Skip to contentSkip to conversation

cifs: don't declare smb_vol info on the stack

PATCH Steve French 2 participants 4 messages Nov 30, 2008 – Dec 2, 2008 2 messages not in the archive VFS

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#2 lore raw

I have thought about this a few times, but always hesitated because I was hoping that eventually cifs mount option parsing could use new parsing functions (which would also shrink the "parse_mount_options" routine which we call during cifs mount), and would cause us to rewrite this. IIRC nfs was changed a few years ago to use the new mount mechanism (match_table and tokens).

Quoted 225 lines · Jeff Layton

struct smb_vol is fairly large, it's probably best to kzalloc it...

---
 fs/cifs/connect.c |   91 ++++++++++++++++++++++++++++------------------------
 1 files changed, 49 insertions(+), 42 deletions(-)
fs/cifs/connect.c +49 -43
index 291ccd6..b08c6a3 100644
@@ -2186,27 +2186,30 @@ cifs_mount(struct super_block *sb, struct cifs_sb_info *cifs_sb,
 {
       int rc = 0;
       int xid;
-       struct smb_vol volume_info;
+       struct smb_vol *volume_info;
       struct cifsSesInfo *pSesInfo = NULL;
       struct cifsTconInfo *tcon = NULL;
       struct TCP_Server_Info *srvTcp = NULL;
 
       xid = GetXid();
 
-/* cFYI(1, ("Entering cifs_mount. Xid: %d with: %s", xid, mount_data)); */
+       volume_info = kzalloc(sizeof(struct smb_vol), GFP_KERNEL);
+       if (!volume_info) {
+               rc = -ENOMEM;
+               goto out;
+       }
 
-       memset(&volume_info, 0, sizeof(struct smb_vol));
-       if (cifs_parse_mount_options(mount_data, devname, &volume_info)) {
+       if (cifs_parse_mount_options(mount_data, devname, volume_info)) {
               rc = -EINVAL;
               goto out;
       }
 
-       if (volume_info.nullauth) {
+       if (volume_info->nullauth) {
               cFYI(1, ("null user"));
-               volume_info.username = "";
-       } else if (volume_info.username) {
+               volume_info->username = "";
+       } else if (volume_info->username) {
               /* BB fixme parse for domain name here */
-               cFYI(1, ("Username: %s", volume_info.username));
+               cFYI(1, ("Username: %s", volume_info->username));
       } else {
               cifserror("No username specified");
       /* In userspace mount helper we can get user name from alternate
@@ -2217,27 +2220,27 @@ cifs_mount(struct super_block *sb, struct cifs_sb_info *cifs_sb,
 
 
       /* this is needed for ASCII cp to Unicode converts */
-       if (volume_info.iocharset == NULL) {
+       if (volume_info->iocharset == NULL) {
               cifs_sb->local_nls = load_nls_default();
       /* load_nls_default can not return null */
       } else {
-               cifs_sb->local_nls = load_nls(volume_info.iocharset);
+               cifs_sb->local_nls = load_nls(volume_info->iocharset);
               if (cifs_sb->local_nls == NULL) {
                       cERROR(1, ("CIFS mount error: iocharset %s not found",
-                                volume_info.iocharset));
+                                volume_info->iocharset));
                       rc = -ELIBACC;
                       goto out;
               }
       }
 
       /* get a reference to a tcp session */
-       srvTcp = cifs_get_tcp_session(&volume_info);
+       srvTcp = cifs_get_tcp_session(volume_info);
       if (IS_ERR(srvTcp)) {
               rc = PTR_ERR(srvTcp);
               goto out;
       }
 
-       pSesInfo = cifs_find_smb_ses(srvTcp, volume_info.username);
+       pSesInfo = cifs_find_smb_ses(srvTcp, volume_info->username);
       if (pSesInfo) {
               cFYI(1, ("Existing smb sess found (status=%d)",
                       pSesInfo->status));
@@ -2275,24 +2278,24 @@ cifs_mount(struct super_block *sb, struct cifs_sb_info *cifs_sb,
               list_add(&pSesInfo->smb_ses_list, &srvTcp->smb_ses_list);
               write_unlock(&cifs_tcp_ses_lock);
 
-               /* volume_info.password freed at unmount */
-               if (volume_info.password) {
-                       pSesInfo->password = volume_info.password;
+               /* volume_info->password freed at unmount */
+               if (volume_info->password) {
+                       pSesInfo->password = volume_info->password;
                       /* set to NULL to prevent freeing on exit */
-                       volume_info.password = NULL;
+                       volume_info->password = NULL;
               }
-               if (volume_info.username)
-                       strncpy(pSesInfo->userName, volume_info.username,
+               if (volume_info->username)
+                       strncpy(pSesInfo->userName, volume_info->username,
                               MAX_USERNAME_SIZE);
-               if (volume_info.domainname) {
-                       int len = strlen(volume_info.domainname);
+               if (volume_info->domainname) {
+                       int len = strlen(volume_info->domainname);
                       pSesInfo->domainName = kmalloc(len + 1, GFP_KERNEL);
                       if (pSesInfo->domainName)
                               strcpy(pSesInfo->domainName,
-                                       volume_info.domainname);
+                                       volume_info->domainname);
               }
-               pSesInfo->linux_uid = volume_info.linux_uid;
-               pSesInfo->overrideSecFlg = volume_info.secFlg;
+               pSesInfo->linux_uid = volume_info->linux_uid;
+               pSesInfo->overrideSecFlg = volume_info->secFlg;
               down(&pSesInfo->sesSem);
 
