/0x24 [ 1.791672] spin_dump+0x78/0x88 [ 1.791678] do_raw_spin_lock+0x110/0x140 [ 1.791684] _raw_spin_lock_irqsave+0x58/0x6c [ 1.791690] of_get_parent+0x28/0x74 [ 1.791694] of_fwnode_get_parent+0x38/0x7c [ 1.791700] fwnode_count_parents+0x34/0xf0 [ 1.791705] fwnode_full_name_string+0x28/0x120 [ 1.791710] device_node_string+0x3e4/0x50c [ 1.791715] pointer+0x294/0x430 [ 1.791718] vsnprintf+0x21c/0x5bc [ 1.791722] vprintk_store+0x108/0x47c [ 1.791728] vprintk_emit+0xc4/0x350 [ 1.791732] vprintk_default+0x34/0x40 [ 1.791736] vprintk+0x24/0x30 [ 1.791740] _printk+0x60/0x8c [ 1.791744] of_node_release+0x154/0x194 [ 1.791749] kobject_put+0xa0/0x120 [ 1.791753] of_node_put+0x18/0x28 [ 1.791756] of_find_node_with_property+0x74/0x100 [ 1.791761] scpsys_probe+0x338/0x5e0 [ 1.791765] platform_probe+0x5c/0xa4 [ 1.791770] really_probe+0xbc/0x2ac [ 1.791774] __driver_probe_device+0x78/0x118 [ 1.791779] driver_probe_device+0x3c/0x170 [ 1.791783] __device_attach_driver+0xb8/0x150 [ 1.791788] bus_for_each_drv+0x88/0xe8 [ 1.791792] __device_attach+0x9c/0x1a0 [ 1.791796] device_initial_probe+0x14/0x20 [ 1.791801] bus_probe_device+0xa0/0xa4 [ 1.791805] deferred_probe_work_func+0x88/0xd0 [ 1.791809] process_one_work+0x1e8/0x448 [ 1.791813] worker_thread+0x1ac/0x340 [ 1.791816] kthread+0x138/0x220 [ 1.791821] ret_from_fork+0x10/0x20 Fixes: c29345fa5f66 ("pmdomain: mediatek: Refactor bus protection regmaps retrieval") Signed-off-by: AngeloGioacchino Del Regno Tested-by: Louis-Alexis Eyraud Tested-by: Macpaul Lin Reviewed-by: Macpaul Lin Signed-off-by: Ulf Hansson

pmdomain: Merge branch fixes into next

2025-09-23T14:35:23+00:00

Merge the pmdomain fixes for v6.17-rc[n] into the next branch, to allow them to get tested together with the new changes that are targeted for v6.18. Signed-off-by: Ulf Hansson

pmdomain: mediatek: set default off flag for MT8195 AUDIO power domain

2025-09-23T14:33:34+00:00

In MT8195 power domain data array, set the KEEP_DEFAULT_OFF and ACTIVE_WAKEUP flags for the AUDIO power domain entry to avoid having this domain being on during boot sequence when unneeded. Fixes: 0e789b491ba0 ("pmdomain: core: Leave powered-on genpds on until sync_state") Fixes: 13a4b7fb6260 ("pmdomain: core: Leave powered-on genpds on until late_initcall_sync") Signed-off-by: Louis-Alexis Eyraud Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Ulf Hansson

pmdomain: mediatek: airoha: convert from round_rate() to determine_rate()

2025-09-17T14:28:18+00:00

The round_rate() clk ops is deprecated, so migrate this driver from round_rate() to determine_rate() using the Coccinelle semantic patch appended to the "under-the-cut" portion of the patch. Note that prior to running the Coccinelle, airoha_cpu_pmdomain_clk_round() was renamed to airoha_cpu_pmdomain_clk_round_rate(). Signed-off-by: Brian Masney Signed-off-by: Ulf Hansson

pmdomain: mediatek: Convert all SoCs to new style regmap retrieval

2025-08-19T12:12:41+00:00

Add the bus_prot_blocks handle and declare num_bus_prot_blocks to allow all of the currently supported AArch64 MediaTek SoCs to use the new style regmap retrieval in the driver when a new style devicetree declaring the mediatek,bus-protection phandle(s) in the main power controller node is found. Reviewed-by: Nícolas F. R. A. Prado Signed-off-by: AngeloGioacchino Del Regno Link: https://lore.kernel.org/r/20250805074746.29457-10-angelogioacchino.delregno@collabora.com Signed-off-by: Ulf Hansson

pmdomain: mediatek: Add support for RTFF Hardware in MT8196/MT6991

2025-08-19T12:12:40+00:00

New generation SoCs use a new RTFF Hardware to save power during operation of various IPs, other than managing isolation of the internal buck converters during powerup/down of power domains. Since some of the power domains need different RTFF handling, add a new scpys_rtff_type enumeration and hold the value for each power domain in struct scpsys_domain_data. If RTFF HW is available, the RTFF additional power sequences are handled in scpsys_ctl_pwrseq_{on,off}(). Reviewed-by: Nícolas F. R. A. Prado Signed-off-by: AngeloGioacchino Del Regno Link: https://lore.kernel.org/r/20250805074746.29457-9-angelogioacchino.delregno@collabora.com Signed-off-by: Ulf Hansson

pmdomain: mediatek: Add support for modem power sequences

2025-08-19T12:12:40+00:00

Add support for the modem power domains by adding its specific power sequence in functions scpsys_modem_pwrseq_{on,off}() and call them if the flag MTK_SCPD_MODEM_PWRSEQ is present. While at it, since some SoC models need to skip setting/clearing the PWR_RST_B_BIT, also add a MTK_SCPD_SKIP_RESET_B flag for that. Reviewed-by: Nícolas F. R. A. Prado Signed-off-by: AngeloGioacchino Del Regno Link: https://lore.kernel.org/r/20250805074746.29457-8-angelogioacchino.delregno@collabora.com Signed-off-by: Ulf Hansson

pmdomain: mediatek: Move ctl sequences out of power_on/off functions

2025-08-19T12:12:40+00:00

In preparation to support power domains of new SoCs and the modem power domains for both new and already supported chips, move the generic control power sequences out of the scpsys_power_on() and scpsys_power_off() and put them in new scpsys_ctl_pwrseq_on(), scpsys_ctl_pewseq_off() functions. Reviewed-by: Nícolas F. R. A. Prado Signed-off-by: AngeloGioacchino Del Regno Link: https://lore.kernel.org/r/20250805074746.29457-7-angelogioacchino.delregno@collabora.com Signed-off-by: Ulf Hansson

pmdomain: mediatek: Handle SoCs with inverted SRAM power-down bits

2025-08-19T12:12:40+00:00

Some SoCs, and even some subsystems in the same SoC, may have the logic for SRAM power-down inverted, as in, setting the bit means "power down" and unsetting means "power up": this is because some hardware subsystems use this as a power-lock indication and some use this as a power down one (for example, usually, the modem ss has it inverted!). In preparation for adding support for power domains with inverted SRAM_PDN bits, add a new MTK_SCPD_SRAM_PDN_INVERTED flag and check for it in scpsys_sram_enable() and scpsys_sram_disable(). Reviewed-by: Nícolas F. R. A. Prado Signed-off-by: AngeloGioacchino Del Regno Link: https://lore.kernel.org/r/20250805074746.29457-6-angelogioacchino.delregno@collabora.com Signed-off-by: Ulf Hansson