Home » Mailing lists » Devel » [RFC PATCH 0/2] net: connect to UNIX sockets from specified root
[RFC PATCH 0/2] net: connect to UNIX sockets from specified root [message #47407] |
Fri, 10 August 2012 12:57 |
Stanislav Kinsbursky
Messages: 683 Registered: October 2011
|
Senior Member |
|
|
Today, there is a problem in connecting of local SUNRPC thansports. These
transports uses UNIX sockets and connection itself is done by rpciod
workqueue.
But UNIX sockets lookup is done in context of process file system root. I.e.
all local thunsports are connecting in rpciod context.
This works nice until we will try to mount NFS from process with other root -
for example in container. This container can have it's own (nested) root and
rcpbind process, listening on it's own unix sockets. But NFS mount attempt in
this container will register new service (Lockd for example) in global rpcbind
- not containers's one.
This patch set introduces kernel connect helper for UNIX stream sockets and
modifies unix_find_other() to be able to search from specified root.
It also replaces generic socket connect call for local transports by new
helper in SUNRPC layer.
The following series implements...
---
Stanislav Kinsbursky (2):
unix sockets: add ability for search for peer from passed root
SUNRPC: connect local transports with unix_stream_connect_root() helper
include/net/af_unix.h | 2 ++
net/sunrpc/xprtsock.c | 28 +++++++++++++++++++++++++---
net/unix/af_unix.c | 25 ++++++++++++++++++-------
3 files changed, 45 insertions(+), 10 deletions(-)
|
|
|
[RFC PATCH 2/2] SUNRPC: connect local transports with unix_stream_connect_root() helper [message #47408 is a reply to message #47407] |
Fri, 10 August 2012 12:57 |
Stanislav Kinsbursky
Messages: 683 Registered: October 2011
|
Senior Member |
|
|
Today, there is a problem in connecting of local SUNRPC thansports. These
transports uses UNIX sockets and connection itself is done by rpciod
workqueue.
But UNIX sockets lookup is done in context of process file system root. I.e.
all local thunsports are connecting in rpciod context.
This works nice until we will try to mount NFS from process with other root -
for example in container. This container can have it's own (nested) root and
rcpbind process, listening on it's own unix sockets. But NFS mount attempt in
this container will register new service (Lockd for example) in global rpcbind
- not containers's one.
This patch solves the problem by using special helper
unix_stream_connect_root(), which lookup socket file starting from passed
root.
Signed-off-by: Stanislav Kinsbursky <skinsbursky@parallels.com>
---
net/sunrpc/xprtsock.c | 28 +++++++++++++++++++++++++---
1 files changed, 25 insertions(+), 3 deletions(-)
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 890b03f..01a6f2a 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -37,6 +37,7 @@
#include <linux/sunrpc/svcsock.h>
#include <linux/sunrpc/xprtsock.h>
#include <linux/file.h>
+#include <linux/fs_struct.h>
#ifdef CONFIG_SUNRPC_BACKCHANNEL
#include <linux/sunrpc/bc_xprt.h>
#endif
@@ -45,6 +46,7 @@
#include <net/checksum.h>
#include <net/udp.h>
#include <net/tcp.h>
+#include <net/af_unix.h>
#include "sunrpc.h"
@@ -255,6 +257,11 @@ struct sock_xprt {
void (*old_state_change)(struct sock *);
void (*old_write_space)(struct sock *);
void (*old_error_report)(struct sock *);
+
+ /*
+ * Saved transport creator root. Required for local transports only.
