linux-stable.git/drivers/cpufreq, branch v5.17.7

cpufreq: qcom-cpufreq-hw: Clear dcvs interrupts

2022-05-09T07:16:24+00:00

[ Upstream commit e4e6448638a01905faeda9bf96aa9df7c8ef463c ]

It's noted that dcvs interrupts are not self-clearing, thus an interrupt
handler runs constantly, which leads to a severe regression in runtime.
To fix the problem an explicit write to clear interrupt register is
required, note that on OSM platforms the register may not be present.

Fixes: 275157b367f4 ("cpufreq: qcom-cpufreq-hw: Add dcvs interrupt support")
Signed-off-by: Vladimir Zapolskiy 
Signed-off-by: Viresh Kumar 
Signed-off-by: Sasha Levin

cpufreq: fix memory leak in sun50i_cpufreq_nvmem_probe

2022-05-09T07:16:22+00:00

[ Upstream commit 1aa24a8f3b5133dae4bc1e57427e345445f3e902 ]

--------------------------------------------
unreferenced object 0xffff000010742a00 (size 128):
  comm "swapper/0", pid 1, jiffies 4294902015 (age 1187.652s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<00000000b4dfebaa>] __kmalloc+0x338/0x474
    [<00000000d6e716db>] sun50i_cpufreq_nvmem_probe+0xc4/0x36c
    [<000000007d6082a0>] platform_probe+0x98/0x11c
    [<00000000c990f549>] really_probe+0x234/0x5a0
    [<000000002d9fecc6>] __driver_probe_device+0x194/0x224
    [<00000000cf0b94fa>] driver_probe_device+0x64/0x13c
    [<00000000f238e4cf>] __device_attach_driver+0xf8/0x180
    [<000000006720e418>] bus_for_each_drv+0xf8/0x160
    [<00000000df4f14f6>] __device_attach+0x174/0x29c
    [<00000000782002fb>] device_initial_probe+0x20/0x30
    [<00000000c2681b06>] bus_probe_device+0xfc/0x110
    [<00000000964cf3bd>] device_add+0x5f0/0xcd0
    [<000000004b9264e3>] platform_device_add+0x198/0x390
    [<00000000fa82a9d0>] platform_device_register_full+0x178/0x210
    [<000000009a5daf13>] sun50i_cpufreq_init+0xf8/0x168
    [<000000000377cc7c>] do_one_initcall+0xe4/0x570
--------------------------------------------

if sun50i_cpufreq_get_efuse failed, then opp_tables leak.

Fixes: f328584f7bff ("cpufreq: Add sun50i nvmem based CPU scaling driver")
Signed-off-by: Xiaobing Luo 
Reviewed-by: Samuel Holland 
Signed-off-by: Viresh Kumar 
Signed-off-by: Sasha Levin

cpufreq: qcom-cpufreq-hw: Fix throttle frequency value on EPSS platforms

2022-05-09T07:16:16+00:00

[ Upstream commit f84ccad5f5660f86a642a3d7e2bfdc4e7a8a2d49 ]

On QCOM platforms with EPSS flavour of cpufreq IP a throttled frequency is
obtained from another register REG_DOMAIN_STATE, thus the helper function
qcom_lmh_get_throttle_freq() should be modified accordingly, as for now
it returns gibberish since .reg_current_vote is unset for EPSS hardware.

To exclude a hardcoded magic number 19200 it is replaced by "xo" clock rate
in KHz.

Fixes: 275157b367f4 ("cpufreq: qcom-cpufreq-hw: Add dcvs interrupt support")
Reviewed-by: Bjorn Andersson 
Signed-off-by: Vladimir Zapolskiy 
Signed-off-by: Viresh Kumar 
Signed-off-by: Sasha Levin

cpufreq: qcom-hw: fix the opp entries refcounting

2022-05-09T07:16:16+00:00

[ Upstream commit 6240aaad75e1a623872a830d13393d7aabf1052c ]

The qcom_lmh_dcvs_notify() will get the dev_pm_opp instance for
throttling, but will not put it, ending up with leaking a reference
count and the following backtrace when putting the CPU offline.

Correctly put the reference count of the returned opp instance.

