Re: [PATCH 2/2] clk: pmc-atom: use devm_kstrdup_const()

[Mike Rapoport <rppt@linux.vnet.ibm.com>]

On Mon, Aug 27, 2018 at 02:58:46PM +0200, Bartosz Golaszewski wrote:
> 2018-08-27 14:52 GMT+02:00 Mike Rapoport <rppt@linux.vnet.ibm.com>:
> > On Mon, Aug 27, 2018 at 02:28:45PM +0200, Bartosz Golaszewski wrote:
> >> 2018-08-27 12:39 GMT+02:00 Mike Rapoport <rppt@linux.vnet.ibm.com>:
> >> > On Mon, Aug 27, 2018 at 10:21:01AM +0200, Bartosz Golaszewski wrote:
> >> >> Use devm_kstrdup_const() in the pmc-atom driver. This mostly serves as
> >> >> an example of how to use this new routine to shrink driver code.
> >> >>
> >> >> While we're at it: replace a call to kcalloc() with devm_kcalloc().
> >> >>
> >> >> Signed-off-by: Bartosz Golaszewski <brgl@bgdev.pl>
> >> >> ---
> >> >>  drivers/clk/x86/clk-pmc-atom.c | 19 ++++---------------
> >> >>  1 file changed, 4 insertions(+), 15 deletions(-)
> >> >>
> >> >> diff --git a/drivers/clk/x86/clk-pmc-atom.c b/drivers/clk/x86/clk-pmc-atom.c
> >> >> index 08ef69945ffb..daa2192e6568 100644
> >> >> --- a/drivers/clk/x86/clk-pmc-atom.c
> >> >> +++ b/drivers/clk/x86/clk-pmc-atom.c
> >> >> @@ -253,14 +253,6 @@ static void plt_clk_unregister_fixed_rate_loop(struct clk_plt_data *data,
> >> >>               plt_clk_unregister_fixed_rate(data->parents[i]);
> >> >>  }
> >> >>
> >> >> -static void plt_clk_free_parent_names_loop(const char **parent_names,
> >> >> -                                        unsigned int i)
> >> >> -{
> >> >> -     while (i--)
> >> >> -             kfree_const(parent_names[i]);
> >> >> -     kfree(parent_names);
> >> >> -}
> >> >> -
> >> >>  static void plt_clk_unregister_loop(struct clk_plt_data *data,
> >> >>                                   unsigned int i)
> >> >>  {
> >> >> @@ -286,8 +278,8 @@ static const char **plt_clk_register_parents(struct platform_device *pdev,
> >> >>       if (!data->parents)
> >> >>               return ERR_PTR(-ENOMEM);
> >> >>
> >> >> -     parent_names = kcalloc(nparents, sizeof(*parent_names),
> >> >> -                            GFP_KERNEL);
> >> >> +     parent_names = devm_kcalloc(&pdev->dev, nparents,
> >> >> +                                 sizeof(*parent_names), GFP_KERNEL);
> >> >>       if (!parent_names)
> >> >>               return ERR_PTR(-ENOMEM);
> >> >>
> >> >> @@ -300,7 +292,8 @@ static const char **plt_clk_register_parents(struct platform_device *pdev,
> >> >>                       err = PTR_ERR(data->parents[i]);
> >> >>                       goto err_unreg;
> >> >>               }
> >> >> -             parent_names[i] = kstrdup_const(clks[i].name, GFP_KERNEL);
> >> >> +             parent_names[i] = devm_kstrdup_const(&pdev->dev,
> >> >> +                                                  clks[i].name, GFP_KERNEL);
> >> >>       }
> >> >>
> >> >>       data->nparents = nparents;
> >> >> @@ -308,7 +301,6 @@ static const char **plt_clk_register_parents(struct platform_device *pdev,
> >> >>
> >> >>  err_unreg:
> >> >>       plt_clk_unregister_fixed_rate_loop(data, i);
> >> >> -     plt_clk_free_parent_names_loop(parent_names, i);
> >> >
> >> > What happens if clks[i].name is not a part of RO data? The devm_kstrdup_const
> >> > will allocate memory and nothing will ever free it...
> >> >
> >>
> >> I'm looking at it and trying to see if I'm missing something, but
> >> AFAIK the whole concept of devm_* is to leave out the resource
> >> management part.
> >>
> >> devm_kstrdup_const() will internally call devm_kstrdup() for strings
> >> that are not in .rodata and once the device is detached, the string
> >> will be freed (or not if it's in .rodata).
> >
> > And when it's going to be freed, how the resource management will know
> > whether it's .rodata or not?
> >
> 
> If the string to be duplicated is in .rodata, it's returned as is from
> devm_kstrdup_const(). Never gets added to the devres list, never get's
> freed. When the string to be duplicated is not in .rodata,
> devm_kstrdup() is called from devm_kstrdup_const(). Now the string is
> in the devres list of this device and it will get freed on driver
> detach. I really don't see what else could be a problem here.

Thanks for the clarification.
I think that it's worth adding a similar explanation to the changelog of
the first patch.
 
> BR
> Bart
> 
> >> BR
> >> Bart
> >>
> >> > And, please don't drop kfree(parent_names) here.
> >> >
> >> >>       return ERR_PTR(err);
> >> >>  }
> >> >>
> >> >> @@ -351,15 +343,12 @@ static int plt_clk_probe(struct platform_device *pdev)
> >> >>               goto err_unreg_clk_plt;
> >> >>       }
> >> >>
> >> >> -     plt_clk_free_parent_names_loop(parent_names, data->nparents);
> >> >> -
> >> >>       platform_set_drvdata(pdev, data);
> >> >>       return 0;
> >> >>
> >> >>  err_unreg_clk_plt:
> >> >>       plt_clk_unregister_loop(data, i);
> >> >>       plt_clk_unregister_parents(data);
> >> >> -     plt_clk_free_parent_names_loop(parent_names, data->nparents);
> >> >
> >> > Ditto.
> >> >
> >> >>       return err;
> >> >>  }
> >> >>
> >> >> --
> >> >> 2.18.0
> >> >>
> >> >
> >> > --
> >> > Sincerely yours,
> >> > Mike.
> >> >
> >>
> >
> > --
> > Sincerely yours,
> > Mike.
> >
> 

-- 
Sincerely yours,
Mike.