fs: Add destroy_inode callbacks to filesystems ...

Several filesystems rely on the default function which frees
the struct inode * rather than the filesystem specific inode
which the inode is embedded in. This works because the inode
is the first element in the filesystem specific inode. Let's
not depend on this behaviour and for clarity add the destroy_inode
callbacks to all filesystems.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

treewide: remove references to CREDITS ...

The CREDITS file was removed from barebox in 2015 by commit 6570288f2d97
("Remove the CREDITS file"). Remove references to it from several files.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

nfs: Do not allow to abort ...

When ctrl-c is pressed then ctrlc() will return true until
ctrlc_handled() is called. This means that once ctrl-c is pressed every
NFS operation will fail until the upper layer calls ctrlc_handled().
When for example we are doing a 'ls -l' on an NFS directory then after
a ctrl-c press not the 'ls -l' aborts, but instead the retrieving of the
directory entries which is not what we want.

Simply do not call ctrlc() in the fs layer. the NFS timeout is 2 seconds
which we have to wait until we have a chance to abort.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: nfs: don't maintain nfs dentries in dcache ...

nfs dentries may change underneath barebox, so do not maintain them in
the dcache.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

nfs: queue received packets ...

It may happen that we receive more than one RPC packet before we come
along to check the result. Instead of handling just a single received
packet and possibly overwriting the last one queue the received RPC
packets.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

nfs: Fix polling for packets ...

It can happen that the first packet we receive is not the desired one.
In this case we have to poll for further packets. set nfs_state back to
STATE_START before polling for more packets, otherwise we check the same
packet again and again.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

nfs: Fix rpc_check_reply() return value for stale packets ...

When we receive a packet with the previous rpc_id then we have
the comment "stale packet, wait a bit longer", but that's not what
the code does. rpc_check_reply() returns 0 in this case and the caller
then interprets the packet as valid.
Always return -EAGAIN for invalid rpc_ids. This lets the caller ignore
the packets as intented.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

nfs: remove unnecessary check ...

In rpc_check_reply() pkt is never NULL, drop the unnecessary patch.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

nfs: Add missing free ...

Add forgotten packet free in error path.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

nfs: Add function to free packets ...

Add to function to free packets rather than freeing them directly. We
will introduce received packet qeueuing with one of the next patches,
this patch is meant to make it better readable.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: nfs: Fix possible buffer overflow ...

nfs_readlink_req() interprets a 32bit value directly received from the
network as length argument to memcpy() without any boundary checking.
Clamp the copy size at the end of the incoming packet.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: nfs: ensure rpc_req message is send ...

Currently we send a rpc message without checking if the send was
succesful and poll for a answer from the server. If the server didn't
answer within the NFS_TIMEOUT window we send the package again. In case
the package send wasn't successful we always run in that timeout. This
gets even worse if the package send fails more than one time.

Check if the package send was successful and resend the package if it
wasn't to fix this behaviour.

Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Merge branch 'for-next/nfs'

fs: let truncate take a loff_t argument ...

loff_t is the correct type for file sizes. Use it to allow to truncate
to sizes bigger than 32bit.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: devfs: Change .lseek callbacks to return 'int' ...

Returning requested offset from .lseek() callback doesn't really give
us any new information while bringing unnecessary
complications. Change all .lseek() types (both in struct struct
cdev_operations and in struct fs_driver_d) to return 'int' and adjust
the rest of the codebase accordingly.

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: Update FILE position in lseek() ...

Instead on relying on driver callbacks to update 'pos' in FILE, do it
as a part of lseek() code. This allows us to drop a bit of repeating
code as well as making lseek() implementation consistent with write()
and read().

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs/nfs: stop using a global variable for nfs packet payload ...

The lifetime of the data provided by a global variable is hard to track.
So move the data pointer into the nfs_priv structure and let rpc_req
return the data to its callers which in turn are responsible to free it.

The callers are changed to use a local variable accordingly. This makes
it plausible that the previous commit catched all usages of the global
data.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs/nfs: copy data from rpc replies to local storage ...

The nfs code uses data provided to the packet handler after net_poll()
returned. But the life time of this data already ended when net_poll()
returns. Most of the time it is possible to get away here but on i.MX28
the data is overwritten since commit 82ec28929cc9 ("net: fec_imx: Do not
use DMA coherent memory for Rx buffers").

So the data from the packet is copied to a malloced buffer that needs
free()ing when the data is not used any more.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs/nfs: don't try to set nfs error code from mount rpc call ...

A packet from a mount rpc call doesn't have an NFS error field, so don't
try to access this.

In the case of the MOUNT_UMOUNT procedure the reply package is short
such that accessing the u32 after the rpc_reply structure is already
after the end of the packet. Apart from the access to out-of-packet data
there is no harm because the wrongly read value is unused. But make this
more explicit by only using nfserr if the call was an NFS request.

Fixes: 9ede56ad2476 ("fs: Add NFS support")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

net: dns: return error codes ...

The resolv() function used to return the IP address. When net_udp_new()
fails we return an error code though which the callers of resolv() take
as an IP address. This is wrong of course and we could return 0 in this
case. Instead we return an error code and pass the resolved IP as a
pointer which allows us to return proper error codes.

This patch also adds error messages and error returns to the various
callers of resolv() which used to just continue with a zero IP and
let the user figure out what went wrong.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: nfs: Switch to dentry cache implementation ...

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

move parseopt to lib/ ...

Signed-off-by: Antony Pavlov <antonynpavlov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: nfs: pick up network interface bootargs parameter ...

This adds the linux.bootarg device parameter from the network device of
the current nfs connection and adds it to the nfs bootargs line.

This allows booting from nfs without manually setting a ip=dhcp or
ip=<ipaddr> option.

Signed-off-by: Enrico Jorns <ejo@pengutronix.de>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

convert users to %pI4 ...

Convert users of ip_to_string() and print_IPaddr() to %pI4 and
remove the now unused functions.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

string: Fix (v)asprintf prototypes ...

Our asprintf and vasprintf have different prototypes than the glibc
functions. This causes trouble when we want to share barebox code
with userspace code. Change the prototypes for (v)asprintf to match
the glibc prototypes. Since the current (v)asprintf are convenient
to use change the existing functions to b(v)asprintf.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

nfs: forward filesystem options to the kernel command line ...

Using NFS in conjunction with boot spec and the feature to let Barebox
auto generate a kernel command line must keep the options the NFS filesystem
was mounted in Barebox. This patch extends the kernel command line parameter
on demand if something different than the defaults are used.

The command:

barebox:/ boot nfs://myhost//root

expands to the kernel command line:

nfsroot=myhost:/root,v3,tcp

while the command:

barebox:/ boot nfs://myhost:2049//root

expands now to the kernel command line:

nfsroot=myhost:/root,v3,tcp,mountport=2049,port=2049

Signed-off-by: Juergen Borleis <jbe@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs/nfs.c: use SUNRPC_PORT remote port by default ...

pico_socket_connect() needs remote_port != 0.

Signed-off-by: Antony Pavlov <antonynpavlov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs/nfs.c: use uint16_t for port numbers ...

Signed-off-by: Antony Pavlov <antonynpavlov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: nfs: Implement Linux rootarg ...

Add the root= parameter for NFS filesystems. We currently hardcode
v3 and tcp which probably needs to become configurable at some point.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

fs: rename inode member of struct filep to priv ...

Because that's what it is. 'inode' will become confusing
once we support real inodes.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>