| Message ID | 20181210210310.12677-9-wsa+renesas@sang-engineering.com (mailing list archive) |
|---|---|
| State | RFC, archived |
| Headers | show |
| Series | i2c: move handling of suspended adapters to the core | expand |
Hi, On 2018-12-10 22:03, Wolfram Sang wrote: > Rejecting transfers should be handled by the core. Also, this will > ensure proper locking which was forgotten in this open coded version. > > Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Works fine on TM2e and OdroidXU3. The only minor issue is the following warning: WARNING: CPU: 4 PID: 28 at drivers/i2c/i2c-core-base.c:1869 __i2c_transfer+0x428/0x5e0 Modules linked in: CPU: 4 PID: 28 Comm: cpuhp/4 Tainted: G W 4.20.0-rc1-00031-ga1c77790d03f #5130 Hardware name: Samsung TM2E board (DT) pstate: 20000005 (nzCv daif -PAN -UAO) pc : __i2c_transfer+0x428/0x5e0 lr : i2c_transfer+0x64/0xb8 ... Call trace: __i2c_transfer+0x428/0x5e0 i2c_transfer+0x64/0xb8 regmap_i2c_read+0x5c/0xa0 _regmap_raw_read+0xac/0x2a8 _regmap_bus_read+0x3c/0x70 _regmap_read+0x64/0x1a0 regmap_read+0x48/0x70 regulator_is_enabled_regmap+0x38/0xb8 _regulator_is_enabled.part.1+0x1c/0x30 create_regulator+0x264/0x270 _regulator_get+0x90/0x2b0 regulator_get_optional+0x10/0x18 dev_pm_opp_set_regulators+0xb8/0x1e8 cpufreq_init+0xd0/0x2f8 cpufreq_online+0xc0/0x660 cpuhp_cpufreq_online+0xc/0x18 cpuhp_invoke_callback+0xac/0x7b0 cpuhp_thread_fun+0xec/0x180 smpboot_thread_fn+0x1f0/0x290 kthread+0xfc/0x128 ret_from_fork+0x10/0x1c ---[ end trace bea5412bc21deb81 ]--- Similar warning was already there ('HS-I2C is not initialized' message, without backtrace) during system suspend/resume. It looks that cpu freq is resumed too early, before any device drivers, but this is not an issue for this patchset. > --- > drivers/i2c/busses/i2c-exynos5.c | 11 ++--------- > 1 file changed, 2 insertions(+), 9 deletions(-) > > diff --git a/drivers/i2c/busses/i2c-exynos5.c b/drivers/i2c/busses/i2c-exynos5.c > index c1ce2299a76e..fe2f2031ea36 100644 > --- a/drivers/i2c/busses/i2c-exynos5.c > +++ b/drivers/i2c/busses/i2c-exynos5.c > @@ -183,7 +183,6 @@ enum i2c_type_exynos { > > struct exynos5_i2c { > struct i2c_adapter adap; > - unsigned int suspended:1; > > struct i2c_msg *msg; > struct completion msg_complete; > @@ -715,11 +714,6 @@ static int exynos5_i2c_xfer(struct i2c_adapter *adap, > struct exynos5_i2c *i2c = adap->algo_data; > int i, ret; > > - if (i2c->suspended) { > - dev_err(i2c->dev, "HS-I2C is not initialized.\n"); > - return -EIO; > - } > - > ret = clk_enable(i2c->clk); > if (ret) > return ret; > @@ -847,8 +841,7 @@ static int exynos5_i2c_suspend_noirq(struct device *dev) > { > struct exynos5_i2c *i2c = dev_get_drvdata(dev); > > - i2c->suspended = 1; > - > + i2c_mark_adapter_suspended(&i2c->adap, true); > clk_unprepare(i2c->clk); > > return 0; > @@ -871,7 +864,7 @@ static int exynos5_i2c_resume_noirq(struct device *dev) > > exynos5_i2c_init(i2c); > clk_disable(i2c->clk); > - i2c->suspended = 0; > + i2c_mark_adapter_suspended(&i2c->adap, false); > > return 0; > } Best regards
> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Thanks for testing! > Similar warning was already there ('HS-I2C is not initialized' message, > without backtrace) during system suspend/resume. It looks that cpu freq > is resumed too early, before any device drivers, but this is not an > issue for this patchset. Yes, this is actually a desired outcome :) Raise awareness if the suspend/resume ordering has issues.
diff --git a/drivers/i2c/busses/i2c-exynos5.c b/drivers/i2c/busses/i2c-exynos5.c index c1ce2299a76e..fe2f2031ea36 100644 --- a/drivers/i2c/busses/i2c-exynos5.c +++ b/drivers/i2c/busses/i2c-exynos5.c @@ -183,7 +183,6 @@ enum i2c_type_exynos { struct exynos5_i2c { struct i2c_adapter adap; - unsigned int suspended:1; struct i2c_msg *msg; struct completion msg_complete; @@ -715,11 +714,6 @@ static int exynos5_i2c_xfer(struct i2c_adapter *adap, struct exynos5_i2c *i2c = adap->algo_data; int i, ret; - if (i2c->suspended) { - dev_err(i2c->dev, "HS-I2C is not initialized.\n"); - return -EIO; - } - ret = clk_enable(i2c->clk); if (ret) return ret; @@ -847,8 +841,7 @@ static int exynos5_i2c_suspend_noirq(struct device *dev) { struct exynos5_i2c *i2c = dev_get_drvdata(dev); - i2c->suspended = 1; - + i2c_mark_adapter_suspended(&i2c->adap, true); clk_unprepare(i2c->clk); return 0; @@ -871,7 +864,7 @@ static int exynos5_i2c_resume_noirq(struct device *dev) exynos5_i2c_init(i2c); clk_disable(i2c->clk); - i2c->suspended = 0; + i2c_mark_adapter_suspended(&i2c->adap, false); return 0; }
Rejecting transfers should be handled by the core. Also, this will ensure proper locking which was forgotten in this open coded version. Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com> --- drivers/i2c/busses/i2c-exynos5.c | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-)