[   84.418025] ------------[ cut here ]------------
[   84.422770] WARNING: CPU: 7 PID: 43 at drivers/opp/core.c:1396 _opp_table_kref_release+0x188/0x190
[   84.431966] Modules linked in:
[   84.435106] CPU: 7 PID: 43 Comm: cpuhp/7 Tainted: G S                5.17.0-rc6-00388-g7cf3c0d89c44-dirty #721
[   84.451631] pstate: 82400005 (Nzcv daif +PAN -UAO +TCO -DIT -SSBS BTYPE=--)
[   84.458781] pc : _opp_table_kref_release+0x188/0x190
[   84.463878] lr : _opp_table_kref_release+0x78/0x190
[   84.468885] sp : ffff80000841bc70
[   84.472294] x29: ffff80000841bc70 x28: ffff6664afe3d000 x27: ffff1db6729e5908
[   84.479621] x26: 0000000000000000 x25: 0000000000000000 x24: ffff1db6729e58e0
[   84.486946] x23: ffff8000080a5000 x22: ffff1db40aad80e0 x21: ffff1db4002fec80
[   84.494277] x20: ffff1db40aad8000 x19: ffffb751c3186300 x18: ffffffffffffffff
[   84.501603] x17: 5300326563697665 x16: 645f676e696c6f6f x15: 00001186c1df5448
[   84.508928] x14: 00000000000002e9 x13: 0000000000000000 x12: 0000000000000000
[   84.516256] x11: ffffb751c3186368 x10: ffffb751c39a2a70 x9 : 0000000000000000
[   84.523585] x8 : ffff1db4008edf00 x7 : ffffb751c328c000 x6 : 0000000000000001
[   84.530916] x5 : 0000000000040000 x4 : 0000000000000001 x3 : ffff1db4008edf00
[   84.538247] x2 : 0000000000000000 x1 : ffff1db400aa6100 x0 : ffff1db40aad80d0
[   84.545579] Call trace:
[   84.548101]  _opp_table_kref_release+0x188/0x190
[   84.552842]  dev_pm_opp_remove_all_dynamic+0x8c/0xc0
[   84.557949]  qcom_cpufreq_hw_cpu_exit+0x30/0xdc
[   84.562608]  cpufreq_offline.isra.0+0x1b4/0x1d8
[   84.567270]  cpuhp_cpufreq_offline+0x10/0x6c
[   84.571663]  cpuhp_invoke_callback+0x16c/0x2b0
[   84.576231]  cpuhp_thread_fun+0x190/0x250
[   84.580353]  smpboot_thread_fn+0x12c/0x230
[   84.584568]  kthread+0xfc/0x100
[   84.587810]  ret_from_fork+0x10/0x20
[   84.591490] irq event stamp: 3482
[   84.594901] hardirqs last  enabled at (3481): [] call_rcu+0x39c/0x50c
[   84.603119] hardirqs last disabled at (3482): [] el1_dbg+0x24/0x8c
[   84.611074] softirqs last  enabled at (310): [] _stext+0x410/0x588
[   84.619028] softirqs last disabled at (305): [] __irq_exit_rcu+0x158/0x174
[   84.627691] ---[ end trace 0000000000000000 ]---

Fixes: 275157b367f4 ("cpufreq: qcom-cpufreq-hw: Add dcvs interrupt support")
Reported-by: kernel test robot 
Tested-by: Vladimir Zapolskiy 
Reviewed-by: Vladimir Zapolskiy 
Reviewed-by: Bjorn Andersson 
Signed-off-by: Dmitry Baryshkov 
Signed-off-by: Viresh Kumar 
Signed-off-by: Sasha Levin

cpufreq: qcom-hw: fix the race between LMH worker and cpuhp

2022-05-09T07:16:16+00:00

[ Upstream commit 5e4f009da6be563984ba4db4ef4f32529e9aeb90 ]

The driver would disable the worker when cpu is being put offline, but
it happens closer to the end of cpufreq_offline(). The function
qcom_lmh_dcvs_poll() can be running in parallel with this, when
policy->cpus already has been updated. Read policy->related_cpus
instead.