+ */
+ struct path root;
};
/*
@@ -1873,7 +1880,8 @@ static int xs_local_finish_connecting(struct rpc_xprt *xprt,
/* Tell the socket layer to start connecting... */
xprt->stat.connect_count++;
xprt->stat.connect_start = jiffies;
- return kernel_connect(sock, xs_addr(xprt), xprt->addrlen, 0);
+ return unix_stream_connect_root(&transport->root, sock, xs_addr(xprt),
+ xprt->addrlen, 0);
}
/**
@@ -2213,6 +2221,18 @@ static void xs_connect(struct rpc_task *task)
}
}
+static void xs_local_destroy(struct rpc_xprt *xprt)
+{
+ struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
+ struct path root = transport->root;
+
+ dprintk("RPC: xs_local_destroy xprt %p\n", xprt);
+
+ xs_destroy(xprt);
+
+ path_put(&root);
+}
+
/**
* xs_local_print_stats - display AF_LOCAL socket-specifc stats
* @xprt: rpc_xprt struct containing statistics
@@ -2431,7 +2451,7 @@ static struct rpc_xprt_ops xs_local_ops = {
.send_request = xs_local_send_request,
.set_retrans_timeout = xprt_set_retrans_timeout_def,
.close = xs_close,
- .destroy = xs_destroy,
+ .destroy = xs_local_destroy,
.print_stats = xs_local_print_stats,
};
@@ -2607,8 +2627,10 @@ static struct rpc_xprt *xs_setup_local(struct xprt_create *args)
dprintk("RPC: set up xprt to %s via AF_LOCAL\n",
xprt->address_strings[RPC_DISPLAY_ADDR]);
- if (try_module_get(THIS_MODULE))
+ if (try_module_get(THIS_MODULE)) {
+ get_fs_root(current->fs, &transport->root);
return xprt;
+ }
ret = ERR_PTR(-EINVAL);
out_err:
xprt_free(xprt);
|
|
|
|
|
|
|
|
|
|
|
|
Re: [RFC PATCH 0/2] net: connect to UNIX sockets from specified root [message #47453 is a reply to message #47450] |
Sat, 11 August 2012 06:23 |
Pavel Emelyanov
Messages: 8 Registered: October 2011
|
Junior Member |
|
|
On 08/11/2012 03:09 AM, H. Peter Anvin wrote:
> On 08/10/2012 12:28 PM, Alan Cox wrote:
>> Explicitly for Linux yes - this is not generally true of the AF_UNIX
>> socket domain and even the permissions aspect isn't guaranteed to be
>> supported on some BSD environments !
>
> Yes, but let's worry about what the Linux behavior should be.
>
>> The name is however just a proxy for the socket itself. You don't even
>> get a device node in the usual sense or the same inode in the file system
>> space.
>
>
> No, but it is looked up the same way any other inode is (the difference
> between FIFOs and sockets is that sockets have separate connections,
> which is also why open() on sockets would be nice.)
>
> However, there is a fundamental difference between AF_UNIX sockets and
> open(), and that is how the pathname is delivered. It thus would make
> more sense to provide the openat()-like information in struct
> sockaddr_un, but that may be very hard to do in a sensible way. In that
> sense it perhaps would be cleaner to be able to do an open[at]() on the
> socket node with O_PATH (perhaps there should be an O_SOCKET option,
> even?) and pass the resulting file descriptor to bind() or connect().
I vote for this (openat + O_WHATEVER on a unix socket) as well. It will
help us in checkpoint-restore, making handling of overmounted/unlinked
sockets much cleaner.
> -hpa
Thanks,
Pavel
|
|
|
Re: [RFC PATCH 0/2] net: connect to UNIX sockets from specified root [message #47454 is a reply to message #47453] |
Sat, 11 August 2012 11:15 |
Stanislav Kinsbursky
Messages: 683 Registered: October 2011
|
Senior Member |
|
|
11.08.2012 10:23, Pavel Emelyanov пишет:
> On 08/11/2012 03:09 AM, H. Peter Anvin wrote:
>> On 08/10/2012 12:28 PM, Alan Cox wrote:
>>> Explicitly for Linux yes - this is not generally true of the AF_UNIX
>>> socket domain and even the permissions aspect isn't guaranteed to be
>>> supported on some BSD environments !
>> Yes, but let's worry about what the Linux behavior should be.
>>
>>> The name is however just a proxy for the socket itself. You don't even
>>> get a device node in the usual sense or the same inode in the file system
>>> space.
>>
>> No, but it is looked up the same way any other inode is (the difference
>> between FIFOs and sockets is that sockets have separate connections,
>> which is also why open() on sockets would be nice.)
>>
>> However, there is a fundamental difference between AF_UNIX sockets and
>> open(), and that is how the pathname is delivered. It thus would make
>> more sense to provide the openat()-like information in struct
>> sockaddr_un, but that may be very hard to do in a sensible way. In that
>> sense it perhaps would be cleaner to be able to do an open[at]() on the
>> socket node with O_PATH (perhaps there should be an O_SOCKET option,
>> even?) and pass the resulting file descriptor to bind() or connect().
