From: Tom Tucker <tom@opengridcomputing.com>
To: Chuck Lever <chuck.lever@oracle.com>
Cc: bfields@fieldses.org, aime.le-rouzic@bull.net, linux-nfs@vger.kernel.org
Subject: Re: [PATCH 08/22] SUNRPC: pass buffer size to svc_addsock() and svc_sock_names()
Date: Mon, 15 Dec 2008 15:05:42 -0600 [thread overview]
Message-ID: <4946C6A6.4050602@opengridcomputing.com> (raw)
In-Reply-To: <7998F1F2-05DD-44AB-B023-80C49DABAD9B@oracle.com>
Chuck Lever wrote:
> On Dec 14, 2008, at 12:33 PM, Tom Tucker wrote:
>> Hi Chuck:
>>
>> This whole svc_addsock business is an artifact of rpc.xxx creating a
>> socket in user mode and then passing the fd down to the kernel via a
>> write to the portlist file. Would it be easier, cleaner, etc... to
>> normalize this whole business to use strings exclusively to create
>> these sockets?
>>
>> E.g. echo "tcp6 2049" > /proc/fs/nfsd/portlist
>>
>> You could quite simply extend the syntax as well to support listening
>> on specific interfaces.
>
> That would be Bruce's decision. It doesn't seem like a bad idea.
>
> Some folks question the idea of using netids like "tcp6" for kernel
> interfaces. I'm on the fence. RDMA doesn't have an official netid yet,
> for example.
Ok clearly there are (at least) two issues. One is mechanical; how do we
get the NFS server side transport instantiated. The second is: what do
the "identifiers" represent?
I'm advocating that the mechanism be normalized so that we can get rid
of code that duplicates functionality.
Perhaps this is naive, but could we simply call it an "NFS Transport
Identifier" such that the only thing it "means" is the election of the
transport provider invoked to instantiate the service?
>
>> Tom
>>
>> Chuck Lever wrote:
>>> Pass the size of the output buffer to the RPC functions that construct
>>> the list of socket names in that buffer. Add documenting comments to
>>> these functions.
>>>
>>> This is a cosmetic change for now. A subsequent patch will make sure
>>> the buffer length is passed to one_sock_name(), where the length will
>>> actually be useful.
>>>
>>> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
>>> ---
>>>
>>> fs/nfsd/nfsctl.c | 12 ++++++++----
>>> include/linux/sunrpc/svcsock.h | 6 ++++--
>>> net/sunrpc/svcsock.c | 34 +++++++++++++++++++++++++++++-----
>>> 3 files changed, 41 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
>>> index 22fc8e5..19db9f4 100644
>>> --- a/fs/nfsd/nfsctl.c
>>> +++ b/fs/nfsd/nfsctl.c
>>> @@ -898,7 +898,7 @@ static ssize_t __write_ports_names(char *buf,
>>> size_t size)
>>> static ssize_t __write_ports_addfd(char *buf, size_t size)
>>> {
>>> char *mesg = buf;
>>> - int fd, err;
>>> + int fd, err, len;
>>> err = get_int(&mesg, &fd);
>>> if (err || fd < 0)
>>> @@ -908,13 +908,16 @@ static ssize_t __write_ports_addfd(char *buf,
>>> size_t size)
>>> if (err)
>>> return err;
>>> - err = svc_addsock(nfsd_serv, fd, buf);
>>> + len = SIMPLE_TRANSACTION_LIMIT;
>>> + err = svc_addsock(nfsd_serv, fd, buf, len);
>>> if (err < 0)
>>> return err;
>>> + len -= err;
>>> err = lockd_up();
>>> if (err < 0)
>>> - svc_sock_names(buf + strlen(buf) + 1, nfsd_serv, buf);
