Message ID | 20240711115207.2843133-6-claudiu.beznea.uj@bp.renesas.com (mailing list archive) |
---|---|
State | Superseded |
Delegated to: | Geert Uytterhoeven |
Headers | show |
Series | i2c: riic: Add support for Renesas RZ/G3S | expand |
On Thu, Jul 11, 2024 at 02:52:01PM +0300, Claudiu wrote: > From: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com> > > Add suspend/resume support for the RIIC driver. This is necessary for the > Renesas RZ/G3S SoC which support suspend to deep sleep state where power > to most of the SoC components is turned off. As a result the I2C controller > needs to be reconfigured after suspend/resume. For this, the reset line > was stored in the driver private data structure as well as i2c timings. > The reset line and I2C timings are necessary to re-initialize the > controller after resume. > > Signed-off-by: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com> ? Doesn't apply on top of the previous patches for me? > +static int riic_i2c_resume(struct device *dev) > +{ > + struct riic_dev *riic = dev_get_drvdata(dev); > + int ret; > + > + ret = reset_control_deassert(riic->rstc); > + if (ret) > + return ret; > + > + ret = riic_init_hw(riic); > + if (ret) { > + reset_control_assert(riic->rstc); Is this assertion really needed? It is not done when init_hw fails in probe().
On 08.08.2024 18:15, Wolfram Sang wrote: > On Thu, Jul 11, 2024 at 02:52:01PM +0300, Claudiu wrote: >> From: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com> >> >> Add suspend/resume support for the RIIC driver. This is necessary for the >> Renesas RZ/G3S SoC which support suspend to deep sleep state where power >> to most of the SoC components is turned off. As a result the I2C controller >> needs to be reconfigured after suspend/resume. For this, the reset line >> was stored in the driver private data structure as well as i2c timings. >> The reset line and I2C timings are necessary to re-initialize the >> controller after resume. >> >> Signed-off-by: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com> > > ? Doesn't apply on top of the previous patches for me? I just checked it on next-20240809. It should be due to commit e1571b1fb4ff ("i2c: riic: reword according to newest specification") which introduced changes around riic_algo object, present also in the diff of this patch. > >> +static int riic_i2c_resume(struct device *dev) >> +{ >> + struct riic_dev *riic = dev_get_drvdata(dev); >> + int ret; >> + >> + ret = reset_control_deassert(riic->rstc); >> + if (ret) >> + return ret; >> + >> + ret = riic_init_hw(riic); >> + if (ret) { >> + reset_control_assert(riic->rstc); > > Is this assertion really needed? It is not done when init_hw fails in > probe(). In case riic_init_hw() fails there is no recovering way for this driver, AFAICT, and thus there is no point in keeping the reset signal de-asserted. In probe this is handled by the devres through action or reset function (riic_reset_control_assert) registered by: ret = devm_add_action_or_reset(dev, riic_reset_control_assert, rstc); if (ret) return ret; Thank you, Claudiu Beznea
> I just checked it on next-20240809. It should be due to commit > e1571b1fb4ff ("i2c: riic: reword according to newest specification") > which introduced changes around riic_algo object, present also in the diff > of this patch. Ah, okay, this patch is the culprit. I wonder, though, because it is already in 6.11-rc1 which was the base for my test. But you need to resend anyhow... > In case riic_init_hw() fails there is no recovering way for this driver, > AFAICT, and thus there is no point in keeping the reset signal de-asserted. Right. If it fails in resume(), then the driver will still not be removed.
