Re: [PATCH v15 06/16] of/fdt: add helper functions for handling properties

["AKASHI, Takahiro" <takahiro.akashi@linaro.org>]

Frank,

On Tue, Oct 09, 2018 at 11:02:29AM -0700, Frank Rowand wrote:
> On 10/04/18 22:06, AKASHI, Takahiro wrote:
> > Frank,
> > 
> > # I haven't reply to your comments.
> > 
> > On Fri, Sep 28, 2018 at 02:13:58PM -0700, Frank Rowand wrote:
> >> On 09/28/18 06:44, Rob Herring wrote:
> >>> +David Gibson
> >>>
> >>> On Fri, Sep 28, 2018 at 1:48 AM AKASHI Takahiro
> >>> <takahiro.akashi@linaro.org> wrote:
> >>>>
> >>>> These functions will be used later to handle kexec-specific properties
> >>>> in arm64's kexec_file implementation.
> >>
> >> As I requested in version 14:
> >>
> >>   The intent of the helper functions is related to properties whose values are
> >>    tuples of the same format as the "reg" property of the "/memory" nodes.  For
> >>    example, the "linux,usable-memory-range" and "linux,elfcoredhr" properties of
> >>    the "/chosen" node.
> >>
> >>    The patch header and the function names should be updated to reflect this intent.
> > 
> > I agree regarding the patch header.
> > 
> >>
> >>>>
> >>>> Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
> >>>> Cc: Rob Herring <robh+dt@kernel.org>
> >>>> Cc: Frank Rowand <frowand.list@gmail.com>
> >>>> Cc: devicetree@vger.kernel.org
> >>>> ---
> >>
> >> Missing list of changes since version 14.
> > 
> > Sorry for the inconvenience, but a whole change list goes into
> > the cover letter, not individual patches.
> 
> That works for me, but please add us to the distro list for the cover letter.

Sure, I will try to do so.

> 
> >>>>  drivers/of/fdt.c       | 56 ++++++++++++++++++++++++++++++++++++++++++
> >>>>  include/linux/of_fdt.h |  4 +++
> >>>>  2 files changed, 60 insertions(+)
> >>>>
> >>>> diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
> >>>> index 800ad252cf9c..c65c31562ccb 100644
> >>>> --- a/drivers/of/fdt.c
> >>>> +++ b/drivers/of/fdt.c
> >>>> @@ -25,6 +25,7 @@
> >>>>  #include <linux/debugfs.h>
> >>>>  #include <linux/serial_core.h>
> >>>>  #include <linux/sysfs.h>
> >>>> +#include <linux/types.h>
> >>>>
> >>>>  #include <asm/setup.h>  /* for COMMAND_LINE_SIZE */
> >>>>  #include <asm/page.h>
> >>>> @@ -1323,3 +1324,58 @@ late_initcall(of_fdt_raw_init);
> >>>>  #endif
> >>>>
> >>>>  #endif /* CONFIG_OF_EARLY_FLATTREE */
> >>
> >> In v14 I requested:
> >>
> >>    Please add comment:
> >>
> >>    /* helper functions for arm64 kexec */
> > 
> > Okay.
> > 
> >>
> >>>> +
> >>>> +#define FDT_ALIGN(x, a)        (((x) + (a) - 1) & ~((a) - 1))
> >>>> +#define FDT_TAGALIGN(x)        (FDT_ALIGN((x), FDT_TAGSIZE))
> >>>> +
> >>>> +int fdt_prop_len(const char *prop_name, int len)
> >>
> >> In v14, I requested:
> >>
> >>    Please rename as fdt_len_added_prop()
> > 
> > Anyhow, I will drop this function, preferring to new
> > fdt_[address|size]_cells().
> > 
> >> I'm not really happy with my suggested name, but do not have a
> >> better one yet.  As Rob notes, maybe David G will have a helpful
> >> comment.
> >>
> >>>> +{
> >>>> +       return (strlen(prop_name) + 1) +
> >>>> +               sizeof(struct fdt_property) +
> >>>> +               FDT_TAGALIGN(len);
> >>>
> >>> Looks like you are using this to calculate how much space you need to
> >>> allocate in addition to the current DTB for a couple of new or
> >>> replaced properties. I'm not sure that this calculation is completely
> >>> accurate. And it is strange there doesn't seem to be any libfdt
> >>> function for this already. It would be simpler to just add some fixed
> >>> additional amount.
> >>>
> >>> Maybe David G has comments on this?
> > 
> > I'm not quit sure why it's not that accurate, but as I said in a reply to
> > David's comment, I will take your suggestion.
> > 
> >>>> +}
> >>>> +
> >>>
> >>> The rest of this should go in drivers/of/fdt_address.c. Ultimately, it
> >>> should go into libfdt, but I'm fine with having it in the kernel for
> >>> now.
> >>>
> >>>> +static void fill_property(void *buf, u64 val64, int cells)
> >>
> >> In v14 I requested:
> >>
> >> Please rename as cpu64_to_fdt_cells()
> > 
> > I don't mind, but this function may be dropped if Rob sticks to
> > u-boot's fdt_pack_reg() over my fdt_setprop_reg().
> 
> I have another comment a couple of minutes ago in the longer thread
> about this function.

