Re: [PATCH] procfs: set d_type on returned direntries

[Samuel Thibault]

Hello,

dnietoc@gmail.com, le dim. 19 janv. 2025 23:50:10 +0000, a ecrit:
> diff --git a/procfs/netfs.c b/procfs/netfs.c
> index 4ed5eab6..3cf7a8e2 100644
> --- a/procfs/netfs.c
> +++ b/procfs/netfs.c
> @@ -115,6 +115,29 @@ error_t netfs_attempt_readlink (struct iouser *user, 
> struct node *np,
>    return 0;
>  }
>  
> +static unsigned char entry_type(char *name)
> +{
> +  char c;
> +  int only_numbers;
> +
> +  if (name[0] == '.' && name[1] == 0)
> +    return DT_DIR;
> +
> +  if (name[0] == '.' && name[1] == '.' && name[2] == 0)
> +    return DT_DIR;
> +
> +  if (strcmp(name, "self") == 0)
> +    return DT_DIR;
> +
> +  for(only_numbers = 1, c = *name; only_numbers && c; c = *++name)
> +    {
> +      if (c < '0' || c > '9')
> +        only_numbers = 0;
> +    }

This is quite fragile indeed, and will fail for subdirectories.

It'd be better to call procfs_lookup to get the actual node and get its
inode type. That'll be much less expensive than applications using
stat() to get it, which will require RPC etc.

Samuel