               /* BB FIXME need to pass vol->secFlgs BB */
@@ -2303,14 +2306,14 @@ cifs_mount(struct super_block *sb, struct cifs_sb_info *cifs_sb,
 
       /* search for existing tcon to this server share */
       if (!rc) {
-               setup_cifs_sb(&volume_info, cifs_sb);
+               setup_cifs_sb(volume_info, cifs_sb);
 
-               tcon = cifs_find_tcon(pSesInfo, volume_info.UNC);
+               tcon = cifs_find_tcon(pSesInfo, volume_info->UNC);
               if (tcon) {
                       cFYI(1, ("Found match on UNC path"));
                       /* existing tcon already has a reference */
                       cifs_put_smb_ses(pSesInfo);
-                       if (tcon->seal != volume_info.seal)
+                       if (tcon->seal != volume_info->seal)
                               cERROR(1, ("transport encryption setting "
                                          "conflicts with existing tid"));
               } else {
@@ -2322,8 +2325,8 @@ cifs_mount(struct super_block *sb, struct cifs_sb_info *cifs_sb,
                       tcon->ses = pSesInfo;
 
                       /* check for null share name ie connect to dfs root */
-                       if ((strchr(volume_info.UNC + 3, '\\') == NULL)
-                           && (strchr(volume_info.UNC + 3, '/') == NULL)) {
+                       if ((strchr(volume_info->UNC + 3, '\\') == NULL)
+                           && (strchr(volume_info->UNC + 3, '/') == NULL)) {
                               /* rc = connect_to_dfs_path(...) */
                               cFYI(1, ("DFS root not supported"));
                               rc = -ENODEV;
@@ -2332,10 +2335,10 @@ cifs_mount(struct super_block *sb, struct cifs_sb_info *cifs_sb,
                               /* BB Do we need to wrap sesSem around
                                * this TCon call and Unix SetFS as
                                * we do on SessSetup and reconnect? */
-                               rc = CIFSTCon(xid, pSesInfo, volume_info.UNC,
+                               rc = CIFSTCon(xid, pSesInfo, volume_info->UNC,
                                             tcon, cifs_sb->local_nls);
                               cFYI(1, ("CIFS Tcon rc = %d", rc));
-                               if (volume_info.nodfs) {
+                               if (volume_info->nodfs) {
                                       tcon->Flags &= ~SMB_SHARE_IS_IN_DFS;
                                       cFYI(1, ("DFS disabled (%d)",
                                               tcon->Flags));
@@ -2343,7 +2346,7 @@ cifs_mount(struct super_block *sb, struct cifs_sb_info *cifs_sb,
                       }
                       if (rc)
                               goto mount_fail_check;
-                       tcon->seal = volume_info.seal;
+                       tcon->seal = volume_info->seal;
                       write_lock(&cifs_tcp_ses_lock);
                       list_add(&tcon->tcon_list, &pSesInfo->tcon_list);
                       write_unlock(&cifs_tcp_ses_lock);
@@ -2353,9 +2356,9 @@ cifs_mount(struct super_block *sb, struct cifs_sb_info *cifs_sb,
                  to a share so for resources mounted more than once
                  to the same server share the last value passed in
                  for the retry flag is used */
-               tcon->retry = volume_info.retry;
-               tcon->nocase = volume_info.nocase;
-               tcon->local_lease = volume_info.local_lease;
+               tcon->retry = volume_info->retry;
+               tcon->nocase = volume_info->nocase;
+               tcon->local_lease = volume_info->local_lease;
       }
       if (pSesInfo) {
               if (pSesInfo->capabilities & CAP_LARGE_FILES) {
@@ -2392,7 +2395,7 @@ mount_fail_check:
       if (tcon->ses->capabilities & CAP_UNIX)
               /* reset of caps checks mount to see if unix extensions
                  disabled for just this mount */
-               reset_cifs_unix_caps(xid, tcon, sb, &volume_info);
+               reset_cifs_unix_caps(xid, tcon, sb, volume_info);
       else
               tcon->unix_ext = 0; /* server does not support them */
 
@@ -2411,18 +2414,22 @@ mount_fail_check:
               cifs_sb->rsize = min(cifs_sb->rsize,
                              (tcon->ses->server->maxBuf - MAX_CIFS_HDR_SIZE));
 