Okay.

-Takahiro Akashi

> 
> >>
> >>>> +{
> >>>> +       __be32 val32;
> >>>> +
> >>>> +       while (cells) {
> >>>> +               val32 = cpu_to_fdt32((val64 >> (32 * (--cells))) & U32_MAX);
> >>>> +               memcpy(buf, &val32, sizeof(val32));
> >>>> +               buf += sizeof(val32);
> >>>
> >>> This is kind of hard to read. I would copy u-boot's fdt_pack_reg function.
> >>>
> >>> BTW, for purposes of moving to libfdt, we'll need the authors'
> >>> (Masahiro Yamada and Hans de Goede) permission to dual license.
> >>>
> >>>> +       }
> >>>> +}
> >>>> +
> >>>> +int fdt_setprop_reg(void *fdt, int nodeoffset, const char *name,
> >>>> +                                               u64 addr, u64 size)
> >>>> +{
> >>>> +       int addr_cells, size_cells;
> >>
> >> unsigned
> > 
> > fdt_[address|size]_cell() returns an int.
> 
> I stand corrected.  They take advantage of the fact that valid values
> are in the range 0..4 and return a negative value for error.
> 
> 
> >>>> +       char buf[sizeof(__be32) * 2 * 2];
> >>>> +               /* assume dt_root_[addr|size]_cells <= 2 */
> >>>> +       void *prop;
> >>>> +       size_t buf_size;
> >>>> +
> >>>> +       addr_cells = fdt_address_cells(fdt, 0);
> >>>> +       if (addr_cells < 0)
> >>>> +               return addr_cells;
> >>>> +       size_cells = fdt_size_cells(fdt, 0);
> >>>> +       if (size_cells < 0)
> >>>> +               return size_cells;
> >>>> +
> >>>> +       /* if *_cells >= 2, cells can hold 64-bit values anyway */
> >>>> +       if ((addr_cells == 1) && (addr > U32_MAX))
> >>>> +               return -FDT_ERR_BADVALUE;
> >>>> +
> >>>> +       if ((size_cells == 1) && (size > U32_MAX))
> >>>> +               return -FDT_ERR_BADVALUE;
> >>
> >> In v14 I requested:
> >>
> >>    Should also check that base + size does not wrap around.
> > 
> > Okay, I will start discussion, as you have suggested, in devicetree-spec ML.
> > 
> > Thanks,
> > -Takahiro Akashi
> > 
> >>
> >>>> +
> >>>> +       buf_size = (addr_cells + size_cells) * sizeof(u32);
> >>>> +       prop = buf;
> >>>> +
> >>>> +       fill_property(prop, addr, addr_cells);
> >>>> +       prop += addr_cells * sizeof(u32);
> >>>> +
> >>>> +       fill_property(prop, size, size_cells);
> >>>> +
> >>>> +       return fdt_setprop(fdt, nodeoffset, name, buf, buf_size);
> >>>> +}
> >>>> diff --git a/include/linux/of_fdt.h b/include/linux/of_fdt.h
> >>>> index b9cd9ebdf9b9..842af6ea92ea 100644
> >>>> --- a/include/linux/of_fdt.h
> >>>> +++ b/include/linux/of_fdt.h
> >>>> @@ -108,5 +108,9 @@ static inline void unflatten_device_tree(void) {}
> >>>>  static inline void unflatten_and_copy_device_tree(void) {}
> >>>>  #endif /* CONFIG_OF_EARLY_FLATTREE */
> >>>>
> >>>> +int fdt_prop_len(const char *prop_name, int len);
> >>>> +int fdt_setprop_reg(void *fdt, int nodeoffset, const char *name,
> >>>> +                                               u64 addr, u64 size);
> >>>> +
> >>>>  #endif /* __ASSEMBLY__ */
> >>>>  #endif /* _LINUX_OF_FDT_H */
> >>>> --
> >>>> 2.19.0
> >>>>
> >>>
> >>
> > 
>