linux-kernel - Re: [PATCH 2/3] clk: qcom: ipq6018: add USB GDSCs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 13 Nov 2023 12:58:42 +0100
From:   Konrad Dybcio <konrad.dybcio@...aro.org>
To:     Robert Marko <robimarko@...il.com>
Cc:     agross@...nel.org, andersson@...nel.org, robh+dt@...nel.org,
        krzysztof.kozlowski+dt@...aro.org, conor+dt@...nel.org,
        mturquette@...libre.com, sboyd@...nel.org,
        linux-arm-msm@...r.kernel.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-clk@...r.kernel.org,
        Kathiravan Thirumoorthy <quic_kathirav@...cinc.com>
Subject: Re: [PATCH 2/3] clk: qcom: ipq6018: add USB GDSCs

On 11.11.2023 12:28, Robert Marko wrote:
> On Tue, 7 Nov 2023 at 22:51, Konrad Dybcio <konrad.dybcio@...aro.org> wrote:
>>
>>
>>
>> On 10/31/23 10:01, Robert Marko wrote:
>>> On Mon, 30 Oct 2023 at 22:12, Konrad Dybcio <konrad.dybcio@...aro.org> wrote:
>>>>
>>>> On 30.10.2023 21:37, Robert Marko wrote:
>>>>> On Mon, 30 Oct 2023 at 20:37, Konrad Dybcio <konrad.dybcio@...aro.org> wrote:
>>>>>>
>>>>>> On 29.10.2023 12:04, Robert Marko wrote:
>>>>>>> On Wed, 25 Oct 2023 at 12:45, Robert Marko <robimarko@...il.com> wrote:
>>>>>>>>
>>>>>>>> IPQ6018 has GDSC-s for each of the USB ports, so lets define them as such
>>>>>>>> and drop the curent code that is de-asserting the USB GDSC-s as part of
>>>>>>>> the GCC probe.
>>>>>>>>
>>>>>>>> Signed-off-by: Robert Marko <robimarko@...il.com>
>>>>>>>
>>>>>>> Unfortunately, after testing on multiple devices I hit the same GDSC
>>>>>>> issue I had a long time ago
>>>>>>> that was the reason I did not send this upstream.
>>>>>>> It seems that USB3 port GDSC (USB0 GDSC in code) works just fine,
>>>>>>> however the USB2 one
>>>>>>> (USB1 GDSC in code) it is stuck off and USB2 port will fail due to this:
>>>>>>>      1.607531] ------------[ cut here ]------------
>>>>>>> [    1.607559] usb1_gdsc status stuck at 'off'
>>>>>>> [    1.607592] WARNING: CPU: 0 PID: 35 at gdsc_toggle_logic+0x16c/0x174
>>>>>>> [    1.615120] Modules linked in:
>>>>>> Can you dump GDSCR (the entire 32-bit register) at boot and when toggling?
>>>>>
>>>>> Sure, here it is:
>>>>> [    0.023760] qcom,gcc-ipq6018 1800000.gcc: reg: 0x3e078 val: 0x8222004 init
>>>>> [    0.023782] qcom,gcc-ipq6018 1800000.gcc: reg: 0x3f078 val: 0x8222004 init
>>>>> [    0.988626] qcom,gcc-ipq6018 1800000.gcc: reg: 0x3f078 val:
>>>>> 0x8282000 before toggle
>>>>> [    1.202506] qcom,gcc-ipq6018 1800000.gcc: reg: 0x3f078 val:
>>>>> 0x8282000 after toggle
>>>>> [    1.207208] qcom,gcc-ipq6018 1800000.gcc: reg: 0x3e078 val:
>>>>> 0xa0282000 before toggle
>>>> Any chance
>>>>
>>>> .en_few_wait_val = 0x2
>>>>
>>>> (turning BIT(19) into BIT(17))
>>>>
>>>> will make a difference?
>>>
>>> Sadly, it makes no difference and GDSC status bit newer comes up which is
>>> rather weird as USB0 one seems to work just fine.
>> What if you add clk_ignore_unused?
> 
> To the USB1 master clock or?
That's a command line parameter, effectively setting it on all clks.

> 
> There is definitively something broken regarding the GDSC as
> GDSC_STATE bits (30-27)
> change from 0 to something on the USB0 GDSC but on GDSC1 they are 0 even after
> SW_OVERRIDE BIT(2) is set to 1, and the POWER BIT(31) newer changes to 1.
> 
> However, if you manually set BIT(2) to 1 then the USB1 master clock
> can come up so
> GDSC seems to work.
> USB1 (The USB2.0 HS) port is still broken after this if USB mass storage is used
> but that was present before the GDSC changes as well and I still need
> to figure out
> which quirk is missing for this.
Please try clk_ignore_unused and see if toggling the GDSC is still broken.

Konrad