From: Steve French <smfrench@gmail.com>
To: Paulo Alcantara <pc@manguebit.com>
Cc: linux-cifs@vger.kernel.org
Subject: Re: [PATCH v2 2/2] cifs: prevent data race in cifs_reconnect_tcon()
Date: Tue, 28 Feb 2023 16:16:11 -0600 [thread overview]
Message-ID: <CAH2r5mso-anitA0n85+MgAX9G52SVF0NjJ+PLw7Y+1Q+OjOx0Q@mail.gmail.com> (raw)
In-Reply-To: <20230228220155.23394-2-pc@manguebit.com>
updated cifs-2.6.git for-next with these two patches
On Tue, Feb 28, 2023 at 4:02 PM Paulo Alcantara <pc@manguebit.com> wrote:
>
> Make sure to get an up-to-date TCP_Server_Info::nr_targets value prior
> to waiting the server to be reconnected in cifs_reconnect_tcon(). It
> is set in cifs_tcp_ses_needs_reconnect() and protected by
> TCP_Server_Info::srv_lock.
>
> Create a new cifs_wait_for_server_reconnect() helper that can be used
> by both SMB2+ and CIFS reconnect code.
>
> Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
> ---
> v1 -> v2: fixed double unlock pointed out by Steve
>
> fs/cifs/cifsproto.h | 1 +
> fs/cifs/cifssmb.c | 43 ++----------------------
> fs/cifs/misc.c | 44 ++++++++++++++++++++++++
> fs/cifs/smb2pdu.c | 82 ++++++++++++---------------------------------
> 4 files changed, 69 insertions(+), 101 deletions(-)
>
> diff --git a/fs/cifs/cifsproto.h b/fs/cifs/cifsproto.h
> index 20a2f0f3f682..e2eff66eefab 100644
> --- a/fs/cifs/cifsproto.h
> +++ b/fs/cifs/cifsproto.h
> @@ -694,5 +694,6 @@ static inline int cifs_create_options(struct cifs_sb_info *cifs_sb, int options)
>
> struct super_block *cifs_get_tcon_super(struct cifs_tcon *tcon);
> void cifs_put_tcon_super(struct super_block *sb);
> +int cifs_wait_for_server_reconnect(struct TCP_Server_Info *server, bool retry);
>
> #endif /* _CIFSPROTO_H */
> diff --git a/fs/cifs/cifssmb.c b/fs/cifs/cifssmb.c
> index a24e4ddf8043..a43c78396dd8 100644
> --- a/fs/cifs/cifssmb.c
> +++ b/fs/cifs/cifssmb.c
> @@ -72,7 +72,6 @@ cifs_reconnect_tcon(struct cifs_tcon *tcon, int smb_command)
> struct cifs_ses *ses;
> struct TCP_Server_Info *server;
> struct nls_table *nls_codepage;
> - int retries;
>
> /*
> * SMBs NegProt, SessSetup, uLogoff do not have tcon yet so check for
> @@ -102,45 +101,9 @@ cifs_reconnect_tcon(struct cifs_tcon *tcon, int smb_command)
> }
> spin_unlock(&tcon->tc_lock);
>
> - retries = server->nr_targets;
> -
> - /*
> - * Give demultiplex thread up to 10 seconds to each target available for
> - * reconnect -- should be greater than cifs socket timeout which is 7
> - * seconds.
