linux-kernel - Re: [PATCH] cifs: Add information about noserverino

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20101206095725.78422138@tlielax.poochiereds.net>
Date:	Mon, 6 Dec 2010 09:57:25 -0500
From:	Jeff Layton <jlayton@...ba.org>
To:	Bernhard Walle <bernhard@...lle.de>
Cc:	sfrench@...ba.org, linux-cifs@...r.kernel.org,
	samba-technical@...ts.samba.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] cifs: Add information about noserverino

On Sun,  5 Dec 2010 18:07:35 +0100
Bernhard Walle <bernhard@...lle.de> wrote:

> In my case I had the problem that 32 bit userspace applications in an
> amd64 environment was not able to list the directories of a CIFS-mounted
> share. The simple userspace code
> 
>     int main(int argc, char *argv[])
>     {
>         DIR *dir;
>         struct dirent *dirent;
> 
>         if (!argv[1]) {
>             fprintf(stderr, "Usage: %s <directory>\n", argv[0]);
>             return -1;
>         }
> 
>         dir = opendir(argv[1]);
>         if (!dir) {
>             perror("Unable to open directory");
>             return -1;
>         }
> 
>         while ((dirent = readdir(dir)) != NULL)
>             puts(dirent->d_name);
> 
>         closedir(dir);
> 
>         return 0;
>     }
> 
> was sufficient to trigger the problem.
> 
> I discovered that the problem was that the inodes were too large to fit
> in a 32 bit (unsigned long) integer, so the compat_filldir() function
> returned -EOVERFLOW.
> 
> While that is okay it would have saved me a some hours of debugging if
> the message below would have appeared in my kernel log.
> 
> The target was a Samba server (I guess) of a Buffalo LinkStation Duo
> with the unmodified vendor firmware 1.34.
> 
> Signed-off-by: Bernhard Walle <bernhard@...lle.de>
> ---
>  fs/cifs/readdir.c |   23 +++++++++++++++++++----
>  1 files changed, 19 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c
> index d5e591f..d979826 100644
> --- a/fs/cifs/readdir.c
> +++ b/fs/cifs/readdir.c
> @@ -773,6 +773,7 @@ int cifs_readdir(struct file *file, void *direntry, filldir_t filldir)
>  	char *tmp_buf = NULL;
>  	char *end_of_smb;
>  	unsigned int max_len;
> +	int err;
>  
>  	xid = GetXid();
>  
> @@ -783,17 +784,31 @@ int cifs_readdir(struct file *file, void *direntry, filldir_t filldir)
>  
>  	switch ((int) file->f_pos) {
>  	case 0:
> -		if (filldir(direntry, ".", 1, file->f_pos,
> -		     file->f_path.dentry->d_inode->i_ino, DT_DIR) < 0) {
> +		err = filldir(direntry, ".", 1, file->f_pos,
> +		     file->f_path.dentry->d_inode->i_ino, DT_DIR);
> +		if (err < 0) {
>  			cERROR(1, "Filldir for current dir failed");
> +			if (err == -EOVERFLOW) {
> +				cERROR(1, "Server inodes are too large for 32 "
> +						"bit userspace. You might "
> +						"consider using 'noserverino' "
> +						"mount option for this mount.");
> +			}
>  			rc = -ENOMEM;
>  			break;
>  		}
>  		file->f_pos++;
>  	case 1:
> -		if (filldir(direntry, "..", 2, file->f_pos,
> -		     file->f_path.dentry->d_parent->d_inode->i_ino, DT_DIR) < 0) {
> +		err = filldir(direntry, "..", 2, file->f_pos,
> +		     file->f_path.dentry->d_parent->d_inode->i_ino, DT_DIR);
> +		if (err < 0) {
>  			cERROR(1, "Filldir for parent dir failed");
> +			if (err == -EOVERFLOW) {
> +				cERROR(1, "Server inodes are too large for 32 "
> +						"bit userspace. You might "
> +						"consider using 'noserverino' "
> +						"mount option for this mount.");
> +			}
>  			rc = -ENOMEM;
>  			break;
>  		}

This doesn't look right to me...

i_ino is an unsigned long. The code in filldir() is this:

        unsigned long d_ino;

[...]

        d_ino = ino;
        if (sizeof(d_ino) < sizeof(ino) && d_ino != ino) {
                buf->error = -EOVERFLOW;
                return -EOVERFLOW;
        }

...so the first condition will return true on a 32-bit arch, but it's
not clear to me how you'd get the second one to return true in this
situation.

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