Re: [PATCH 3/3] firmware loader: allow distribution to choose default search paths

[Takashi Iwai]

At Wed, 5 Jun 2013 17:35:26 +0800,
Ming Lei wrote:
> 
> On Wed, Jun 5, 2013 at 3:10 PM, Takashi Iwai <tiwai@xxxxxxx> wrote:
> > At Wed,  5 Jun 2013 13:42:49 +0800,
> > Ming Lei wrote:
> >>
> >> For some distributions(e.g. android), firmware images aren't put
> >> under kernel built-in search paths, so introduce one Kconfig
> >> option to allow distributions or users to choose its specific default
> >> search paths, which are always tried before searching from kernel
> >> built-in paths in direct loading.
> >>
> >> Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> >> Cc: Takashi Iwai <tiwai@xxxxxxx>
> >> Signed-off-by: Ming Lei <ming.lei@xxxxxxxxxxxxx>
> >> ---
> >>  drivers/base/Kconfig          |   12 +++++++
> >>  drivers/base/firmware_class.c |   76 ++++++++++++++++++++++++++++++++++++-----
> >>  2 files changed, 80 insertions(+), 8 deletions(-)
> >>
> >> diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig
> >> index 07abd9d..5b0c909 100644
> >> --- a/drivers/base/Kconfig
> >> +++ b/drivers/base/Kconfig
> >> @@ -156,6 +156,18 @@ config FW_LOADER_USER_HELPER
> >>         no longer required unless you have a special firmware file that
> >>         resides in a non-standard path.
> >>
> >> +config FW_CUSTOMIZED_PATH
> >> +     string "default firmware search paths for direct loading"
> >> +     help
> >> +       On some distribution(e.g. android), firmware images aren't
> >> +       put under kernel built-in search paths, so provide this option
> >> +       for distributions to choose a distribution specific firmware
> >> +       search path. The option allows to choose more than one path,
> >> +       and paths are seperated with semicolon(e.g. on android, the
> >> +       option might look as "/etc/firmware;/vendor/firmware").
> >> +
> >> +       If you are unsure about this, don't choose here.
> >> +
> >>  config DEBUG_DRIVER
> >>       bool "Driver Core verbose debug messages"
> >>       depends on DEBUG_KERNEL
> >> diff --git a/drivers/base/firmware_class.c b/drivers/base/firmware_class.c
> >> index c743409..50b5913 100644
> >> --- a/drivers/base/firmware_class.c
> >> +++ b/drivers/base/firmware_class.c
> >> @@ -267,6 +267,9 @@ static void fw_free_buf(struct firmware_buf *buf)
> >>  static char fw_path_para[256];
> >>  static const char * const fw_path[] = {
> >>       fw_path_para,
> >> +#ifdef CONFIG_FW_CUSTOMIZED_PATH
> >> +     CONFIG_FW_CUSTOMIZED_PATH,
> >> +#endif
> >>       "/lib/firmware/updates/" UTS_RELEASE,
> >>       "/lib/firmware/updates",
> >>       "/lib/firmware/" UTS_RELEASE,
> >> @@ -314,6 +317,59 @@ static bool fw_read_file_contents(struct file *file, struct firmware_buf *fw_buf
> >>       return true;
> >>  }
> >>
> >> +static bool fw_get_file_firmware(const char *path,
> >> +                              struct firmware_buf *buf)
> >> +{
> >> +     struct file *file;
> >> +     bool success;
> >> +
> >> +     file = filp_open(path, O_RDONLY, 0);
> >> +     if (IS_ERR(file))
> >> +             return false;
> >> +     success = fw_read_file_contents(file, buf);
> >> +     fput(file);
> >> +
> >> +     return success;
> >> +}
> >> +
> >> +#ifdef CONFIG_FW_CUSTOMIZED_PATH
> >> +/* The path in @paths is seperated by ';' */
> >> +static bool fw_get_fw_file_from_paths(const char *paths, char *path,
> >> +                                   struct firmware_buf *buf)
> >> +{
> >> +     int len, start, end;
> >> +     char *pos;
> >> +
> >> +     end = -1;
> >> +     do {
> >> +             start = end + 1;
> >> +             pos = strchr(&paths[start], ';');
> >> +             if (pos) {
> >> +                     end = (int)(pos - paths);
> >> +                     len = end - start;
> >> +             } else {
> >> +                     len = strlen(&paths[start]);
> >> +             }
> >> +
> >> +             if (PATH_MAX < len + strlen(buf->fw_id))
> >> +                     continue;
> >> +             strncpy(path, &paths[start], len);
> >> +             snprintf(&path[len], PATH_MAX - len, "/%s", buf->fw_id);
> >> +
> >> +             if (fw_get_file_firmware(path, buf))
> >> +                     return true;
> >> +     } while (pos && end < strlen(paths) - 1);
> >> +
> >> +     return false;
> >> +}
> >> +#else
> >> +static bool fw_get_fw_file_from_paths(const char *paths, char *path,
> >> +                                   struct firmware_buf *buf)
> >> +{
> >> +     return false;
> >> +}
> >> +#endif
> >> +
> >>  static bool fw_get_filesystem_firmware(struct device *device,
> >>                                      struct firmware_buf *buf)
> >>  {
> >> @@ -322,19 +378,23 @@ static bool fw_get_filesystem_firmware(struct device *device,
> >>       char *path = __getname();
> >>
> >>       for (i = 0; i < ARRAY_SIZE(fw_path); i++) {
> >> -             struct file *file;
> >>
> >>               /* skip the unset customized path */
> >>               if (!fw_path[i][0])
> >>                       continue;
> >>
> >> -             snprintf(path, PATH_MAX, "%s/%s", fw_path[i], buf->fw_id);
> >> -
> >> -             file = filp_open(path, O_RDONLY, 0);
> >> -             if (IS_ERR(file))
> >> -                     continue;
> >> -             success = fw_read_file_contents(file, buf);
> >> -             fput(file);
> >> +             /*
> >> +              * If CONFIG_FW_CUSTOMIZED_PATH is set, search from
> >> +              * these paths first
> >> +              */
> >> +             if (i == 1 && ARRAY_SIZE(fw_path) > 5) {
> >> +                     success = fw_get_fw_file_from_paths(fw_path[1],
> >> +                                                         path, buf);
> >> +             } else {
> >> +                     snprintf(path, PATH_MAX, "%s/%s", fw_path[i],
> >> +                              buf->fw_id);
> >> +                     success = fw_get_file_firmware(path, buf);
> >
> > Shouldn't fw_get_fw_file_from_paths() be applied unconditionally?
> > It'll be benefit for the customized path passed via module option,
> > too.  The only drawback is a slight code growth, but the code can be
> > reduced a bit with strcspn() or such.
> 
> You mean that both the 1st two items should be covered by
> fw_get_fw_file_from_paths()? If so, the function may become
> a bit ugly since two strings are required to pass in, and we can't
> merge one runtime string and one ro string created in compiling.

I meant to simply call fw_get_fw_file_from_paths() for all fw_path[]
entries.  So far, the module option can pass only a single path.
But if it's handled through fw_get_file_from_paths(), you can pass
multiple paths there, too.

> Looks we can let fw_get_fw_file_from_paths() handle all
> predefined paths(CONFIG_FW_CUSTOMIZED_PATH and
> kernel built-in paths), then fw_get_filesystem_firmware()
> may become simple, just check fw_path_para and all
> other paths by fw_get_fw_file_from_paths(). How about the
> idea?
> 
> > BTW, I now wonder what happens if you pass a relative path.
> > Did you already test it?
> 
> I tested absolute paths, and not test relative paths. Do you mean
> it may cause security problem?

Yes.  It just came to my mind while reviewing your patch.

> If so, we can check and ignore them,
> but it should be OK since the paths are provided by kernel builder.
> >From view of function, I don't think there are much difference with
> absolute paths.

The path can be provided via module option, too, so we need to check
it in anyway.


thanks,

Takashi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/