Re: [PATCH -next] clk: tegra: clk-dfll: indicate correct error reason in tegra_dfll_register

[Thierry Reding <thierry.reding@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2027 bytes --]

On Wed, Sep 23, 2020 at 04:32:40PM -0700, Stephen Boyd wrote:
> Quoting Thierry Reding (2020-09-23 01:16:54)
> > On Tue, Sep 22, 2020 at 12:57:46PM -0700, Stephen Boyd wrote:
> > > Quoting Qilong Zhang (2020-09-18 02:46:42)
> > > > From: Zhang Qilong <zhangqilong3@huawei.com>
> > > > 
> > > > Calling devm_ioremap means getting devices resource have been
> > > > successful. When remap operation failed, we should return '-ENOMEM'
> > > > instead of '-ENODEV' to differentiate between getting resource and
> > > > mapping memory for reminding callers. Moreover, it is not consistent
> > > > with devm_kzalloc operation.
> > > > 
> > > > Signed-off-by: Zhang Qilong <zhangqilong3@huawei.com>
> > > > ---
> > > >  drivers/clk/tegra/clk-dfll.c | 8 ++++----
> > > >  1 file changed, 4 insertions(+), 4 deletions(-)
> > > > 
> > > > diff --git a/drivers/clk/tegra/clk-dfll.c b/drivers/clk/tegra/clk-dfll.c
> > > > index cfbaa90c7adb..6637b73be9f1 100644
> > > > --- a/drivers/clk/tegra/clk-dfll.c
> > > > +++ b/drivers/clk/tegra/clk-dfll.c
> > > > @@ -1993,7 +1993,7 @@ int tegra_dfll_register(struct platform_device *pdev,
> > > >         td->base = devm_ioremap(td->dev, mem->start, resource_size(mem));
> > > >         if (!td->base) {
> > > >                 dev_err(td->dev, "couldn't ioremap DFLL control registers\n");
> > > > -               return -ENODEV;
> > > > +               return -ENOMEM;
> > > 
> > > Can you remove the dev_err() lines too? They're pretty much useless.
> > 
> > I find them somewhat useful because they indicate which particular
> > resource wasn't properly mapped. If we get an -ENOMEM without the error
> > message, we'll have to go and guess which one it is.
> > 
> 
> Doesn't that print the stacktrace when we run out of memory?

slab allocator functions like kmalloc() and friends do, but I'm not
aware of ioremap() doing so. Perhaps if it runs out of virtual memory
for the mapping it would, but there are other reasons why this can fail.

Thierry

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]