> - */
> - while (server->tcpStatus == CifsNeedReconnect) {
> - rc = wait_event_interruptible_timeout(server->response_q,
> - (server->tcpStatus != CifsNeedReconnect),
> - 10 * HZ);
> - if (rc < 0) {
> - cifs_dbg(FYI, "%s: aborting reconnect due to a received signal by the process\n",
> - __func__);
> - return -ERESTARTSYS;
> - }
> -
> - /* are we still trying to reconnect? */
> - spin_lock(&server->srv_lock);
> - if (server->tcpStatus != CifsNeedReconnect) {
> - spin_unlock(&server->srv_lock);
> - break;
> - }
> - spin_unlock(&server->srv_lock);
> -
> - if (retries && --retries)
> - continue;
> -
> - /*
> - * on "soft" mounts we wait once. Hard mounts keep
> - * retrying until process is killed or server comes
> - * back on-line
> - */
> - if (!tcon->retry) {
> - cifs_dbg(FYI, "gave up waiting on reconnect in smb_init\n");
> - return -EHOSTDOWN;
> - }
> - retries = server->nr_targets;
> - }
> + rc = cifs_wait_for_server_reconnect(server, tcon->retry);
> + if (rc)
> + return rc;
>
> spin_lock(&ses->chan_lock);
> if (!cifs_chan_needs_reconnect(ses, server) && !tcon->need_reconnect) {
> diff --git a/fs/cifs/misc.c b/fs/cifs/misc.c
> index 0c6c1fc8dae9..a0d286ee723d 100644
> --- a/fs/cifs/misc.c
> +++ b/fs/cifs/misc.c
> @@ -1266,3 +1266,47 @@ int cifs_inval_name_dfs_link_error(const unsigned int xid,
> return 0;
> }
> #endif
> +
> +int cifs_wait_for_server_reconnect(struct TCP_Server_Info *server, bool retry)
> +{
> + int timeout = 10;
> + int rc;
> +
> + spin_lock(&server->srv_lock);
> + if (server->tcpStatus != CifsNeedReconnect) {
> + spin_unlock(&server->srv_lock);
> + return 0;
> + }
> + timeout *= server->nr_targets;
> + spin_unlock(&server->srv_lock);
> +
> + /*
> + * Give demultiplex thread up to 10 seconds to each target available for
> + * reconnect -- should be greater than cifs socket timeout which is 7
> + * seconds.
> + *
> + * On "soft" mounts we wait once. Hard mounts keep retrying until
> + * process is killed or server comes back on-line.
> + */
> + do {
> + rc = wait_event_interruptible_timeout(server->response_q,
> + (server->tcpStatus != CifsNeedReconnect),
> + timeout * HZ);
> + if (rc < 0) {
> + cifs_dbg(FYI, "%s: aborting reconnect due to received signal\n",
> + __func__);
> + return -ERESTARTSYS;
> + }
> +
> + /* are we still trying to reconnect? */
> + spin_lock(&server->srv_lock);
> + if (server->tcpStatus != CifsNeedReconnect) {
> + spin_unlock(&server->srv_lock);
> + return 0;
> + }
> + spin_unlock(&server->srv_lock);
> + } while (retry);
> +
> + cifs_dbg(FYI, "%s: gave up waiting on reconnect\n", __func__);
> + return -EHOSTDOWN;
> +}
> diff --git a/fs/cifs/smb2pdu.c b/fs/cifs/smb2pdu.c
> index ca9d7110ddcb..0e53265e1462 100644
> --- a/fs/cifs/smb2pdu.c
> +++ b/fs/cifs/smb2pdu.c
> @@ -139,66 +139,6 @@ smb2_hdr_assemble(struct smb2_hdr *shdr, __le16 smb2_cmd,
> return;
> }
>
> -static int wait_for_server_reconnect(struct TCP_Server_Info *server,
> - __le16 smb2_command, bool retry)
> -{
> - int timeout = 10;
> - int rc;
> -
> - spin_lock(&server->srv_lock);
> - if (server->tcpStatus != CifsNeedReconnect) {
> - spin_unlock(&server->srv_lock);
> - return 0;
> - }
> - timeout *= server->nr_targets;
> - spin_unlock(&server->srv_lock);
> -
> - /*
> - * Return to caller for TREE_DISCONNECT and LOGOFF and CLOSE
> - * here since they are implicitly done when session drops.
> - */
> - switch (smb2_command) {
> - /*
> - * BB Should we keep oplock break and add flush to exceptions?
