linux/drivers/acpi/cppc_acpi.c, branch v6.17

ACPI: CPPC: Fix NULL pointer dereference when nosmp is used

2025-06-10T18:52:50Z

With nosmp in cmdline, other CPUs are not brought up, leaving their cpc_desc_ptr NULL. CPU0's iteration via for_each_possible_cpu() dereferences these NULL pointers, causing panic. Panic backtrace: [ 0.401123] Unable to handle kernel NULL pointer dereference at virtual address 00000000000000b8 ... [ 0.403255] [] cppc_allow_fast_switch+0x6a/0xd4 ... Kernel panic - not syncing: Attempted to kill init! Fixes: 3cc30dd00a58 ("cpufreq: CPPC: Enable fast_switch") Reported-by: Xu Lu Signed-off-by: Yunhui Cui Link: https://patch.msgid.link/20250604023036.99553-1-cuiyunhui@bytedance.com [ rjw: New subject ] Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Add three functions related to autonomous selection

2025-04-30T20:01:31Z

cppc_set_epp() - write energy performance preference register value, based on ACPI 6.5, s8.4.6.1.7 cppc_get_auto_act_window() - read autonomous activity window register value, based on ACPI 6.5, s8.4.6.1.6 cppc_set_auto_act_window() - write autonomous activity window register value, based on ACPI 6.5, s8.4.6.1.6 Reviewed-by: Pierre Gondois Signed-off-by: Lifeng Zheng Reviewed-by: Mario Limonciello Link: https://patch.msgid.link/20250411093855.982491-9-zhenglifeng1@huawei.com Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Modify cppc_get_auto_sel_caps() to cppc_get_auto_sel()

2025-04-30T20:01:31Z

Modify cppc_get_auto_sel_caps() to cppc_get_auto_sel(). Using a cppc_perf_caps to carry the value is unnecessary. Add a check to ensure the pointer 'enable' is not null. Reviewed-by: Pierre Gondois Signed-off-by: Lifeng Zheng Reviewed-by: Mario Limonciello Link: https://patch.msgid.link/20250411093855.982491-8-zhenglifeng1@huawei.com Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Refactor register value get and set ABIs

2025-04-30T20:01:31Z

Refactor register value get and set ABIs by using cppc_get_reg_val(), cppc_set_reg_val() and CPPC_REG_VAL_READ(). Reviewed-by: Pierre Gondois Signed-off-by: Lifeng Zheng Reviewed-by: Mario Limonciello Link: https://patch.msgid.link/20250411093855.982491-7-zhenglifeng1@huawei.com Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Add cppc_set_reg_val()

2025-04-30T20:01:30Z

Add cppc_set_reg_val() as a generic function for setting CPPC register values, with this features: 1. Check register. If a register is writeable, it must be a buffer and can not be null. 2. Extract the operations if register is in PCC out as cppc_set_reg_val_in_pcc(). This function can be used to reduce some existing code duplication. Reviewed-by: Pierre Gondois Signed-off-by: Lifeng Zheng Reviewed-by: Mario Limonciello Link: https://patch.msgid.link/20250411093855.982491-6-zhenglifeng1@huawei.com Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Extract cppc_get_reg_val_in_pcc()

2025-04-30T20:01:30Z

Extract the operations if register is in pcc out from cppc_get_reg_val() as cppc_get_reg_val_in_pcc(). Reviewed-by: Pierre Gondois Signed-off-by: Lifeng Zheng Reviewed-by: Mario Limonciello Link: https://patch.msgid.link/20250411093855.982491-5-zhenglifeng1@huawei.com Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Rename cppc_get_perf() to cppc_get_reg_val()

2025-04-30T20:01:30Z

Rename cppc_get_perf() to cppc_get_reg_val() as a generic function to read CPPC registers. Reviewed-by: Pierre Gondois Signed-off-by: Lifeng Zheng Reviewed-by: Mario Limonciello Link: https://patch.msgid.link/20250411093855.982491-4-zhenglifeng1@huawei.com Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Optimize cppc_get_perf()

2025-04-30T20:01:30Z

Optimize cppc_get_perf() with three changes: 1. Change the error kind to "no such device" when pcc_ss_id < 0, as other register value getting functions. 2. Add a check to ensure the pointer 'perf' is no null. 3. Add a check to verify if the register is supported to be read before using it. The logic is: (1) If the register is of the integer type, check whether the register is optional and its value is 0. If yes, the register is not supported. (2) If the register is of other types, a null one is not supported. 4. Return the result of cpc_read() instead of 0. Reviewed-by: Pierre Gondois Signed-off-by: Lifeng Zheng Reviewed-by: Mario Limonciello Link: https://patch.msgid.link/20250411093855.982491-3-zhenglifeng1@huawei.com Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Add IS_OPTIONAL_CPC_REG macro to judge if a cpc_reg is optional

2025-04-30T20:01:30Z

In ACPI 6.5, s8.4.6.1 _CPC (Continuous Performance Control), whether each of the per-cpu cpc_regs[] is mandatory or optional is defined. Since the CPC_SUPPORTED() check is only for optional _CPC fields, another macro to check if the field is optional is needed. Reviewed-by: Pierre Gondois Signed-off-by: Lifeng Zheng Reviewed-by: Mario Limonciello Link: https://patch.msgid.link/20250411093855.982491-2-zhenglifeng1@huawei.com Signed-off-by: Rafael J. Wysocki

ACPI: CPPC: Simplify PCC shared memory region handling

2025-04-30T19:51:23Z

The PCC driver now handles mapping and unmapping of shared memory areas as part of pcc_mbox_{request,free}_channel(). Without these before, this ACPI CPPC driver did handling of those mappings like several other PCC mailbox client drivers. There were redundant operations, leading to unnecessary code. Maintaining the consistency across these driver was harder due to scattered handling of shmem. Just use the mapped shmem and remove all redundant operations from this driver. Signed-off-by: Sudeep Holla Link: https://patch.msgid.link/20250411113144.1151094-2-sudeep.holla@arm.com Signed-off-by: Rafael J. Wysocki