-       /* volume_info.password is freed above when existing session found
+       /* volume_info->password is freed above when existing session found
       (in which case it is not needed anymore) but when new sesion is created
       the password ptr is put in the new session structure (in which case the
       password will be freed at unmount time) */
 out:
       /* zero out password before freeing */
-       if (volume_info.password != NULL) {
-               memset(volume_info.password, 0, strlen(volume_info.password));
-               kfree(volume_info.password);
+       if (volume_info) {
+               if (volume_info->password != NULL) {
+                       memset(volume_info->password, 0,
+                               strlen(volume_info->password));
+                       kfree(volume_info->password);
+               }
+               kfree(volume_info->UNC);
+               kfree(volume_info->prepath);
+               kfree(volume_info);
       }
-       kfree(volume_info.UNC);
-       kfree(volume_info.prepath);
       FreeXid(xid);
       return rc;
 }
--

1.5.5.1

Signature · 3 lines
Thanks,

Steve
#3 lore raw
Quoted 7 lines · On Sun, 30 Nov 2008 14:49:04 -0600 "Steve French" <[email protected]> wrote:

I have thought about this a few times, but always hesitated because I was hoping that eventually cifs mount option parsing could use new parsing functions (which would also shrink the "parse_mount_options" routine which we call during cifs mount), and would cause us to rewrite this. IIRC nfs was changed a few years ago to use the new mount mechanism (match_table and tokens).

I think we'll still have a smb_vol struct or something like it, regardless of how we end up doing the parsing. The results of the parsing have to be recorded somewhere. There's a similar struct for NFS as well (nfs_parsed_mount_data). We definitely should switch to the standard parser, but I don't think this patch will make much difference in how that is implemented.

Signature · 1 line
Jeff Layton <[email protected]>
#4 lore raw

Re: [PATCH 0/7] cifs: clean up socket creation, connection and sending (try #2)

In reply to a message not in the archive

After review of Jeff's 2 recent patch series, and discussion with Jeff on IRC (as a result Jeff respun and made minor modifications to a couple), I have just merged 10 of his 14 patches into cifs-2.6.git. See below:

3 min ago Jeff Layton cifs: make ipv6_connect take a TCP_Server_Info
3 min ago Jeff Layton cifs: make ipv4_connect take a TCP_Server_Info
14 min ago Jeff Layton cifs: don't declare smb_vol info on the stack
15 min ago Jeff Layton cifs: move allocation of new TCP_Server_Info
into separ ...
7 min ago Jeff Layton cifs: account for IPv6 in ses->serverName and clean ...
21 min ago Jeff Layton cifs: make dnotify thread experimental code
5 hours ago Jeff Layton cifs: convert tcpSem to a mutex
5 hours ago Jeff Layton cifs: take module reference when starting cifsd
6 hours ago Jeff Layton cifs: display addr and prefixpath options in /proc ...
6 hours ago Jeff Layton cifs: remove unused SMB session pointer from
struct ...

Quoted 44 lines · Jeff Layton

This patchset is intended as a cleanup of the code that creates and connects sockets. It's also intended to unify the smb_send and smb_send2 functions, and simplify the sending code in the blocking I/O case. There is a lot of duplicate code in those functions, and this should make it easier to ensure that we fix bugs with it properly. This should also fix the case where we get a partial send in smb_send. Previously, we did not force a reconnect in that case.

This is the second post of this patchset. The main differences from the original set are:

1) fix some symbol naming style issues that HCH pointed out

2) move the address handling into cifs_get_tcp_server and out of cifs_mount altogether (also pointed out by HCH)

3) the addition of a patch to reduce stack consumption in cifs_mount by dynamically allocating the volume_info

This patchset is also intended for 2.6.29 and depends on the 5 patch set of cleanups that I sent last week.

I've only given this set cursory testing. If it looks OK, I'll plan to do some more with it. The last set was more heavily tested, and showed no functional or performance regressions as best I could tell. Since the differences between that set and this one are fairly superficial, I don't expect much difference in performance here.

Jeff Layton (7):
cifs: convert tcpSem to a mutex
cifs: move allocation of new TCP_Server_Info into separate function
cifs: don't declare smb_vol info on the stack
cifs: make ipv4_connect take a TCP_Server_Info arg
cifs: make ipv6_connect take a TCP_Server_Info arg
cifs: turn smb_send into a wrapper around smb_sendv
cifs: don't retry on blocking sends

 fs/cifs/cifsglob.h  |    2 +-
 fs/cifs/cifsproto.h |    4 +-
 fs/cifs/connect.c   |  609 ++++++++++++++++++++++++++-------------------------
 fs/cifs/transport.c |  147 ++++---------
 4 files changed, 359 insertions(+), 403 deletions(-)
Signature · 3 lines
Thanks,

Steve
#5 lore raw

Re: [PATCH 0/7] cifs: clean up socket creation, connection and sending (try #2)

Thanks Steve,

Quoted 1 line

21 min ago Jeff Layton cifs: make dnotify thread experimental code

I think that this patch ^^^ may cause processes to hang indefinitely if the server goes unresponsive. I think the dnotify thread is the only thing that prevents this since it wakes things up every 15s. It would be good to get the wait_for_response patch in soon...

Thanks,

Signature · 1 line
Jeff Layton <[email protected]>