> - */
> - case SMB2_TREE_DISCONNECT:
> - case SMB2_CANCEL:
> - case SMB2_CLOSE:
> - case SMB2_OPLOCK_BREAK:
> - return -EAGAIN;
> - }
> -
> - /*
> - * Give demultiplex thread up to 10 seconds to each target available for
> - * reconnect -- should be greater than cifs socket timeout which is 7
> - * seconds.
> - *
> - * On "soft" mounts we wait once. Hard mounts keep retrying until
> - * process is killed or server comes back on-line.
> - */
> - do {
> - rc = wait_event_interruptible_timeout(server->response_q,
> - (server->tcpStatus != CifsNeedReconnect),
> - timeout * HZ);
> - if (rc < 0) {
> - cifs_dbg(FYI, "%s: aborting reconnect due to received signal\n",
> - __func__);
> - return -ERESTARTSYS;
> - }
> -
> - /* are we still trying to reconnect? */
> - spin_lock(&server->srv_lock);
> - if (server->tcpStatus != CifsNeedReconnect) {
> - spin_unlock(&server->srv_lock);
> - return 0;
> - }
> - spin_unlock(&server->srv_lock);
> - } while (retry);
> -
> - cifs_dbg(FYI, "%s: gave up waiting on reconnect\n", __func__);
> - return -EHOSTDOWN;
> -}
> -
> static int
> smb2_reconnect(__le16 smb2_command, struct cifs_tcon *tcon,
> struct TCP_Server_Info *server)
> @@ -243,7 +183,27 @@ smb2_reconnect(__le16 smb2_command, struct cifs_tcon *tcon,
> (!tcon->ses->server) || !server)
> return -EIO;
>
> - rc = wait_for_server_reconnect(server, smb2_command, tcon->retry);
> + spin_lock(&server->srv_lock);
> + if (server->tcpStatus == CifsNeedReconnect) {
> + /*
> + * Return to caller for TREE_DISCONNECT and LOGOFF and CLOSE
> + * here since they are implicitly done when session drops.
> + */
> + switch (smb2_command) {
> + /*
> + * BB Should we keep oplock break and add flush to exceptions?
> + */
> + case SMB2_TREE_DISCONNECT:
> + case SMB2_CANCEL:
> + case SMB2_CLOSE:
> + case SMB2_OPLOCK_BREAK:
> + spin_unlock(&server->srv_lock);
> + return -EAGAIN;
> + }
> + }
> + spin_unlock(&server->srv_lock);
> +
> + rc = cifs_wait_for_server_reconnect(server, tcon->retry);
> if (rc)
> return rc;
>
> --
> 2.39.2
>
--
Thanks,
Steve
next prev parent reply other threads:[~2023-02-28 22:16 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-27 13:53 [PATCH 1/2] cifs: improve checking of DFS links over STATUS_OBJECT_NAME_INVALID Paulo Alcantara
2023-02-27 13:53 ` [PATCH 2/2] cifs: prevent data race in cifs_reconnect_tcon() Paulo Alcantara
2023-02-27 22:40 ` Steve French
2023-02-28 0:09 ` Paulo Alcantara
2023-02-28 22:01 ` [PATCH v2 1/2] cifs: improve checking of DFS links over STATUS_OBJECT_NAME_INVALID Paulo Alcantara
2023-02-28 22:01 ` [PATCH v2 2/2] cifs: prevent data race in cifs_reconnect_tcon() Paulo Alcantara
2023-02-28 22:16 ` Steve French [this message]
2023-02-28 22:20 ` [PATCH v2 1/2] cifs: improve checking of DFS links over STATUS_OBJECT_NAME_INVALID ronnie sahlberg
2023-02-28 23:35 ` Steve French
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=CAH2r5mso-anitA0n85+MgAX9G52SVF0NjJ+PLw7Y+1Q+OjOx0Q@mail.gmail.com \
--to=smfrench@gmail.com \
--cc=linux-cifs@vger.kernel.org \
--cc=pc@manguebit.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).