> I vote for this (openat + O_WHATEVER on a unix socket) as well. It will
> help us in checkpoint-restore, making handling of overmounted/unlinked
> sockets much cleaner.
I have to notice, that it's not enough and doesn't solve the issue.
There should be some way how to connect/bind already existent unix
socket (from kernel, at least), because socket can be created in user space.
And this way (sock operation or whatever) have to provide an ability to
lookup UNIX socket starting from specified root to support containers.
>
>> -hpa
> Thanks,
> Pavel
>
|
|
|
Re: [RFC PATCH 0/2] net: connect to UNIX sockets from specified root [message #47490 is a reply to message #47454] |
Mon, 13 August 2012 16:47 |
bfields
Messages: 107 Registered: September 2007
|
Senior Member |
|
|
On Sat, Aug 11, 2012 at 03:15:24PM +0400, Stanislav Kinsbursky wrote:
> 11.08.2012 10:23, Pavel Emelyanov пишет:
> >On 08/11/2012 03:09 AM, H. Peter Anvin wrote:
> >>On 08/10/2012 12:28 PM, Alan Cox wrote:
> >>>Explicitly for Linux yes - this is not generally true of the AF_UNIX
> >>>socket domain and even the permissions aspect isn't guaranteed to be
> >>>supported on some BSD environments !
> >>Yes, but let's worry about what the Linux behavior should be.
> >>
> >>>The name is however just a proxy for the socket itself. You don't even
> >>>get a device node in the usual sense or the same inode in the file system
> >>>space.
> >>
> >>No, but it is looked up the same way any other inode is (the difference
> >>between FIFOs and sockets is that sockets have separate connections,
> >>which is also why open() on sockets would be nice.)
> >>
> >>However, there is a fundamental difference between AF_UNIX sockets and
> >>open(), and that is how the pathname is delivered. It thus would make
> >>more sense to provide the openat()-like information in struct
> >>sockaddr_un, but that may be very hard to do in a sensible way. In that
> >>sense it perhaps would be cleaner to be able to do an open[at]() on the
> >>socket node with O_PATH (perhaps there should be an O_SOCKET option,
> >>even?) and pass the resulting file descriptor to bind() or connect().
> >I vote for this (openat + O_WHATEVER on a unix socket) as well. It will
> >help us in checkpoint-restore, making handling of overmounted/unlinked
> >sockets much cleaner.
>
> I have to notice, that it's not enough and doesn't solve the issue.
> There should be some way how to connect/bind already existent unix
> socket (from kernel, at least), because socket can be created in
> user space.
> And this way (sock operation or whatever) have to provide an ability
> to lookup UNIX socket starting from specified root to support
> containers.
I don't understand--the rpcbind sockets are created by the kernel. What
am I missing?
--b.
|
|
|
Re: [RFC PATCH 0/2] net: connect to UNIX sockets from specified root [message #47491 is a reply to message #47490] |
Mon, 13 August 2012 17:39 |
Stanislav Kinsbursky
Messages: 683 Registered: October 2011
|
Senior Member |
|
|
13.08.2012 20:47, J. Bruce Fields пишет:
> On Sat, Aug 11, 2012 at 03:15:24PM +0400, Stanislav Kinsbursky wrote:
>> 11.08.2012 10:23, Pavel Emelyanov пишет:
>>> On 08/11/2012 03:09 AM, H. Peter Anvin wrote:
>>>> On 08/10/2012 12:28 PM, Alan Cox wrote:
>>>>> Explicitly for Linux yes - this is not generally true of the AF_UNIX
>>>>> socket domain and even the permissions aspect isn't guaranteed to be
>>>>> supported on some BSD environments !
>>>> Yes, but let's worry about what the Linux behavior should be.
>>>>
>>>>> The name is however just a proxy for the socket itself. You don't even
>>>>> get a device node in the usual sense or the same inode in the file system
>>>>> space.
>>>> No, but it is looked up the same way any other inode is (the difference
>>>> between FIFOs and sockets is that sockets have separate connections,
>>>> which is also why open() on sockets would be nice.)
