public inbox for u-boot@lists.denx.de
 help / color / mirror / Atom feed
From: Andre Przywara <andre.przywara@arm.com>
To: Simon Glass <sjg@chromium.org>
Cc: "Lothar Waßmann" <LW@karo-electronics.de>,
	"Tom Rini" <trini@konsulko.com>,
	"Heinrich Schuchardt" <xypron.glpk@gmx.de>,
	u-boot@lists.denx.de
Subject: Re: [PATCH v2 3/3] fdt: introduce fsapply command
Date: Mon, 13 Feb 2023 17:26:15 +0000	[thread overview]
Message-ID: <20230213172615.17fb108c@donnerap.cambridge.arm.com> (raw)
In-Reply-To: <CAPnjgZ040ckPUN1SmJ3PtOd93XQW3F4L35b4RCq=O9-kezUpgg@mail.gmail.com>

On Fri, 10 Feb 2023 09:05:34 -0700
Simon Glass <sjg@chromium.org> wrote:

Hi Simon,

> On Fri, 10 Feb 2023 at 04:32, Lothar Waßmann <LW@karo-electronics.de> wrote:
> >
> > Hi,
> >
> > On Fri, 10 Feb 2023 11:02:13 +0000 Andre Przywara wrote:  
> > > Explicitly specifying the exact filenames of devicetree overlays files
> > > on a U-Boot command line can be quite tedious for users, especially
> > > when it should be made persistent for every boot.
> > >
> > > To simplify the task of applying (custom) DT overlays, introduce a
> > > "fdt fsapply" subcommand, that iterates a given directory in any
> > > supported filesystem, and tries to apply every .dtbo file found it
> > > there.
> > >
> > > This allows users to simply drop a DT overlay file into a magic
> > > directory, and it will be applied on the next boot automatically,
> > > by the virtue of just a generic U-Boot command call.
> > >
> > > Signed-off-by: Andre Przywara <andre.przywara@arm.com>
> > > ---
> > >  cmd/fdt.c | 87 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > >  1 file changed, 87 insertions(+)  
> 
> Please add some help at doc/usage/cmd

Right, will do.

> Also please add a test for this subcommand in test/cmd

Yeah, I knew you would say that ;-)
It's still the same problem as last time: sandboxfs doesn't implement
.readdir, so this doesn't work easily there. So I started with filling this
gap, and was just wondering if I should piggy back on the already existing
sandbox_fs_ls abstraction, and somewhat re-translate this back into dirent
structures, or whether I should properly wrap
{open,read,close}dir in os_*dir() helpers, and build sandbox_fs_readdir()
based on that?

Any advice? Both seem equally doable.

Cheers,
Andre

> > > diff --git a/cmd/fdt.c b/cmd/fdt.c
> > > index 1972490bdc2..00f92dbbb5d 100644
> > > --- a/cmd/fdt.c
> > > +++ b/cmd/fdt.c
> > > @@ -127,6 +129,81 @@ static int fdt_get_header_value(int argc, char *const argv[])
> > >       return CMD_RET_FAILURE;
> > >  }
> > >
> > > +#ifdef CONFIG_OF_LIBFDT_OVERLAY
> > > +static int apply_all_overlays(const char *ifname, const char *dev_part_str,
> > > +                           const char *dirname)
> > > +{
> > > +     unsigned long addr;
> > > +     struct fdt_header *dtbo;
> > > +     const char *addr_str;
> > > +     struct fs_dir_stream *dirs;
> > > +     struct fs_dirent *dent;
> > > +     char fname[256], *name_beg;
> > > +     int ret;
> > > +
> > > +     addr_str = env_get("fdtoverlay_addr_r");
> > > +     if (!addr_str) {
> > > +             printf("Invalid fdtoverlay_addr_r for loading overlays\n");
> > > +             return CMD_RET_FAILURE;
> > > +     }
> > > +     addr = hextoul(addr_str, NULL);
> > > +
> > > +     ret = fs_set_blk_dev(ifname, dev_part_str, FS_TYPE_ANY);
> > > +     if (ret)
> > > +             return CMD_RET_FAILURE;
> > > +
> > > +     if (!dirname)
> > > +             dirname = "/";
> > > +     dirs = fs_opendir(dirname);
> > > +     if (!dirs) {
> > > +             printf("Cannot find directory \"%s\"\n", dirname);
> > > +             return CMD_RET_FAILURE;
> > > +     }
> > > +
> > > +     strcpy(fname, dirname);
> > > +     name_beg = strchr(fname, 0);
> > > +     if (name_beg[-1] != '/')
> > > +             *name_beg++ = '/';
> > > +
> > > +     dtbo = map_sysmem(addr, 0);
> > > +     while ((dent = fs_readdir(dirs))) {
> > > +             loff_t size = 0;
> > > +
> > > +             if (dent->type == FS_DT_DIR)
> > > +                     continue;
> > > +
> > > +             if (strcmp(dent->name + strlen(dent->name) - 5, ".dtbo"))
> > > +                     continue;
> > > +
> > > +             printf("%s: ", dent->name);
> > > +             strcpy(name_beg, dent->name);
> > > +             fs_set_blk_dev(ifname, dev_part_str, FS_TYPE_ANY);
> > > +             if (dent->size > SZ_2M)
> > > +                     size = SZ_2M;
> > > +             else
> > > +                     size = dent->size;
> > > +             ret = fs_read(fname, addr, 0, size, &size);
> > > +             if (ret) {
> > > +                     printf("  errno: %d\n", ret);
> > > +                     continue;
> > > +             }
> > > +             if (!fdt_valid(&dtbo)) {
> > > +                     /* fdt_valid() clears the pointer upon failure */
> > > +                     dtbo = map_sysmem(addr, 0);
> > > +                     continue;
> > > +             }
> > > +
> > > +             if (fdt_overlay_apply_verbose(working_fdt, dtbo) == 0)
> > > +                     printf("applied\n");
> > > +     }
> > > +     unmap_sysmem(dtbo);
> > > +
> > > +     fs_closedir(dirs);
> > > +
> > > +     return 0;  
> > return CMD_RET_SUCCESS;  
> 
> There is no need for that...0 means success in U-Boot. It is shorter
> and clearer IMO.
> 
> Regards,
> SImon


  reply	other threads:[~2023-02-13 17:26 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-02-10 11:02 [PATCH v2 0/3] fdt: introduce "fsapply" command Andre Przywara
2023-02-10 11:02 ` [PATCH v2 1/3] cmd: fdt: move: Use map_sysmem to convert pointers Andre Przywara
2023-02-13  0:34   ` Simon Glass
2023-02-10 11:02 ` [PATCH v2 2/3] cmd: fdt: allow standalone "fdt move" Andre Przywara
2023-02-10 11:34   ` Lothar Waßmann
2023-02-13  0:34     ` Simon Glass
2023-02-10 11:02 ` [PATCH v2 3/3] fdt: introduce fsapply command Andre Przywara
2023-02-10 11:32   ` Lothar Waßmann
2023-02-10 16:05     ` Simon Glass
2023-02-13 17:26       ` Andre Przywara [this message]
2023-02-13 18:12         ` Simon Glass
2023-02-16 16:13 ` [PATCH v2 0/3] fdt: introduce "fsapply" command Heinrich Schuchardt

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230213172615.17fb108c@donnerap.cambridge.arm.com \
    --to=andre.przywara@arm.com \
    --cc=LW@karo-electronics.de \
    --cc=sjg@chromium.org \
    --cc=trini@konsulko.com \
    --cc=u-boot@lists.denx.de \
    --cc=xypron.glpk@gmx.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox