Re: [PATCH] mfd: rk808: Always use poweroff when requested

[Soeren Moch]

On 10.12.19 13:55, Markus Reichl wrote:
> Hi Anand,
>
> Am 10.12.19 um 13:42 schrieb Anand Moon:
>> Hi Soeren,
>>
>> On Mon, 9 Dec 2019 at 17:28, Soeren Moch <smoch@xxxxxx> wrote:
>>>
>>> With the device tree property "rockchip,system-power-controller" we
>>> explicitly request to use this PMIC to power off the system. So always
>>> register our poweroff function, even if some other handler (probably
>>> PSCI poweroff) was registered before.
>>>
>>> Signed-off-by: Soeren Moch <smoch@xxxxxx>
>>> ---
>>> Cc: Lee Jones <lee.jones@xxxxxxxxxx>
>>> Cc: Heiko Stuebner <heiko@xxxxxxxxx>
>>> Cc: linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
>>> Cc: linux-rockchip@xxxxxxxxxxxxxxxxxxx
>>> Cc: linux-kernel@xxxxxxxxxxxxxxx
>>> ---
>>> Âdrivers/mfd/rk808.c | 11 ++---------
>>> Â1 file changed, 2 insertions(+), 9 deletions(-)
>>>
>>> diff --git a/drivers/mfd/rk808.c b/drivers/mfd/rk808.c
>>> index a69a6742ecdc..616e44e7ef98 100644
>>> --- a/drivers/mfd/rk808.c
>>> +++ b/drivers/mfd/rk808.c
>>> @@ -550,7 +550,7 @@ static int rk808_probe(struct i2c_client *client,
>>> ÂÂÂÂÂÂÂ const struct mfd_cell *cells;
>>> ÂÂÂÂÂÂÂ int nr_pre_init_regs;
>>> ÂÂÂÂÂÂÂ int nr_cells;
>>> -ÂÂÂÂÂÂ int pm_off = 0, msb, lsb;
>>> +ÂÂÂÂÂÂ int msb, lsb;
>>> ÂÂÂÂÂÂÂ unsigned char pmic_id_msb, pmic_id_lsb;
>>> ÂÂÂÂÂÂÂ int ret;
>>> ÂÂÂÂÂÂÂ int i;
>>> @@ -674,16 +674,9 @@ static int rk808_probe(struct i2c_client *client,
>>> ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ goto err_irq;
>>> ÂÂÂÂÂÂÂ }
>>>
>>> -ÂÂÂÂÂÂ pm_off = of_property_read_bool(np,
>>> -ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ "rockchip,system-power-controller");
>>> -ÂÂÂÂÂÂ if (pm_off && !pm_power_off) {
>>> +ÂÂÂÂÂÂ if (of_property_read_bool(np,
>>> "rockchip,system-power-controller")) {
>>> ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ rk808_i2c_client = client;
>>> ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ pm_power_off = rk808->pm_pwroff_fn;
>>> -ÂÂÂÂÂÂ }
>>> -
>>> -ÂÂÂÂÂÂ if (pm_off && !pm_power_off_prepare) {
>>> -ÂÂÂÂÂÂÂÂÂÂÂÂÂÂ if (!rk808_i2c_client)
>>> -ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ rk808_i2c_client = client;
>>> ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ pm_power_off_prepare = rk808->pm_pwroff_prep_fn;
>>> ÂÂÂÂÂÂÂ }
>>>
>>
>> I gave this a try on my Rock960 and Odroid N1
>> both got kernel panic below.
>
> I see the same on rk3399-roc-pc.
This is no panic, it's a harmless warning.
The i2c core nowadays expects a specially marked i2c transfer function
late in the powerdown cycle:

diff --git a/drivers/i2c/busses/i2c-rk3x.c b/drivers/i2c/busses/i2c-rk3x.c
index 1a33007b03e9..cec115e0afa4 100644
--- a/drivers/i2c/busses/i2c-rk3x.c
+++ b/drivers/i2c/busses/i2c-rk3x.c
@@ -1126,6 +1126,7 @@ static u32 rk3x_i2c_func(struct i2c_adapter *adap)
Â
Âstatic const struct i2c_algorithm rk3x_i2c_algorithm = {
ÂÂÂÂ .master_xferÂÂÂ ÂÂÂ = rk3x_i2c_xfer,
+ÂÂÂ .master_xfer_atomicÂÂÂ = rk3x_i2c_xfer, /* usable for PMIC poweroff */
ÂÂÂÂ .functionalityÂÂÂ ÂÂÂ = rk3x_i2c_func,
Â};
Â
---
It is only used for powerdown. The regular i2c xfer function works.

Heiko, should I send a formal patch for that?