>>> + svc_sock_names(nfsd_serv, buf + strlen(buf) + 1,
>>> + len - strlen(buf) - 1, buf);
>>> /* Decrease the count, but don't shut down the the service */
>>> nfsd_serv->sv_nrthreads--;
>>> @@ -935,7 +938,8 @@ static ssize_t __write_ports_delfd(char *buf,
>>> size_t size)
>>> return -ENOMEM;
>>> if (nfsd_serv)
>>> - len = svc_sock_names(buf, nfsd_serv, toclose);
>>> + len = svc_sock_names(nfsd_serv, buf,
>>> + SIMPLE_TRANSACTION_LIMIT, toclose);
>>> if (len >= 0)
>>> lockd_down();
>>> diff --git a/include/linux/sunrpc/svcsock.h
>>> b/include/linux/sunrpc/svcsock.h
>>> index 483e103..f57ce85 100644
>>> --- a/include/linux/sunrpc/svcsock.h
>>> +++ b/include/linux/sunrpc/svcsock.h
>>> @@ -38,8 +38,10 @@ int svc_recv(struct svc_rqst *, long);
>>> int svc_send(struct svc_rqst *);
>>> void svc_drop(struct svc_rqst *);
>>> void svc_sock_update_bufs(struct svc_serv *serv);
>>> -int svc_sock_names(char *buf, struct svc_serv *serv, char
>>> *toclose);
>>> -int svc_addsock(struct svc_serv *serv, int fd, char
>>> *name_return);
>>> +int svc_sock_names(struct svc_serv *serv, char *buf, size_t
>>> buflen,
>>> + char *toclose);
>>> +int svc_addsock(struct svc_serv *serv, int fd, char
>>> *name_return,
>>> + size_t len);
>>> void svc_init_xprt_sock(void);
>>> void svc_cleanup_xprt_sock(void);
>>> diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
>>> index a1951dc..39f5015 100644
>>> --- a/net/sunrpc/svcsock.c
>>> +++ b/net/sunrpc/svcsock.c
>>> @@ -263,8 +263,23 @@ static int one_sock_name(char *buf, struct
>>> svc_sock *svsk)
>>> return len;
>>> }
>>> -int
>>> -svc_sock_names(char *buf, struct svc_serv *serv, char *toclose)
>>> +/**
>>> + * svc_sock_names - construct a list of listener names in a string
>>> + * @serv: pointer to RPC service
>>> + * @buf: pointer to a buffer to fill in with socket names
>>> + * @buflen: size of the buffer to be filled
>>> + * @toclose: pointer to '\0'-terminated C string containing the name
>>> + * of a listener to be closed
>>> + *
>>> + * Fills in @buf with a '\n'-separated list of names of listener
>>> + * sockets. If @toclose is not NULL, the socket named by @toclose
>>> + * is closed, and is not included in the output list.
>>> + *
>>> + * Returns positive length of the socket name string, or a negative
>>> + * errno value on error.
>>> + */
>>> +int svc_sock_names(struct svc_serv *serv, char *buf, size_t buflen,
>>> + char *toclose)
>>> {
>>> struct svc_sock *svsk, *closesk = NULL;
>>> int len = 0;
>>> @@ -1165,9 +1180,18 @@ static struct svc_sock
>>> *svc_setup_socket(struct svc_serv *serv,
>>> return svsk;
>>> }
>>> -int svc_addsock(struct svc_serv *serv,
>>> - int fd,
>>> - char *name_return)
>>> +/**
>>> + * svc_addsock - add a listener socket to an RPC service
>>> + * @serv: pointer to RPC service to which to add a new listener
>>> + * @fd: file descriptor of the new listener
>>> + * @name_return: pointer to buffer to pass back name of listener
>>> + * @len: size of the buffer
>>> + *
>>> + * Fills in socket name and returns positive length of name if
>>> successful.
>>> + * Name is terminated with '\n'. On error, returns a negative errno
>>> + * value.
