user/sven/linux.git/drivers/cpufreq, branch v6.1.59

cpufreq: brcmstb-avs-cpufreq: Fix -Warray-bounds bug

2023-09-13T07:43:04Z

commit e520d0b6be950ce3738cf4b9bd3b392be818f1dc upstream. Allocate extra space for terminating element at: drivers/cpufreq/brcmstb-avs-cpufreq.c: 449 table[i].frequency = CPUFREQ_TABLE_END; and add code comment to make this clear. This fixes the following -Warray-bounds warning seen after building ARM with multi_v7_defconfig (GCC 13): In function 'brcm_avs_get_freq_table', inlined from 'brcm_avs_cpufreq_init' at drivers/cpufreq/brcmstb-avs-cpufreq.c:623:15: drivers/cpufreq/brcmstb-avs-cpufreq.c:449:28: warning: array subscript 5 is outside array bounds of 'void[60]' [-Warray-bounds=] 449 | table[i].frequency = CPUFREQ_TABLE_END; In file included from include/linux/node.h:18, from include/linux/cpu.h:17, from include/linux/cpufreq.h:12, from drivers/cpufreq/brcmstb-avs-cpufreq.c:44: In function 'devm_kmalloc_array', inlined from 'devm_kcalloc' at include/linux/device.h:328:9, inlined from 'brcm_avs_get_freq_table' at drivers/cpufreq/brcmstb-avs-cpufreq.c:437:10, inlined from 'brcm_avs_cpufreq_init' at drivers/cpufreq/brcmstb-avs-cpufreq.c:623:15: include/linux/device.h:323:16: note: at offset 60 into object of size 60 allocated by 'devm_kmalloc' 323 | return devm_kmalloc(dev, bytes, flags); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ This helps with the ongoing efforts to tighten the FORTIFY_SOURCE routines on memcpy() and help us make progress towards globally enabling -Warray-bounds. Link: https://github.com/KSPP/linux/issues/324 Fixes: de322e085995 ("cpufreq: brcmstb-avs-cpufreq: AVS CPUfreq driver for Broadcom STB SoCs") Cc: stable@vger.kernel.org Signed-off-by: Gustavo A. R. Silva Reviewed-by: Florian Fainelli Signed-off-by: Viresh Kumar Signed-off-by: Greg Kroah-Hartman

cpufreq: Fix the race condition while updating the transition_task of policy

2023-09-13T07:42:59Z

[ Upstream commit 61bfbf7951ba561dcbdd5357702d3cbc2d447812 ] The field 'transition_task' of policy structure is used to track the task which is performing the frequency transition. Using this field to print a warning once detect a case where the same task is calling _begin() again before completing the preivous frequency transition via the _end(). However, there is a potential race condition in _end() and _begin() APIs while updating the field 'transition_task' of policy, the scenario is depicted below: Task A Task B /* 1st freq transition */ Invoke _begin() { ... ... } /* 2nd freq transition */ Invoke _begin() { ... //waiting for A to ... //clear ... //transition_ongoing ... //in _end() for ... //the 1st transition | Change the frequency | | Invoke _end() { | ... | ... | transition_ongoing = false; V transition_ongoing = true; transition_task = current; transition_task = NULL; ... //A overwrites the task ... //performing the transition ... //result in error warning. } To fix this race condition, the transition_lock of policy structure is now acquired before updating policy structure in _end() API. Which ensure that only one task can update the 'transition_task' field at a time. Link: https://lore.kernel.org/all/b3c61d8a-d52d-3136-fbf0-d1de9f1ba411@huawei.com/ Fixes: ca654dc3a93d ("cpufreq: Catch double invocations of cpufreq_freq_transition_begin/end") Signed-off-by: Liao Chang Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki Signed-off-by: Sasha Levin

cpufreq: powernow-k8: Use related_cpus instead of cpus in driver.exit()

2023-09-13T07:42:30Z

[ Upstream commit 03997da042dac73c69e60d91942c727c76828b65 ] Since the 'cpus' field of policy structure will become empty in the cpufreq core API, it is better to use 'related_cpus' in the exit() callback of driver. Fixes: c3274763bfc3 ("cpufreq: powernow-k8: Initialize per-cpu data-structures properly") Signed-off-by: Liao Chang Signed-off-by: Viresh Kumar Signed-off-by: Sasha Levin

