The current error handling for calls such as devm_clk_get_enabled() in
the adi-axi-adc probe() function means that, if a property such as
'clocks' (for example) is not present in the devicetree when booting a
kernel with the driver enabled, the resulting error message will be
vague, e.g.:
|adi_axi_adc 44a00000.backend: probe with driver adi_axi_adc failed with error -2
Change the devm_clk_get_enabled(), devm_regmap_init_mmio(), and
devm_iio_backend_register() checks to use dev_err_probe() with some
context for easier debugging.
After the fix:
|adi_axi_adc 44a00000.backend: error -ENOENT: failed to get clock
|adi_axi_adc 44a00000.backend: probe with driver adi_axi_adc failed with error -2
Signed-off-by: Trevor Gamblin <[email protected]>
---
drivers/iio/adc/adi-axi-adc.c | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/drivers/iio/adc/adi-axi-adc.c b/drivers/iio/adc/adi-axi-adc.c
index 0cf0d81358fd..99998c2769d5 100644
--- a/drivers/iio/adc/adi-axi-adc.c
+++ b/drivers/iio/adc/adi-axi-adc.c
@@ -292,7 +292,8 @@ static int adi_axi_adc_probe(struct platform_device *pdev)
st->regmap = devm_regmap_init_mmio(&pdev->dev, base,
&axi_adc_regmap_config);
if (IS_ERR(st->regmap))
- return PTR_ERR(st->regmap);
+ return dev_err_probe(&pdev->dev, PTR_ERR(st->regmap),
+ "failed to init register map\n");
expected_ver = device_get_match_data(&pdev->dev);
if (!expected_ver)
@@ -300,7 +301,8 @@ static int adi_axi_adc_probe(struct platform_device *pdev)
clk = devm_clk_get_enabled(&pdev->dev, NULL);
if (IS_ERR(clk))
- return PTR_ERR(clk);
+ return dev_err_probe(&pdev->dev, PTR_ERR(clk),
+ "failed to get clock\n");
/*
* Force disable the core. Up to the frontend to enable us. And we can
@@ -328,7 +330,8 @@ static int adi_axi_adc_probe(struct platform_device *pdev)
ret = devm_iio_backend_register(&pdev->dev, &adi_axi_adc_generic, st);
if (ret)
- return ret;
+ return dev_err_probe(&pdev->dev, ret,
+ "failed to register iio backend\n");
dev_info(&pdev->dev, "AXI ADC IP core (%d.%.2d.%c) probed\n",
ADI_AXI_PCORE_VER_MAJOR(ver),
--
2.45.2
On Thu, 2024-06-13 at 12:34 -0400, Trevor Gamblin wrote:
> The current error handling for calls such as devm_clk_get_enabled() in
> the adi-axi-adc probe() function means that, if a property such as
> 'clocks' (for example) is not present in the devicetree when booting a
> kernel with the driver enabled, the resulting error message will be
> vague, e.g.:
>
> > adi_axi_adc 44a00000.backend: probe with driver adi_axi_adc failed with error -2
>
> Change the devm_clk_get_enabled(), devm_regmap_init_mmio(), and
> devm_iio_backend_register() checks to use dev_err_probe() with some
> context for easier debugging.
>
> After the fix:
>
> > adi_axi_adc 44a00000.backend: error -ENOENT: failed to get clock
> > adi_axi_adc 44a00000.backend: probe with driver adi_axi_adc failed with error -2
>
> Signed-off-by: Trevor Gamblin <[email protected]>
> ---
Somehow feel that in these cases the error log should come from the functions we're
calling but bah... likely not going happen/change:
Reviewed-by: Nuno Sa <[email protected]>
(As a suggestion, you may do similar work in the axi-dac driver)
- Nuno Sá
On 2024-06-14 5:11 a.m., Nuno Sá wrote:
> On Thu, 2024-06-13 at 12:34 -0400, Trevor Gamblin wrote:
>> The current error handling for calls such as devm_clk_get_enabled() in
>> the adi-axi-adc probe() function means that, if a property such as
>> 'clocks' (for example) is not present in the devicetree when booting a
>> kernel with the driver enabled, the resulting error message will be
>> vague, e.g.:
>>
>>> adi_axi_adc 44a00000.backend: probe with driver adi_axi_adc failed with error -2
>> Change the devm_clk_get_enabled(), devm_regmap_init_mmio(), and
>> devm_iio_backend_register() checks to use dev_err_probe() with some
>> context for easier debugging.
>>
>> After the fix:
>>
>>> adi_axi_adc 44a00000.backend: error -ENOENT: failed to get clock
>>> adi_axi_adc 44a00000.backend: probe with driver adi_axi_adc failed with error -2
>> Signed-off-by: Trevor Gamblin <[email protected]>
>> ---
> Somehow feel that in these cases the error log should come from the functions we're
> calling but bah... likely not going happen/change:
>
> Reviewed-by: Nuno Sa <[email protected]>
>
> (As a suggestion, you may do similar work in the axi-dac driver)
Thanks. I'll send that early next week after a quick test on a board.
Trevor
>
> - Nuno Sá
>
>
On Fri, 14 Jun 2024 15:18:12 -0400
Trevor Gamblin <[email protected]> wrote:
> On 2024-06-14 5:11 a.m., Nuno Sá wrote:
> > On Thu, 2024-06-13 at 12:34 -0400, Trevor Gamblin wrote:
> >> The current error handling for calls such as devm_clk_get_enabled() in
> >> the adi-axi-adc probe() function means that, if a property such as
> >> 'clocks' (for example) is not present in the devicetree when booting a
> >> kernel with the driver enabled, the resulting error message will be
> >> vague, e.g.:
> >>
> >>> adi_axi_adc 44a00000.backend: probe with driver adi_axi_adc failed with error -2
> >> Change the devm_clk_get_enabled(), devm_regmap_init_mmio(), and
> >> devm_iio_backend_register() checks to use dev_err_probe() with some
> >> context for easier debugging.
> >>
> >> After the fix:
> >>
> >>> adi_axi_adc 44a00000.backend: error -ENOENT: failed to get clock
> >>> adi_axi_adc 44a00000.backend: probe with driver adi_axi_adc failed with error -2
> >> Signed-off-by: Trevor Gamblin <[email protected]>
> >> ---
> > Somehow feel that in these cases the error log should come from the functions we're
> > calling but bah... likely not going happen/change:
> >
> > Reviewed-by: Nuno Sa <[email protected]>
> >
> > (As a suggestion, you may do similar work in the axi-dac driver)
>
> Thanks. I'll send that early next week after a quick test on a board.
I tweaked the commit text to say
After the change:
as I don't want to see this picked up by a bot for stable.
As a side note, you could have taken the opportunity to add
struct device *dev = pdev->dev;
which I think would have reduced a few line lengths without loosing
readability. Not important though and perhaps not even a good idea :)
Applied to the togreg branch of iio.git and pushed out as testing.
Note I plan to rebase my tree shortly to avoid some merge conflicts
that otherwise occur. I'll probably do that before I push out
other than as testing.
Thanks,
Jonathan
>
> Trevor
>
> >
> > - Nuno Sá
> >
> >
>