>>> + */
>>> +int svc_addsock(struct svc_serv *serv, int fd, char *name_return,
>>> size_t len)
>>> {
>>> int err = 0;
>>> struct socket *so = sockfd_lookup(fd, &err);
>>>
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
>>> the body of a message to majordomo@vger.kernel.org
>>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>>>
>>
>
> --
> Chuck Lever
> chuck[dot]lever[at]oracle[dot]com
>
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2008-12-15 21:05 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-12-12 21:57 [PATCH 00/22] IPv6 support NFSD Chuck Lever
[not found] ` <20081212215340.24332.88416.stgit-07a7zB5ZJzbwdl/1UfZZQIVfYA8g3rJ/@public.gmane.org>
2008-12-12 21:57 ` [PATCH 01/22] NFSD: clean up failover sysctl function naming Chuck Lever
2008-12-12 21:57 ` [PATCH 02/22] NFSD: Fix a handful of coding style issues in write_filehandle() Chuck Lever
2008-12-12 21:57 ` [PATCH 03/22] NFSD: Replace open-coded integer with macro Chuck Lever
2008-12-12 21:57 ` [PATCH 04/22] NFSD: Add documenting comments for nfsctl interface Chuck Lever
2008-12-12 21:57 ` [PATCH 05/22] NFSD: Add helper functions for __write_ports() Chuck Lever
[not found] ` <20081212215742.24332.36578.stgit-07a7zB5ZJzbwdl/1UfZZQIVfYA8g3rJ/@public.gmane.org>
2008-12-25 4:15 ` J. Bruce Fields
2008-12-29 17:04 ` Chuck Lever
2008-12-29 18:42 ` J. Bruce Fields
2008-12-12 21:57 ` [PATCH 06/22] NFSD: Refactor __write_ports() Chuck Lever
2008-12-12 21:57 ` [PATCH 07/22] NFSD: Prevent a buffer overflow in svc_xprt_names() Chuck Lever
[not found] ` <20081212215757.24332.77904.stgit-07a7zB5ZJzbwdl/1UfZZQIVfYA8g3rJ/@public.gmane.org>
2008-12-25 4:32 ` J. Bruce Fields
2008-12-12 21:58 ` [PATCH 08/22] SUNRPC: pass buffer size to svc_addsock() and svc_sock_names() Chuck Lever
[not found] ` <20081212215804.24332.24605.stgit-07a7zB5ZJzbwdl/1UfZZQIVfYA8g3rJ/@public.gmane.org>
2008-12-14 17:33 ` Tom Tucker
2008-12-15 16:40 ` Chuck Lever
2008-12-15 21:05 ` Tom Tucker [this message]
2008-12-15 21:17 ` J. Bruce Fields
2008-12-25 4:43 ` J. Bruce Fields
2008-12-29 19:24 ` Chuck Lever
2008-12-30 19:38 ` J. Bruce Fields
2008-12-12 21:58 ` [PATCH 09/22] SUNRPC: Switch one_sock_name() to use snprintf() Chuck Lever
2008-12-12 21:58 ` [PATCH 10/22] SUNRPC: Support AF_INET6 in one_sock_name() Chuck Lever
2008-12-12 21:58 ` [PATCH 11/22] SUNRPC: Clean up one_sock_name() Chuck Lever
2008-12-12 21:58 ` [PATCH 12/22] NFSD: Support AF_INET6 in svc_addsock() function Chuck Lever
2008-12-12 21:58 ` [PATCH 13/22] NFS: Move NFS client's IP address parser to nfs_common/ Chuck Lever
[not found] ` <20081212215842.24332.47093.stgit-07a7zB5ZJzbwdl/1UfZZQIVfYA8g3rJ/@public.gmane.org>
2008-12-25 4:47 ` J. Bruce Fields
2008-12-12 21:58 ` [PATCH 14/22] NFSD: Support IPv6 addresses in write_failover_ip() Chuck Lever
2008-12-12 21:58 ` [PATCH 15/22] NFSD: Enable NFS server use of AF_INET6 Chuck Lever
2008-12-12 21:59 ` [PATCH 16/22] NFSD: Prevent buffer overflow in write_threads() Chuck Lever
2008-12-12 21:59 ` [PATCH 17/22] NFSD: Prevent buffer overflow in write_versions() Chuck Lever
2008-12-12 21:59 ` [PATCH 18/22] NFSD: Prevent buffer overflow in write_maxblksize() Chuck Lever
2008-12-12 21:59 ` [PATCH 19/22] NFSD: Prevent buffer overflow in write_leasetime() Chuck Lever
2008-12-12 21:59 ` [PATCH 20/22] NFSD: Prevent buffer overflow in write_recoverydir() Chuck Lever
2008-12-12 21:59 ` [PATCH 21/22] NLM: Refactor make_socks() function Chuck Lever
2008-12-12 21:59 ` [PATCH 22/22] NLM: Clean up flow of control in " Chuck Lever
2008-12-16 16:53 ` [PATCH 00/22] IPv6 support NFSD J. Bruce Fields
2008-12-25 5:01 ` J. Bruce Fields
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4946C6A6.4050602@opengridcomputing.com \
--to=tom@opengridcomputing.com \
--cc=aime.le-rouzic@bull.net \
--cc=bfields@fieldses.org \
--cc=chuck.lever@oracle.com \
--cc=linux-nfs@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.