diff --git a/drivers/i2c/busses/i2c-riic.c b/drivers/i2c/busses/i2c-riic.c index 46765715d39f..5e4f56ee97e3 100644 --- a/drivers/i2c/busses/i2c-riic.c +++ b/drivers/i2c/busses/i2c-riic.c @@ -105,6 +105,8 @@ struct riic_dev { struct completion msg_done; struct i2c_adapter adapter; struct clk *clk; + struct reset_control *rstc; + struct i2c_timings i2c_t; }; struct riic_irq_desc { @@ -302,11 +304,12 @@ static const struct i2c_algorithm riic_algo = { .functionality = riic_func, }; -static int riic_init_hw(struct riic_dev *riic, struct i2c_timings *t) +static int riic_init_hw(struct riic_dev *riic) { int ret; unsigned long rate; int total_ticks, cks, brl, brh; + struct i2c_timings *t = &riic->i2c_t; struct device *dev = riic->adapter.dev.parent; if (t->bus_freq_hz > I2C_MAX_FAST_MODE_FREQ) { @@ -423,8 +426,6 @@ static int riic_i2c_probe(struct platform_device *pdev) struct device *dev = &pdev->dev; struct riic_dev *riic; struct i2c_adapter *adap; - struct i2c_timings i2c_t; - struct reset_control *rstc; int i, ret; riic = devm_kzalloc(dev, sizeof(*riic), GFP_KERNEL); @@ -441,16 +442,16 @@ static int riic_i2c_probe(struct platform_device *pdev) return PTR_ERR(riic->clk); } - rstc = devm_reset_control_get_optional_exclusive(dev, NULL); - if (IS_ERR(rstc)) - return dev_err_probe(dev, PTR_ERR(rstc), + riic->rstc = devm_reset_control_get_optional_exclusive(dev, NULL); + if (IS_ERR(riic->rstc)) + return dev_err_probe(dev, PTR_ERR(riic->rstc), "Error: missing reset ctrl\n"); - ret = reset_control_deassert(rstc); + ret = reset_control_deassert(riic->rstc); if (ret) return ret; - ret = devm_add_action_or_reset(dev, riic_reset_control_assert, rstc); + ret = devm_add_action_or_reset(dev, riic_reset_control_assert, riic->rstc); if (ret) return ret; @@ -479,13 +480,13 @@ static int riic_i2c_probe(struct platform_device *pdev) init_completion(&riic->msg_done); - i2c_parse_fw_timings(dev, &i2c_t, true); + i2c_parse_fw_timings(dev, &riic->i2c_t, true); pm_runtime_set_autosuspend_delay(dev, 0); pm_runtime_use_autosuspend(dev); pm_runtime_enable(dev); - ret = riic_init_hw(riic, &i2c_t); + ret = riic_init_hw(riic); if (ret) goto out; @@ -495,7 +496,7 @@ static int riic_i2c_probe(struct platform_device *pdev) platform_set_drvdata(pdev, riic); - dev_info(dev, "registered with %dHz bus speed\n", i2c_t.bus_freq_hz); + dev_info(dev, "registered with %dHz bus speed\n", riic->i2c_t.bus_freq_hz); return 0; out: @@ -552,6 +553,50 @@ static const struct riic_of_data riic_rz_v2h_info = { }, }; +static int riic_i2c_suspend(struct device *dev) +{ + struct riic_dev *riic = dev_get_drvdata(dev); + int ret; + + ret = pm_runtime_resume_and_get(dev); + if (ret) + return ret; + + i2c_mark_adapter_suspended(&riic->adapter); + + /* Disable output on SDA, SCL pins. */ + riic_clear_set_bit(riic, ICCR1_ICE, 0, RIIC_ICCR1); + + pm_runtime_mark_last_busy(dev); + pm_runtime_put_sync(dev); + + return reset_control_assert(riic->rstc); +} + +static int riic_i2c_resume(struct device *dev) +{ + struct riic_dev *riic = dev_get_drvdata(dev); + int ret; + + ret = reset_control_deassert(riic->rstc); + if (ret) + return ret; + + ret = riic_init_hw(riic); + if (ret) { + reset_control_assert(riic->rstc); + return ret; + } + + i2c_mark_adapter_resumed(&riic->adapter); + + return 0; +} + +static const struct dev_pm_ops riic_i2c_pm_ops = { + SYSTEM_SLEEP_PM_OPS(riic_i2c_suspend, riic_i2c_resume) +}; + static const struct of_device_id riic_i2c_dt_ids[] = { { .compatible = "renesas,riic-rz", .data = &riic_rz_a_info }, { .compatible = "renesas,riic-r9a09g057", .data = &riic_rz_v2h_info }, @@ -564,6 +609,7 @@ static struct platform_driver riic_i2c_driver = { .driver = { .name = "i2c-riic", .of_match_table = riic_i2c_dt_ids, + .pm = pm_ptr(&riic_i2c_pm_ops), }, };