>>>>
>>>> However, there is a fundamental difference between AF_UNIX sockets and
>>>> open(), and that is how the pathname is delivered. It thus would make
>>>> more sense to provide the openat()-like information in struct
>>>> sockaddr_un, but that may be very hard to do in a sensible way. In that
>>>> sense it perhaps would be cleaner to be able to do an open[at]() on the
>>>> socket node with O_PATH (perhaps there should be an O_SOCKET option,
>>>> even?) and pass the resulting file descriptor to bind() or connect().
>>> I vote for this (openat + O_WHATEVER on a unix socket) as well. It will
>>> help us in checkpoint-restore, making handling of overmounted/unlinked
>>> sockets much cleaner.
>> I have to notice, that it's not enough and doesn't solve the issue.
>> There should be some way how to connect/bind already existent unix
>> socket (from kernel, at least), because socket can be created in
>> user space.
>> And this way (sock operation or whatever) have to provide an ability
>> to lookup UNIX socket starting from specified root to support
>> containers.
> I don't understand--the rpcbind sockets are created by the kernel. What
> am I missing?
Kernel preform connect to rpcbind socket (i.e. user-space binds it),
doesn't it?
>
> --b.
|
|
|
Re: [RFC PATCH 0/2] net: connect to UNIX sockets from specified root [message #47492 is a reply to message #47491] |
Mon, 13 August 2012 18:24 |
bfields
Messages: 107 Registered: September 2007
|
Senior Member |
|
|
On Mon, Aug 13, 2012 at 09:39:53PM +0400, Stanislav Kinsbursky wrote:
> 13.08.2012 20:47, J. Bruce Fields пишет:
> >On Sat, Aug 11, 2012 at 03:15:24PM +0400, Stanislav Kinsbursky wrote:
> >>11.08.2012 10:23, Pavel Emelyanov пишет:
> >>>On 08/11/2012 03:09 AM, H. Peter Anvin wrote:
> >>>>On 08/10/2012 12:28 PM, Alan Cox wrote:
> >>>>>Explicitly for Linux yes - this is not generally true of the AF_UNIX
> >>>>>socket domain and even the permissions aspect isn't guaranteed to be
> >>>>>supported on some BSD environments !
> >>>>Yes, but let's worry about what the Linux behavior should be.
> >>>>
> >>>>>The name is however just a proxy for the socket itself. You don't even
> >>>>>get a device node in the usual sense or the same inode in the file system
> >>>>>space.
> >>>>No, but it is looked up the same way any other inode is (the difference
> >>>>between FIFOs and sockets is that sockets have separate connections,
> >>>>which is also why open() on sockets would be nice.)
> >>>>
> >>>>However, there is a fundamental difference between AF_UNIX sockets and
> >>>>open(), and that is how the pathname is delivered. It thus would make
> >>>>more sense to provide the openat()-like information in struct
> >>>>sockaddr_un, but that may be very hard to do in a sensible way. In that
> >>>>sense it perhaps would be cleaner to be able to do an open[at]() on the
> >>>>socket node with O_PATH (perhaps there should be an O_SOCKET option,
> >>>>even?) and pass the resulting file descriptor to bind() or connect().
> >>>I vote for this (openat + O_WHATEVER on a unix socket) as well. It will
> >>>help us in checkpoint-restore, making handling of overmounted/unlinked
> >>>sockets much cleaner.
> >>I have to notice, that it's not enough and doesn't solve the issue.
> >>There should be some way how to connect/bind already existent unix
> >>socket (from kernel, at least), because socket can be created in
> >>user space.
> >>And this way (sock operation or whatever) have to provide an ability
> >>to lookup UNIX socket starting from specified root to support
> >>containers.
> >I don't understand--the rpcbind sockets are created by the kernel. What
> >am I missing?
>
> Kernel preform connect to rpcbind socket (i.e. user-space binds it),
> doesn't it?
I'm confused, possibly because there are three "sockets" here: the
client-side socket that's connected, the server-side socket that's
bound, and the common object that exists in the filesystem namespace.
Userland creates the server-side socket and binds to it. All of that is
done in the context of the rpcbind process, so is created in rpcbind's
namespace. That should be OK, right?
The client side socket is created and connected in
xs_local_setup_socket().
Making sure they both end up with the same thing is a matter of making
sure they lookup the same path in the same namespace. The difficult
part of that is the in-kernel client-side socket connect, where we don't
have the right process context any more.