[   37.122433] ------------[ cut here ]------------
[   37.127225] WARNING: CPU: 0 PID: 187 at drivers/base/arch_topology.c:180 topology_update_thermal_pressure+0xec/0x100
[   37.138098] Modules linked in:
[   37.141279] CPU: 0 PID: 187 Comm: kworker/0:3 Tainted: G S                5.17.0-rc6-00389-g37c83d0b8710-dirty #713
[   37.158306] Workqueue: events qcom_lmh_dcvs_poll
[   37.163095] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[   37.170278] pc : topology_update_thermal_pressure+0xec/0x100
[   37.176131] lr : topology_update_thermal_pressure+0x20/0x100
[   37.181977] sp : ffff800009b6bce0
[   37.185402] x29: ffff800009b6bce0 x28: ffffd87abe92b000 x27: ffff04bd7292e205
[   37.192792] x26: ffffd87abe930af8 x25: ffffd87abe94e4c8 x24: 0000000000000000
[   37.200180] x23: ffff04bb01177018 x22: ffff04bb011770c0 x21: ffff04bb01177000
[   37.207567] x20: ffff04bb0a419000 x19: 00000000000c4e00 x18: 0000000000000000
[   37.214954] x17: 000000040044ffff x16: 004000b2b5503510 x15: 0000006aaa1326d2
[   37.222333] x14: 0000000000000232 x13: 0000000000000001 x12: 0000000000000040
[   37.229718] x11: ffff04bb00400000 x10: 968f57bd39f701c8 x9 : ffff04bb0acc8674
[   37.237095] x8 : fefefefefefefeff x7 : 0000000000000018 x6 : ffffd87abd90092c
[   37.244478] x5 : 0000000000000016 x4 : 0000000000000000 x3 : 0000000000000100
[   37.251852] x2 : ffff04bb0a419020 x1 : 0000000000000100 x0 : 0000000000000100
[   37.259235] Call trace:
[   37.261771]  topology_update_thermal_pressure+0xec/0x100
[   37.267266]  qcom_lmh_dcvs_poll+0xbc/0x154
[   37.271505]  process_one_work+0x288/0x69c
[   37.275654]  worker_thread+0x74/0x470
[   37.279450]  kthread+0xfc/0x100
[   37.282712]  ret_from_fork+0x10/0x20
[   37.286417] irq event stamp: 74
[   37.289664] hardirqs last  enabled at (73): [] _raw_spin_unlock_irq+0x44/0x80
[   37.298632] hardirqs last disabled at (74): [] __schedule+0x710/0xa10
[   37.306885] softirqs last  enabled at (58): [] _stext+0x410/0x588
[   37.314778] softirqs last disabled at (51): [] __irq_exit_rcu+0x158/0x174
[   37.323386] ---[ end trace 0000000000000000 ]---

Fixes: 275157b367f4 ("cpufreq: qcom-cpufreq-hw: Add dcvs interrupt support")
Signed-off-by: Dmitry Baryshkov 
Reviewed-by: Bjorn Andersson 
Signed-off-by: Viresh Kumar 
Signed-off-by: Sasha Levin

cpufreq: qcom-hw: drop affinity hint before freeing the IRQ

2022-05-09T07:16:16+00:00

[ Upstream commit be5985b3dbce5ba2af3c8b0f2b7df235c93907e6 ]

Drop affinity hint before freeing the throttling IRQ to fix the
following trace:

[  185.114773] ------------[ cut here ]------------
[  185.119517] WARNING: CPU: 7 PID: 43 at kernel/irq/manage.c:1887 free_irq+0x3a4/0x3dc
[  185.127474] Modules linked in:
[  185.130618] CPU: 7 PID: 43 Comm: cpuhp/7 Tainted: G S      W         5.17.0-rc6-00386-g67382a5b705d-dirty #690
[  185.147125] pstate: 604000c5 (nZCv daIF +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[  185.154269] pc : free_irq+0x3a4/0x3dc
[  185.158031] lr : free_irq+0x33c/0x3dc
[  185.161792] sp : ffff80000841bc90
[  185.165195] x29: ffff80000841bc90 x28: ffffa6edc5c3d000 x27: ffff6d93729e5908
[  185.172515] x26: 0000000000000000 x25: ffff6d910109fc00 x24: ffff6d91011490e0
[  185.179838] x23: ffff6d9101149218 x22: 0000000000000080 x21: 0000000000000000
[  185.187163] x20: ffff6d9101149000 x19: ffff6d910ab61500 x18: ffffffffffffffff
[  185.194487] x17: 2e35202020202020 x16: 2020202020202020 x15: ffff80008841b9a7
[  185.201805] x14: 00000000000003c9 x13: 0000000000000001 x12: 0000000000000040
[  185.209135] x11: ffff6d91005aab58 x10: ffff6d91005aab5a x9 : ffffc6a5ad1c5408
[  185.216455] x8 : ffff6d91005adb88 x7 : 0000000000000000 x6 : ffffc6a5ab5a91f4
[  185.223776] x5 : 0000000000000000 x4 : ffff6d91011490a8 x3 : ffffc6a5ad266108
[  185.231098] x2 : 0000000013033204 x1 : ffff6d9101149000 x0 : ffff6d910a9cc000
[  185.238421] Call trace:
[  185.240932]  free_irq+0x3a4/0x3dc
[  185.244334]  qcom_cpufreq_hw_cpu_exit+0x78/0xcc
[  185.248985]  cpufreq_offline.isra.0+0x228/0x270
[  185.253639]  cpuhp_cpufreq_offline+0x10/0x20
[  185.258027]  cpuhp_invoke_callback+0x16c/0x2b0
[  185.262592]  cpuhp_thread_fun+0x190/0x250
[  185.266710]  smpboot_thread_fn+0x12c/0x230
[  185.270914]  kthread+0xfc/0x100
[  185.274145]  ret_from_fork+0x10/0x20
[  185.277820] irq event stamp: 212
[  185.281136] hardirqs last  enabled at (211): [] _raw_spin_unlock_irqrestore+0x8c/0xa0
[  185.290775] hardirqs last disabled at (212): [] __schedule+0x710/0xa10
[  185.299081] softirqs last  enabled at (0): [] copy_process+0x7d0/0x1a14
[  185.307475] softirqs last disabled at (0): [<0000000000000000>] 0x0