cpufreq: amd-pstate-ut: Fix kernel panic when loading the driver

2023-09-13T07:42:29Z

[ Upstream commit 60dd283804479c4a52f995b713f448e2cd65b8c8 ] After loading the amd-pstate-ut driver, amd_pstate_ut_check_perf() and amd_pstate_ut_check_freq() use cpufreq_cpu_get() to get the policy of the CPU and mark it as busy. In these functions, cpufreq_cpu_put() should be used to release the policy, but it is not, so any other entity trying to access the policy is blocked indefinitely. One such scenario is when amd_pstate mode is changed, leading to the following splat: [ 1332.103727] INFO: task bash:2929 blocked for more than 120 seconds. [ 1332.110001] Not tainted 6.5.0-rc2-amd-pstate-ut #5 [ 1332.115315] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 1332.123140] task:bash state:D stack:0 pid:2929 ppid:2873 flags:0x00004006 [ 1332.123143] Call Trace: [ 1332.123145] [ 1332.123148] __schedule+0x3c1/0x16a0 [ 1332.123154] ? _raw_read_lock_irqsave+0x2d/0x70 [ 1332.123157] schedule+0x6f/0x110 [ 1332.123160] schedule_timeout+0x14f/0x160 [ 1332.123162] ? preempt_count_add+0x86/0xd0 [ 1332.123165] __wait_for_common+0x92/0x190 [ 1332.123168] ? __pfx_schedule_timeout+0x10/0x10 [ 1332.123170] wait_for_completion+0x28/0x30 [ 1332.123173] cpufreq_policy_put_kobj+0x4d/0x90 [ 1332.123177] cpufreq_policy_free+0x157/0x1d0 [ 1332.123178] ? preempt_count_add+0x58/0xd0 [ 1332.123180] cpufreq_remove_dev+0xb6/0x100 [ 1332.123182] subsys_interface_unregister+0x114/0x120 [ 1332.123185] ? preempt_count_add+0x58/0xd0 [ 1332.123187] ? __pfx_amd_pstate_change_driver_mode+0x10/0x10 [ 1332.123190] cpufreq_unregister_driver+0x3b/0xd0 [ 1332.123192] amd_pstate_change_driver_mode+0x1e/0x50 [ 1332.123194] store_status+0xe9/0x180 [ 1332.123197] dev_attr_store+0x1b/0x30 [ 1332.123199] sysfs_kf_write+0x42/0x50 [ 1332.123202] kernfs_fop_write_iter+0x143/0x1d0 [ 1332.123204] vfs_write+0x2df/0x400 [ 1332.123208] ksys_write+0x6b/0xf0 [ 1332.123210] __x64_sys_write+0x1d/0x30 [ 1332.123213] do_syscall_64+0x60/0x90 [ 1332.123216] ? fpregs_assert_state_consistent+0x2e/0x50 [ 1332.123219] ? exit_to_user_mode_prepare+0x49/0x1a0 [ 1332.123223] ? irqentry_exit_to_user_mode+0xd/0x20 [ 1332.123225] ? irqentry_exit+0x3f/0x50 [ 1332.123226] ? exc_page_fault+0x8e/0x190 [ 1332.123228] entry_SYSCALL_64_after_hwframe+0x6e/0xd8 [ 1332.123232] RIP: 0033:0x7fa74c514a37 [ 1332.123234] RSP: 002b:00007ffe31dd0788 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 [ 1332.123238] RAX: ffffffffffffffda RBX: 0000000000000008 RCX: 00007fa74c514a37 [ 1332.123239] RDX: 0000000000000008 RSI: 000055e27c447aa0 RDI: 0000000000000001 [ 1332.123241] RBP: 000055e27c447aa0 R08: 00007fa74c5d1460 R09: 000000007fffffff [ 1332.123242] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000008 [ 1332.123244] R13: 00007fa74c61a780 R14: 00007fa74c616600 R15: 00007fa74c615a00 [ 1332.123247] Fix this by calling cpufreq_cpu_put() wherever necessary. Fixes: 14eb1c96e3a3 ("cpufreq: amd-pstate: Add test module for amd-pstate driver") Reviewed-by: Mario Limonciello Reviewed-by: Meng Li Reviewed-by: Wyes Karny Suggested-by: Wyes Karny Signed-off-by: Swapnil Sapkal [ rjw: Subject and changelog edits ] Signed-off-by: Rafael J. Wysocki Signed-off-by: Sasha Levin