Soeren

>
>>
>> [ÂÂ 58.305868] xhci-hcd xhci-hcd.0.auto: USB bus 5 deregistered
>> [ÂÂ 58.306747] reboot: Power down
>> [ÂÂ 58.307106] ------------[ cut here ]------------
>> [ÂÂ 58.307510] No atomic I2C transfer handler for 'i2c-0'
>> [ÂÂ 58.308007] WARNING: CPU: 0 PID: 1 at drivers/i2c/i2c-core.h:41
>> i2c_transfer+0xe4/0xf8
>> [ÂÂ 58.308696] Modules linked in: snd_soc_hdmi_codec dw_hdmi_i2s_audio
>> rockchipdrm analogix_dp brcmfmac nvme dw_mipi_dsi nvme_core dw_hdmi
>> panfrost cec brcmutil drm_kms_helper gpu_sched cfg80211 hci_uart drm
>> btbcm crct10dif_ce snd_soc_simple_card bluetooth snd_soc_rockchip_i2s
>> snd_soc_simple_card_utils snd_soc_rockchip_pcm phy_rockchip_pcie
>> ecdh_generic rtc_rk808 ecc pcie_rockchip_host rfkill rockchip_thermal
>> ip_tables x_tables ipv6 nf_defrag_ipv6
>> [ÂÂ 58.312150] CPU: 0 PID: 1 Comm: shutdown Not tainted
>> 5.5.0-rc1-dirty #1
>> [ÂÂ 58.312725] Hardware name: 96boards Rock960 (DT)
>> [ÂÂ 58.313131] pstate: 60000085 (nZCv daIf -PAN -UAO)
>> [ÂÂ 58.313551] pc : i2c_transfer+0xe4/0xf8
>> [ÂÂ 58.313889] lr : i2c_transfer+0xe4/0xf8
>> [ÂÂ 58.314225] sp : ffff80001004bb00
>> [ÂÂ 58.314516] x29: ffff80001004bb00 x28: ffff00007d208000
>> [ÂÂ 58.314981] x27: 0000000000000000 x26: 0000000000000000
>> [ÂÂ 58.315446] x25: 0000000000000000 x24: 0000000000000008
>> [ÂÂ 58.315910] x23: 0000000000000000 x22: ffff80001004bc74
>> [ÂÂ 58.316375] x21: 0000000000000002 x20: ffff80001004bb58
>> [ÂÂ 58.316841] x19: ffff0000784f0880 x18: 0000000000000010
>> [ÂÂ 58.317305] x17: 0000000000000001 x16: 0000000000000019
>> [ÂÂ 58.317770] x15: ffffffffffffffff x14: ffff8000118398c8
>> [ÂÂ 58.318236] x13: ffff80009004b867 x12: ffff80001004b86f
>> [ÂÂ 58.318701] x11: ffff800011851000 x10: ffff80001004b7f0
>> [ÂÂ 58.319166] x9 : 00000000ffffffd0 x8 : ffff800010699ad8
>> [ÂÂ 58.319631] x7 : 0000000000000265 x6 : ffff800011a20be9
>> [ÂÂ 58.320096] x5 : 0000000000000000 x4 : 0000000000000000
>> [ÂÂ 58.320561] x3 : 00000000ffffffff x2 : ffff800011851ab8
>> [ÂÂ 58.321026] x1 : d375c0d4f4751f00 x0 : 0000000000000000
>> [ÂÂ 58.321491] Call trace:
>> [ÂÂ 58.321710]Â i2c_transfer+0xe4/0xf8
>> [ÂÂ 58.322020]Â regmap_i2c_read+0x5c/0x98
>> [ÂÂ 58.322350]Â _regmap_raw_read+0xcc/0x138
>> [ÂÂ 58.322694]Â _regmap_bus_read+0x3c/0x70
>> [ÂÂ 58.323034]Â _regmap_read+0x60/0xe0
>> [ÂÂ 58.323341]Â _regmap_update_bits+0xc8/0x108
>> [ÂÂ 58.323707]Â regmap_update_bits_base+0x60/0x90
>> [ÂÂ 58.324099]Â rk808_device_shutdown+0x38/0x50
>> [ÂÂ 58.324476]Â machine_power_off+0x24/0x30
>> [ÂÂ 58.324823]Â kernel_power_off+0x64/0x70
>> [ÂÂ 58.325159]Â __do_sys_reboot+0x15c/0x240
>> [ÂÂ 58.325504]Â __arm64_sys_reboot+0x20/0x28
>> [ÂÂ 58.325858]Â el0_svc_common.constprop.2+0x88/0x150
>> [ÂÂ 58.326279]Â el0_svc_handler+0x20/0x80
>> [ÂÂ 58.326607]Â el0_sync_handler+0x118/0x188
>> [ÂÂ 58.326960]Â el0_sync+0x140/0x180
>> [ÂÂ 58.327251] ---[ end trace b1de39d03d724d01 ]---
>>
>> -Anand
>>
>> _______________________________________________
>> Linux-rockchip mailing list
>> Linux-rockchip@xxxxxxxxxxxxxxxxxxx
>> http://lists.infradead.org/mailman/listinfo/linux-rockchip
>>
>
> GruÃ,