We currently set that up with __sock_create followed by
kernel_connect.
The proposal seems to be to instead do an openat followed by a
kernel_connect, and pass the path in the openat instead of the connect.
(Though in the kernel we won't be able to call openat, so we'll end up
doing something like nfsd does (calling lookup_one_len() and
dentry_open() by hand).)
Have I got all that right?
I don't know if that's better just calling into the unix socket code at
connect time as your patch does. Maybe the answer depends on whether
it's a priority to make this functionality available to userspace.
--b.
|
|
|
Re: [RFC PATCH 0/2] net: connect to UNIX sockets from specified root [message #47495 is a reply to message #47492] |
Tue, 14 August 2012 08:46 |
Stanislav Kinsbursky
Messages: 683 Registered: October 2011
|
Senior Member |
|
|
13.08.2012 22:24, J. Bruce Fields пишет:
> On Mon, Aug 13, 2012 at 09:39:53PM +0400, Stanislav Kinsbursky wrote:
>> 13.08.2012 20:47, J. Bruce Fields пишет:
>>> On Sat, Aug 11, 2012 at 03:15:24PM +0400, Stanislav Kinsbursky wrote:
>>>> 11.08.2012 10:23, Pavel Emelyanov пишет:
>>>>> On 08/11/2012 03:09 AM, H. Peter Anvin wrote:
>>>>>> On 08/10/2012 12:28 PM, Alan Cox wrote:
>>>>>>> Explicitly for Linux yes - this is not generally true of the
>>>>>>> AF_UNIX socket domain and even the permissions aspect isn't
>>>>>>> guaranteed to be supported on some BSD environments !
>>>>>> Yes, but let's worry about what the Linux behavior should be.
>>>>>>
>>>>>>> The name is however just a proxy for the socket itself. You
>>>>>>> don't even get a device node in the usual sense or the same inode
>>>>>>> in the file system space.
>>>>>> No, but it is looked up the same way any other inode is (the
>>>>>> difference between FIFOs and sockets is that sockets have separate
>>>>>> connections, which is also why open() on sockets would be nice.)
>>>>>>
>>>>>> However, there is a fundamental difference between AF_UNIX sockets
>>>>>> and open(), and that is how the pathname is delivered. It thus
>>>>>> would make more sense to provide the openat()-like information in
>>>>>> struct sockaddr_un, but that may be very hard to do in a sensible
>>>>>> way. In that sense it perhaps would be cleaner to be able to do
>>>>>> an open[at]() on the socket node with O_PATH (perhaps there should
>>>>>> be an O_SOCKET option, even?) and pass the resulting file
>>>>>> descriptor to bind() or connect().
>>>>> I vote for this (openat + O_WHATEVER on a unix socket) as well. It
>>>>> will help us in checkpoint-restore, making handling of
>>>>> overmounted/unlinked sockets much cleaner.
>>>> I have to notice, that it's not enough and doesn't solve the issue.
>>>> There should be some way how to connect/bind already existent unix
>>>> socket (from kernel, at least), because socket can be created in user
>>>> space. And this way (sock operation or whatever) have to provide an
>>>> ability to lookup UNIX socket starting from specified root to support
>>>> containers.
>>> I don't understand--the rpcbind sockets are created by the kernel. What
>>> am I missing?
>>
>> Kernel preform connect to rpcbind socket (i.e. user-space binds it),
>> doesn't it?
>
> I'm confused, possibly because there are three "sockets" here: the
> client-side socket that's connected, the server-side socket that's bound,
> and the common object that exists in the filesystem namespace.
>
> Userland creates the server-side socket and binds to it. All of that is
> done in the context of the rpcbind process, so is created in rpcbind's
> namespace. That should be OK, right?
>
> The client side socket is created and connected in xs_local_setup_socket().
>
> Making sure they both end up with the same thing is a matter of making sure
> they lookup the same path in the same namespace. The difficult part of that
> is the in-kernel client-side socket connect, where we don't have the right
> process context any more.
>
Looks like I'm missing something important.
Where are these UNIX in-kernel created and listening sockets (in code, I mean)?
--
Best regards,
Stanislav Kinsbursky
|
|
|
Goto Forum:
Current Time: Fri Dec 13 01:29:43 GMT 2024
Total time taken to generate the page: 0.02546 seconds
|