cpufreq: amd-pstate-ut: Remove module parameter access

2023-09-13T07:42:29Z

[ Upstream commit 8d6e5e8268e89979d86501dbb8385ce2e6154de1 ] In amd-pstate-ut, shared memory-based systems call get_shared_mem() as part of amd_pstate_ut_check_enabled() function. This function was written when CONFIG_X86_AMD_PSTATE was tristate config and amd_pstate can be built as a module. Currently CONFIG_X86_AMD_PSTATE is a boolean config and module parameter shared_mem is removed. But amd-pstate-ut code still accesses this module parameter. Remove those accesses. Fixes: 456ca88d8a52 ("cpufreq: amd-pstate: change amd-pstate driver to be built-in type") Reviewed-by: Mario Limonciello Reviewed-by: Meng Li Reviewed-by: Wyes Karny Suggested-by: Wyes Karny Signed-off-by: Swapnil Sapkal [ rjw: Subject edits ] Signed-off-by: Rafael J. Wysocki Signed-off-by: Sasha Levin

cpufreq: intel_pstate: set stale CPU frequency to minimum

2023-09-13T07:42:26Z

commit d51847acb018d83186e4af67bc93f9a00a8644f7 upstream. The intel_pstate CPU frequency scaling driver does not use policy->cur and it is 0. When the CPU frequency is outdated arch_freq_get_on_cpu() will default to the nominal clock frequency when its call to cpufreq_quick_getpolicy_cur returns the never updated 0. Thus, the listed frequency might be outside of currently set limits. Some users are complaining about the high reported frequency, albeit stale, when their system is idle and/or it is above the reduced maximum they have set. This patch will maintain policy_cur for the intel_pstate driver at the current minimum CPU frequency. Reported-by: Yang Jie Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217597 Signed-off-by: Doug Smythies [ rjw: White space damage fixes and comment adjustment ] Signed-off-by: Rafael J. Wysocki Signed-off-by: Keyon Jie Signed-off-by: Greg Kroah-Hartman

cpufreq: intel_pstate: Drop ACPI _PSS states table patching

2023-08-03T08:24:18Z

commit e8a0e30b742f76ebd0f3b196973df4bf65d8fbbb upstream. After making acpi_processor_get_platform_limit() use the "no limit" value for its frequency QoS request when _PPC returns 0, it is not necessary to replace the frequency corresponding to the first _PSS return package entry with the maximum turbo frequency of the given CPU in intel_pstate_init_acpi_perf_limits() any more, so drop the code doing that along with the comment explaining it. Signed-off-by: Rafael J. Wysocki Tested-by: Hagar Hemdan Signed-off-by: Greg Kroah-Hartman

cpufreq: mediatek: correct voltages for MT7622 and MT7623

2023-07-19T14:21:58Z

[ Upstream commit f85534113f5ae90a52521cdb9e9977a43ee42626 ] The MT6380 regulator typically used together with MT7622 does not support the current maximum processor and SRAM voltage in the cpufreq driver (1360000uV). For MT7622 limit processor and SRAM supply voltages to 1350000uV to avoid having the tracking algorithm request unsupported voltages from the regulator. On MT7623 there is no separate SRAM supply and the maximum voltage used is 1300000uV. Create dedicated platform data for MT7623 to cover that case as well. Fixes: 0883426fd07e3 ("cpufreq: mediatek: Raise proc and sram max voltage for MT7622/7623") Suggested-by: Jia-wei Chang Signed-off-by: Daniel Golle Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Viresh Kumar Signed-off-by: Sasha Levin

cpufreq: intel_pstate: Fix energy_performance_preference for passive

2023-07-19T14:21:01Z

[ Upstream commit 03f44ffb3d5be2fceda375d92c70ab6de4df7081 ] If the intel_pstate driver is set to passive mode, then writing the same value to the energy_performance_preference sysfs twice will fail. This is caused by the wrong return value used (index of the matched energy_perf_string), instead of the length of the passed in parameter. Fix by forcing the internal return value to zero when the same preference is passed in by user. This same issue is not present when active mode is used for the driver. Fixes: f6ebbcf08f37 ("cpufreq: intel_pstate: Implement passive mode with HWP enabled") Reported-by: Niklas Neronin Signed-off-by: Tero Kristo Signed-off-by: Rafael J. Wysocki Signed-off-by: Sasha Levin

cpufreq: amd-pstate: Add ->fast_switch() callback

2023-06-05T07:26:21Z

commit 4badf2eb1e986bdbf34dd2f5d4c979553a86fe54 upstream. Schedutil normally calls the adjust_perf callback for drivers with adjust_perf callback available and fast_switch_possible flag set. However, when frequency invariance is disabled and schedutil tries to invoke fast_switch. So, there is a chance of kernel crash if this function pointer is not set. To protect against this scenario add fast_switch callback to amd_pstate driver. Fixes: 1d215f0319c2 ("cpufreq: amd-pstate: Add fast switch function for AMD P-State") Signed-off-by: Gautham R. Shenoy Signed-off-by: Wyes Karny Signed-off-by: Rafael J. Wysocki Signed-off-by: Greg Kroah-Hartman