Re: [PATCH v2 1/6] clk: qcom: ipq5018: keep XO clock always on

[Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>]

On 5/14/25 10:04 AM, George Moussalem wrote:
> 
> 
> On 5/10/25 01:47, Konrad Dybcio wrote:
>> On 5/6/25 7:43 AM, George Moussalem via B4 Relay wrote:
>>> From: George Moussalem <george.moussalem@outlook.com>
>>>
>>> The XO clock must not be disabled to avoid the kernel trying to disable
>>> the it (when parenting it under the CMN PLL reference clock), else the
>>> kernel will panic and the below message will appear in the kernel logs.
>>> So let's enable the XO and its source CLK and keep them always on.
>>>
>>> [    0.916515] ------------[ cut here ]------------
>>> [    0.918890] gcc_xo_clk_src status stuck at 'on'
>>> [    0.918944] WARNING: CPU: 0 PID: 8 at drivers/clk/qcom/clk-branch.c:86 clk_branch_wait+0x114/0x124
>>> [    0.927926] Modules linked in:
>>> [    0.936945] CPU: 0 PID: 8 Comm: kworker/0:0 Not tainted 6.6.74 #0
>>> [    0.939982] Hardware name: Linksys MX2000 (DT)
>>> [    0.946151] Workqueue: pm pm_runtime_work
>>> [    0.950489] pstate: 604000c5 (nZCv daIF +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
>>> [    0.954566] pc : clk_branch_wait+0x114/0x124
>>> [    0.961335] lr : clk_branch_wait+0x114/0x124
>>> [    0.965849] sp : ffffffc08181bb50
>>> [    0.970101] x29: ffffffc08181bb50 x28: 0000000000000000 x27: 61c8864680b583eb
>>> [    0.973317] x26: ffffff801fec2168 x25: ffffff800000abc0 x24: 0000000000000002
>>> [    0.980437] x23: ffffffc0809f6fd8 x22: 0000000000000000 x21: ffffffc08044193c
>>> [    0.985276] loop: module loaded
>>> [    0.987554] x20: 0000000000000000 x19: ffffffc081749278 x18: 000000000000007c
>>> [    0.987573] x17: 0000000091706274 x16: 000000001985c4f7 x15: ffffffc0816bbdf0
>>> [    0.987587] x14: 0000000000000174 x13: 000000000000007c x12: 00000000ffffffea
>>> [    0.987601] x11: 00000000ffffefff x10: ffffffc081713df0 x9 : ffffffc0816bbd98
>>> [    0.987615] x8 : 0000000000017fe8 x7 : c0000000ffffefff x6 : 0000000000057fa8
>>> [    1.026268] x5 : 0000000000000fff x4 : 0000000000000000 x3 : ffffffc08181b950
>>> [    1.033385] x2 : ffffffc0816bbd30 x1 : ffffffc0816bbd30 x0 : 0000000000000023
>>> [    1.040507] Call trace:
>>> [    1.047618]  clk_branch_wait+0x114/0x124
>>> [    1.049875]  clk_branch2_disable+0x2c/0x3c
>>> [    1.054043]  clk_core_disable+0x60/0xac
>>> [    1.057948]  clk_core_disable+0x68/0xac
>>> [    1.061681]  clk_disable+0x30/0x4c
>>> [    1.065499]  pm_clk_suspend+0xd4/0xfc
>>> [    1.068971]  pm_generic_runtime_suspend+0x2c/0x44
>>> [    1.072705]  __rpm_callback+0x40/0x1bc
>>> [    1.077392]  rpm_callback+0x6c/0x78
>>> [    1.081038]  rpm_suspend+0xf0/0x5c0
>>> [    1.084423]  pm_runtime_work+0xf0/0xfc
>>> [    1.087895]  process_one_work+0x17c/0x2f8
>>> [    1.091716]  worker_thread+0x2e8/0x4d4
>>> [    1.095795]  kthread+0xdc/0xe0
>>> [    1.099440]  ret_from_fork+0x10/0x20
>>> [    1.102480] ---[ end trace 0000000000000000 ]---
>>>
>>> Signed-off-by: George Moussalem <george.moussalem@outlook.com>
>>> ---
>>
>> [...]
>>
>>> +    /* Keep some clocks always-on */
>>> +    qcom_branch_set_clk_en(regmap, 0x30018); /* GCC_XO_CLK_SRC */
>>
>> this clock is not a clk_branch2 - its control register is different and
>> this call is incorrect - you can drop it altogether, as if the XO source
>> clock isn't running, the system is dead
>>
>>> +    qcom_branch_set_clk_en(regmap, 0x30030); /* GCC_XO_CLK */
>>
>> This one actually is likely supposed to be always-on too - does removing
>> these two lines do any harm?
> 
> removing these lines AND the clock structs works AND updating the parents for clocks that reference the xo_clk_src works. There kernel is not complaining about anything. The other option is setting the CLK_IS_CRITICAL flag. What would your preference be?

Let's do critical after all

Konrad