Fixes: 3ed6dfbd3bb98 ("cpufreq: qcom-hw: Set CPU affinity of dcvsh interrupts")
Tested-by: Vladimir Zapolskiy 
Reviewed-by: Vladimir Zapolskiy 
Reviewed-by: Bjorn Andersson 
Signed-off-by: Dmitry Baryshkov 
Signed-off-by: Viresh Kumar 
Signed-off-by: Sasha Levin

cpufreq: CPPC: Fix performance/frequency conversion

2022-04-13T17:27:24+00:00

[ Upstream commit ec1c7ad47664f964c1101fe555b6fde0cb124b38 ]

CPUfreq governors request CPU frequencies using information
on current CPU usage. The CPPC driver converts them to
performance requests. Frequency targets are computed as:
	target_freq = (util / cpu_capacity) * max_freq
target_freq is then clamped between [policy->min, policy->max].

The CPPC driver converts performance values to frequencies
(and vice-versa) using cppc_cpufreq_perf_to_khz() and
cppc_cpufreq_khz_to_perf(). These functions both use two different
factors depending on the range of the input value. For
cppc_cpufreq_khz_to_perf():
- (NOMINAL_PERF / NOMINAL_FREQ) or
- (LOWEST_PERF / LOWEST_FREQ)
and for cppc_cpufreq_perf_to_khz():
- (NOMINAL_FREQ / NOMINAL_PERF) or
- ((NOMINAL_PERF - LOWEST_FREQ) / (NOMINAL_PERF - LOWEST_PERF))

This means:
1- the functions are not inverse for some values:
   (perf_to_khz(khz_to_perf(x)) != x)
2- cppc_cpufreq_perf_to_khz(LOWEST_PERF) can sometimes give
   a different value from LOWEST_FREQ due to integer approximation
3- it is implied that performance and frequency are proportional
   (NOMINAL_FREQ / NOMINAL_PERF) == (LOWEST_PERF / LOWEST_FREQ)

This patch changes the conversion functions to an affine function.
This fixes the 3 points above.

Suggested-by: Lukasz Luba 
Suggested-by: Morten Rasmussen 
Signed-off-by: Pierre Gondois 
Signed-off-by: Viresh Kumar 
Signed-off-by: Sasha Levin

cpufreq: qcom-cpufreq-nvmem: fix reading of PVS Valid fuse

2022-04-08T11:58:24+00:00

[ Upstream commit 4a8a77abf0e2b6468ba0281e33384cbec5fb476a ]

The fuse consists of 64 bits, with this statement we're supposed to get
the upper 32 bits but it actually read out of bounds and got 0 instead
of the desired value which lead to the "PVS bin not set." codepath being
run resetting our pvs value.

Fixes: a8811ec764f9 ("cpufreq: qcom: Add support for krait based socs")
Signed-off-by: Luca Weiss 
Signed-off-by: Viresh Kumar 
Signed-off-by: Sasha Levin

cpufreq: qcom-hw: Delay enabling throttle_irq

2022-02-09T07:48:49+00:00

In the event that the SoC is under thermal pressure while booting it's
possible for the dcvs notification to happen inbetween the cpufreq
framework calling init and it actually updating the policy's
related_cpus cpumask.

Prior to the introduction of the thermal pressure update helper an empty
cpumask would simply result in the thermal pressure of no cpus being
updated, but the new code will attempt to dereference an invalid per_cpu
variable.

Avoid this problem by using the newly reintroduced "ready" callback, to
postpone enabling the IRQ until the related_cpus cpumask is filled in.

Fixes: 0258cb19c77d ("cpufreq: qcom-cpufreq-hw: Use new thermal pressure update function")
Signed-off-by: Bjorn Andersson 
Signed-off-by: Viresh Kumar

cpufreq: Reintroduce ready() callback

2022-02-09T07:48:49+00:00

This effectively revert '4bf8e582119e ("cpufreq: Remove ready()
callback")', in order to reintroduce the ready callback.

This is needed in order to be able to leave the thermal pressure
interrupts in the Qualcomm CPUfreq driver disabled during
initialization, so that it doesn't fire while related_cpus are still 0.

Signed-off-by: Bjorn Andersson 
[ Viresh: Added the Chinese translation as well and updated commit msg ]
Signed-off-by: Viresh Kumar