Messages in this thread | | | Date | Thu, 25 Apr 2013 16:16:43 -0400 | From | Eduardo Valentin <> | Subject | Re: [PATCH 1/1] staging: ti-soc-thermal: remove usage of IS_ERR_OR_NULL |
| |
Russell,
Thanks for your time reviewing this patch.
On 25-04-2013 13:44, Russell King wrote: > On Thu, Apr 25, 2013 at 09:46:16AM -0400, Eduardo Valentin wrote: >> diff --git a/drivers/staging/ti-soc-thermal/ti-bandgap.c b/drivers/staging/ti-soc-thermal/ti-bandgap.c >> index f20c1cf..5027833 100644 >> --- a/drivers/staging/ti-soc-thermal/ti-bandgap.c >> +++ b/drivers/staging/ti-soc-thermal/ti-bandgap.c >> @@ -469,7 +469,7 @@ static inline int ti_bandgap_validate(struct ti_bandgap *bgp, int id) >> { >> int ret = 0; >> >> - if (IS_ERR_OR_NULL(bgp)) { >> + if (!bgp || IS_ERR(bgp)) { >> pr_err("%s: invalid bandgap pointer\n", __func__); >> ret = -EINVAL; > > We really don't need these kinds of "bad pointer passed to a function" > checks in the kernel. Just ensure that all callers have correctly > error checked the return value.
Right.. I will check if it is safe to simply remove the above check.
> >> goto exit; >> @@ -1191,7 +1191,7 @@ int ti_bandgap_probe(struct platform_device *pdev) >> int clk_rate, ret = 0, i; >> >> bgp = ti_bandgap_build(pdev); >> - if (IS_ERR_OR_NULL(bgp)) { >> + if (IS_ERR(bgp)) { >> dev_err(&pdev->dev, "failed to fetch platform data\n"); >> return PTR_ERR(bgp); > > That was definitely a bug, good fix. > >> } >> @@ -1207,14 +1207,14 @@ int ti_bandgap_probe(struct platform_device *pdev) >> } >> >> bgp->fclock = clk_get(NULL, bgp->conf->fclock_name); >> - ret = IS_ERR_OR_NULL(bgp->fclock); >> + ret = IS_ERR(bgp->fclock); >> if (ret) { >> dev_err(&pdev->dev, "failed to request fclock reference\n"); >> goto free_irqs; >> } > > I'm not sure that's correct. If free_irqs ends up returning 'ret' then > it will return 0 or 1, not the error code.
Indeed. Good point.
> > if (IS_ERR(bgp->fclock)) { > dev_err... > ret = PTR_ERR(bgp->fclock); > goto free_irqs; > } > > would be much better in that case.
Yes, I will incorporate the above on next version.
> > Also another fix - clk_get(&pdev->dev, "fclk") and arrange for the > clkdev settings for this device. Same for the one below but call > it "div_clk" or something.
Yeh, that may come in a separated patch. I believe the driver just follows the TI OMAP autogeneration design. It just fetches the clk name fro hw database and auto gen the clkdev settings, I believe.
> >> >> bgp->div_clk = clk_get(NULL, bgp->conf->div_ck_name); >> - ret = IS_ERR_OR_NULL(bgp->div_clk); >> + ret = IS_ERR(bgp->div_clk); >> if (ret) { > > Possibly the same problem here if 'ret' ends up being returned as-is.
agreed.
> >> diff --git a/drivers/staging/ti-soc-thermal/ti-thermal-common.c b/drivers/staging/ti-soc-thermal/ti-thermal-common.c >> index 8e67ebf..4c5f55c37 100644 >> --- a/drivers/staging/ti-soc-thermal/ti-thermal-common.c >> +++ b/drivers/staging/ti-soc-thermal/ti-thermal-common.c >> @@ -101,7 +101,7 @@ static inline int ti_thermal_get_temp(struct thermal_zone_device *thermal, >> >> pcb_tz = data->pcb_tz; >> /* In case pcb zone is available, use the extrapolation rule with it */ >> - if (!IS_ERR_OR_NULL(pcb_tz)) { >> + if (!IS_ERR(pcb_tz)) { >> ret = thermal_zone_get_temp(pcb_tz, &pcb_temp); >> if (!ret) { >> tmp -= pcb_temp; /* got a valid PCB temp */ >> @@ -124,7 +124,7 @@ static int ti_thermal_bind(struct thermal_zone_device *thermal, >> struct ti_thermal_data *data = thermal->devdata; >> int id; >> >> - if (IS_ERR_OR_NULL(data)) >> + if (!data || IS_ERR(data)) >> return -ENODEV; >> >> /* check if this is the cooling device we registered */ >> @@ -146,7 +146,7 @@ static int ti_thermal_unbind(struct thermal_zone_device *thermal, >> { >> struct ti_thermal_data *data = thermal->devdata; >> >> - if (IS_ERR_OR_NULL(data)) >> + if (!data || IS_ERR(data)) >> return -ENODEV; >> >> /* check if this is the cooling device we registered */ >> @@ -282,6 +282,7 @@ static struct ti_thermal_data >> data->sensor_id = id; >> data->bgp = bgp; >> data->mode = THERMAL_DEVICE_ENABLED; >> + /* pcb_tz will be either valid or PTR_ERR() */ >> data->pcb_tz = thermal_zone_get_zone_by_name("pcb"); >> INIT_WORK(&data->thermal_wq, ti_thermal_work); >> >> @@ -295,7 +296,7 @@ int ti_thermal_expose_sensor(struct ti_bandgap *bgp, int id, >> >> data = ti_bandgap_get_sensor_data(bgp, id); >> >> - if (IS_ERR_OR_NULL(data)) >> + if (!data || IS_ERR(data)) >> data = ti_thermal_build_data(bgp, id); >> >> if (!data) >> @@ -306,7 +307,7 @@ int ti_thermal_expose_sensor(struct ti_bandgap *bgp, int id, >> OMAP_TRIP_NUMBER, 0, data, &ti_thermal_ops, >> NULL, FAST_TEMP_MONITORING_RATE, >> FAST_TEMP_MONITORING_RATE); >> - if (IS_ERR_OR_NULL(data->ti_thermal)) { >> + if (IS_ERR(data->ti_thermal)) { >> dev_err(bgp->dev, "thermal zone device is NULL\n"); >> return PTR_ERR(data->ti_thermal); >> } >> @@ -343,7 +344,7 @@ int ti_thermal_register_cpu_cooling(struct ti_bandgap *bgp, int id) >> struct ti_thermal_data *data; >> >> data = ti_bandgap_get_sensor_data(bgp, id); >> - if (IS_ERR_OR_NULL(data)) >> + if (!data || IS_ERR(data)) >> data = ti_thermal_build_data(bgp, id); > > I can't tell whether the above are correct or not as I can't see the > original code.
That is in staging/next.
> >> >> if (!data) >> @@ -356,7 +357,7 @@ int ti_thermal_register_cpu_cooling(struct ti_bandgap *bgp, int id) >> >> /* Register cooling device */ >> data->cool_dev = cpufreq_cooling_register(cpu_present_mask); >> - if (IS_ERR_OR_NULL(data->cool_dev)) { >> + if (IS_ERR(data->cool_dev)) { >> dev_err(bgp->dev, >> "Failed to register cpufreq cooling device\n"); >> return PTR_ERR(data->cool_dev); > > Definitely a bug - and another good fix.
Thanks. >
| |