whiterose

linux unikernel
Log | Files | Refs | README | LICENSE | git clone https://git.ne02ptzero.me/git/whiterose

commit 0f9d140a566532175b4555401ee47ed58b01f9c9
parent e1706720408e72fb883f6b151c2b3b23d8e7e5b2
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Mon, 14 Jan 2019 05:43:40 +1200

Merge tag '5.0-rc1-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fixes from Steve French:
 "A set of cifs/smb3 fixes, 4 for stable, most from Pavel. His patches
  fix an important set of crediting (flow control) problems, and also
  two problems in cifs_writepages, ddressing some large i/o and also
  compounding issues"

* tag '5.0-rc1-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: update internal module version number
  CIFS: Fix error paths in writeback code
  CIFS: Move credit processing to mid callbacks for SMB3
  CIFS: Fix credits calculation for cancelled requests
  cifs: Fix potential OOB access of lock element array
  cifs: Limit memory used by lock request calls to a page
  cifs: move large array from stack to heap
  CIFS: Do not hide EINTR after sending network packets
  CIFS: Fix credit computation for compounded requests
  CIFS: Do not set credits to 1 if the server didn't grant anything
  CIFS: Fix adjustment of credits for MTU requests
  cifs: Fix a tiny potential memory leak
  cifs: Fix a debug message

Diffstat:
Mfs/cifs/cifsfs.h | 2+-
Mfs/cifs/cifsglob.h | 20++++++++++++++++++++
Mfs/cifs/cifssmb.c | 30++++++++++++++++++++----------
Mfs/cifs/connect.c | 7++++---
Mfs/cifs/dfs_cache.c | 1+
Mfs/cifs/file.c | 45+++++++++++++++++++++++++++++++++++----------
Mfs/cifs/inode.c | 10++++++++++
Mfs/cifs/smb2file.c | 8++++++--
Mfs/cifs/smb2pdu.c | 31++++++++++++++++++++++---------
Mfs/cifs/transport.c | 122+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--------------------
10 files changed, 211 insertions(+), 65 deletions(-)

diff --git a/fs/cifs/cifsfs.h b/fs/cifs/cifsfs.h @@ -150,5 +150,5 @@ extern long cifs_ioctl(struct file *filep, unsigned int cmd, unsigned long arg); extern const struct export_operations cifs_export_ops; #endif /* CONFIG_CIFS_NFSD_EXPORT */ -#define CIFS_VERSION "2.15" +#define CIFS_VERSION "2.16" #endif /* _CIFSFS_H */ diff --git a/fs/cifs/cifsglob.h b/fs/cifs/cifsglob.h @@ -1438,6 +1438,7 @@ struct mid_q_entry { int mid_state; /* wish this were enum but can not pass to wait_event */ unsigned int mid_flags; __le16 command; /* smb command code */ + unsigned int optype; /* operation type */ bool large_buf:1; /* if valid response, is pointer to large buf */ bool multiRsp:1; /* multiple trans2 responses for one request */ bool multiEnd:1; /* both received */ @@ -1574,6 +1575,25 @@ static inline void free_dfs_info_array(struct dfs_info3_param *param, kfree(param); } +static inline bool is_interrupt_error(int error) +{ + switch (error) { + case -EINTR: + case -ERESTARTSYS: + case -ERESTARTNOHAND: + case -ERESTARTNOINTR: + return true; + } + return false; +} + +static inline bool is_retryable_error(int error) +{ + if (is_interrupt_error(error) || error == -EAGAIN) + return true; + return false; +} + #define MID_FREE 0 #define MID_REQUEST_ALLOCATED 1 #define MID_REQUEST_SUBMITTED 2 diff --git a/fs/cifs/cifssmb.c b/fs/cifs/cifssmb.c @@ -128,24 +128,31 @@ static int __cifs_reconnect_tcon(const struct nls_table *nlsc, int rc; struct dfs_cache_tgt_list tl; struct dfs_cache_tgt_iterator *it = NULL; - char tree[MAX_TREE_SIZE + 1]; + char *tree; const char *tcp_host; size_t tcp_host_len; const char *dfs_host; size_t dfs_host_len; + tree = kzalloc(MAX_TREE_SIZE, GFP_KERNEL); + if (!tree) + return -ENOMEM; + if (tcon->ipc) { - snprintf(tree, sizeof(tree), "\\\\%s\\IPC$", + snprintf(tree, MAX_TREE_SIZE, "\\\\%s\\IPC$", tcon->ses->server->hostname); - return CIFSTCon(0, tcon->ses, tree, tcon, nlsc); + rc = CIFSTCon(0, tcon->ses, tree, tcon, nlsc); + goto out; } - if (!tcon->dfs_path) - return CIFSTCon(0, tcon->ses, tcon->treeName, tcon, nlsc); + if (!tcon->dfs_path) { + rc = CIFSTCon(0, tcon->ses, tcon->treeName, tcon, nlsc); + goto out; + } rc = dfs_cache_noreq_find(tcon->dfs_path + 1, NULL, &tl); if (rc) - return rc; + goto out; extract_unc_hostname(tcon->ses->server->hostname, &tcp_host, &tcp_host_len); @@ -165,7 +172,7 @@ static int __cifs_reconnect_tcon(const struct nls_table *nlsc, continue; } - snprintf(tree, sizeof(tree), "\\%s", tgt); + snprintf(tree, MAX_TREE_SIZE, "\\%s", tgt); rc = CIFSTCon(0, tcon->ses, tree, tcon, nlsc); if (!rc) @@ -182,6 +189,8 @@ static int __cifs_reconnect_tcon(const struct nls_table *nlsc, rc = -ENOENT; } dfs_cache_free_tgts(&tl); +out: + kfree(tree); return rc; } #else @@ -2114,7 +2123,7 @@ cifs_writev_requeue(struct cifs_writedata *wdata) for (j = 0; j < nr_pages; j++) { unlock_page(wdata2->pages[j]); - if (rc != 0 && rc != -EAGAIN) { + if (rc != 0 && !is_retryable_error(rc)) { SetPageError(wdata2->pages[j]); end_page_writeback(wdata2->pages[j]); put_page(wdata2->pages[j]); @@ -2123,7 +2132,7 @@ cifs_writev_requeue(struct cifs_writedata *wdata) if (rc) { kref_put(&wdata2->refcount, cifs_writedata_release); - if (rc == -EAGAIN) + if (is_retryable_error(rc)) continue; break; } @@ -2132,7 +2141,8 @@ cifs_writev_requeue(struct cifs_writedata *wdata) i += nr_pages; } while (i < wdata->nr_pages); - mapping_set_error(inode->i_mapping, rc); + if (rc != 0 && !is_retryable_error(rc)) + mapping_set_error(inode->i_mapping, rc); kref_put(&wdata->refcount, cifs_writedata_release); } diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c @@ -433,9 +433,10 @@ static void reconn_inval_dfs_target(struct TCP_Server_Info *server, kfree(server->hostname); server->hostname = extract_hostname(name); - if (!server->hostname) { - cifs_dbg(FYI, "%s: failed to extract hostname from target: %d\n", - __func__, -ENOMEM); + if (IS_ERR(server->hostname)) { + cifs_dbg(FYI, + "%s: failed to extract hostname from target: %ld\n", + __func__, PTR_ERR(server->hostname)); } } diff --git a/fs/cifs/dfs_cache.c b/fs/cifs/dfs_cache.c @@ -776,6 +776,7 @@ static int get_tgt_list(const struct dfs_cache_entry *ce, it->it_name = kstrndup(t->t_name, strlen(t->t_name), GFP_KERNEL); if (!it->it_name) { + kfree(it); rc = -ENOMEM; goto err_free_it; } diff --git a/fs/cifs/file.c b/fs/cifs/file.c @@ -733,7 +733,8 @@ reopen_success: if (can_flush) { rc = filemap_write_and_wait(inode->i_mapping); - mapping_set_error(inode->i_mapping, rc); + if (!is_interrupt_error(rc)) + mapping_set_error(inode->i_mapping, rc); if (tcon->unix_ext) rc = cifs_get_inode_info_unix(&inode, full_path, @@ -1132,14 +1133,18 @@ cifs_push_mandatory_locks(struct cifsFileInfo *cfile) /* * Accessing maxBuf is racy with cifs_reconnect - need to store value - * and check it for zero before using. + * and check it before using. */ max_buf = tcon->ses->server->maxBuf; - if (!max_buf) { + if (max_buf < (sizeof(struct smb_hdr) + sizeof(LOCKING_ANDX_RANGE))) { free_xid(xid); return -EINVAL; } + BUILD_BUG_ON(sizeof(struct smb_hdr) + sizeof(LOCKING_ANDX_RANGE) > + PAGE_SIZE); + max_buf = min_t(unsigned int, max_buf - sizeof(struct smb_hdr), + PAGE_SIZE); max_num = (max_buf - sizeof(struct smb_hdr)) / sizeof(LOCKING_ANDX_RANGE); buf = kcalloc(max_num, sizeof(LOCKING_ANDX_RANGE), GFP_KERNEL); @@ -1472,12 +1477,16 @@ cifs_unlock_range(struct cifsFileInfo *cfile, struct file_lock *flock, /* * Accessing maxBuf is racy with cifs_reconnect - need to store value - * and check it for zero before using. + * and check it before using. */ max_buf = tcon->ses->server->maxBuf; - if (!max_buf) + if (max_buf < (sizeof(struct smb_hdr) + sizeof(LOCKING_ANDX_RANGE))) return -EINVAL; + BUILD_BUG_ON(sizeof(struct smb_hdr) + sizeof(LOCKING_ANDX_RANGE) > + PAGE_SIZE); + max_buf = min_t(unsigned int, max_buf - sizeof(struct smb_hdr), + PAGE_SIZE); max_num = (max_buf - sizeof(struct smb_hdr)) / sizeof(LOCKING_ANDX_RANGE); buf = kcalloc(max_num, sizeof(LOCKING_ANDX_RANGE), GFP_KERNEL); @@ -2110,6 +2119,7 @@ static int cifs_writepages(struct address_space *mapping, pgoff_t end, index; struct cifs_writedata *wdata; int rc = 0; + int saved_rc = 0; unsigned int xid; /* @@ -2138,8 +2148,10 @@ retry: rc = server->ops->wait_mtu_credits(server, cifs_sb->wsize, &wsize, &credits); - if (rc) + if (rc != 0) { + done = true; break; + } tofind = min((wsize / PAGE_SIZE) - 1, end - index) + 1; @@ -2147,6 +2159,7 @@ retry: &found_pages); if (!wdata) { rc = -ENOMEM; + done = true; add_credits_and_wake_if(server, credits, 0); break; } @@ -2175,7 +2188,7 @@ retry: if (rc != 0) { add_credits_and_wake_if(server, wdata->credits, 0); for (i = 0; i < nr_pages; ++i) { - if (rc == -EAGAIN) + if (is_retryable_error(rc)) redirty_page_for_writepage(wbc, wdata->pages[i]); else @@ -2183,7 +2196,7 @@ retry: end_page_writeback(wdata->pages[i]); put_page(wdata->pages[i]); } - if (rc != -EAGAIN) + if (!is_retryable_error(rc)) mapping_set_error(mapping, rc); } kref_put(&wdata->refcount, cifs_writedata_release); @@ -2193,6 +2206,15 @@ retry: continue; } + /* Return immediately if we received a signal during writing */ + if (is_interrupt_error(rc)) { + done = true; + break; + } + + if (rc != 0 && saved_rc == 0) + saved_rc = rc; + wbc->nr_to_write -= nr_pages; if (wbc->nr_to_write <= 0) done = true; @@ -2210,6 +2232,9 @@ retry: goto retry; } + if (saved_rc != 0) + rc = saved_rc; + if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0)) mapping->writeback_index = index; @@ -2242,8 +2267,8 @@ cifs_writepage_locked(struct page *page, struct writeback_control *wbc) set_page_writeback(page); retry_write: rc = cifs_partialpagewrite(page, 0, PAGE_SIZE); - if (rc == -EAGAIN) { - if (wbc->sync_mode == WB_SYNC_ALL) + if (is_retryable_error(rc)) { + if (wbc->sync_mode == WB_SYNC_ALL && rc == -EAGAIN) goto retry_write; redirty_page_for_writepage(wbc, page); } else if (rc != 0) { diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c @@ -2257,6 +2257,11 @@ cifs_setattr_unix(struct dentry *direntry, struct iattr *attrs) * the flush returns error? */ rc = filemap_write_and_wait(inode->i_mapping); + if (is_interrupt_error(rc)) { + rc = -ERESTARTSYS; + goto out; + } + mapping_set_error(inode->i_mapping, rc); rc = 0; @@ -2400,6 +2405,11 @@ cifs_setattr_nounix(struct dentry *direntry, struct iattr *attrs) * the flush returns error? */ rc = filemap_write_and_wait(inode->i_mapping); + if (is_interrupt_error(rc)) { + rc = -ERESTARTSYS; + goto cifs_setattr_exit; + } + mapping_set_error(inode->i_mapping, rc); rc = 0; diff --git a/fs/cifs/smb2file.c b/fs/cifs/smb2file.c @@ -122,12 +122,14 @@ smb2_unlock_range(struct cifsFileInfo *cfile, struct file_lock *flock, /* * Accessing maxBuf is racy with cifs_reconnect - need to store value - * and check it for zero before using. + * and check it before using. */ max_buf = tcon->ses->server->maxBuf; - if (!max_buf) + if (max_buf < sizeof(struct smb2_lock_element)) return -EINVAL; + BUILD_BUG_ON(sizeof(struct smb2_lock_element) > PAGE_SIZE); + max_buf = min_t(unsigned int, max_buf, PAGE_SIZE); max_num = max_buf / sizeof(struct smb2_lock_element); buf = kcalloc(max_num, sizeof(struct smb2_lock_element), GFP_KERNEL); if (!buf) @@ -264,6 +266,8 @@ smb2_push_mandatory_locks(struct cifsFileInfo *cfile) return -EINVAL; } + BUILD_BUG_ON(sizeof(struct smb2_lock_element) > PAGE_SIZE); + max_buf = min_t(unsigned int, max_buf, PAGE_SIZE); max_num = max_buf / sizeof(struct smb2_lock_element); buf = kcalloc(max_num, sizeof(struct smb2_lock_element), GFP_KERNEL); if (!buf) { diff --git a/fs/cifs/smb2pdu.c b/fs/cifs/smb2pdu.c @@ -162,24 +162,31 @@ static int __smb2_reconnect(const struct nls_table *nlsc, int rc; struct dfs_cache_tgt_list tl; struct dfs_cache_tgt_iterator *it = NULL; - char tree[MAX_TREE_SIZE + 1]; + char *tree; const char *tcp_host; size_t tcp_host_len; const char *dfs_host; size_t dfs_host_len; + tree = kzalloc(MAX_TREE_SIZE, GFP_KERNEL); + if (!tree) + return -ENOMEM; + if (tcon->ipc) { - snprintf(tree, sizeof(tree), "\\\\%s\\IPC$", + snprintf(tree, MAX_TREE_SIZE, "\\\\%s\\IPC$", tcon->ses->server->hostname); - return SMB2_tcon(0, tcon->ses, tree, tcon, nlsc); + rc = SMB2_tcon(0, tcon->ses, tree, tcon, nlsc); + goto out; } - if (!tcon->dfs_path) - return SMB2_tcon(0, tcon->ses, tcon->treeName, tcon, nlsc); + if (!tcon->dfs_path) { + rc = SMB2_tcon(0, tcon->ses, tcon->treeName, tcon, nlsc); + goto out; + } rc = dfs_cache_noreq_find(tcon->dfs_path + 1, NULL, &tl); if (rc) - return rc; + goto out; extract_unc_hostname(tcon->ses->server->hostname, &tcp_host, &tcp_host_len); @@ -199,7 +206,7 @@ static int __smb2_reconnect(const struct nls_table *nlsc, continue; } - snprintf(tree, sizeof(tree), "\\%s", tgt); + snprintf(tree, MAX_TREE_SIZE, "\\%s", tgt); rc = SMB2_tcon(0, tcon->ses, tree, tcon, nlsc); if (!rc) @@ -216,6 +223,8 @@ static int __smb2_reconnect(const struct nls_table *nlsc, rc = -ENOENT; } dfs_cache_free_tgts(&tl); +out: + kfree(tree); return rc; } #else @@ -3278,12 +3287,14 @@ smb2_async_readv(struct cifs_readdata *rdata) if (rdata->credits) { shdr->CreditCharge = cpu_to_le16(DIV_ROUND_UP(rdata->bytes, SMB2_MAX_BUFFER_SIZE)); - shdr->CreditRequest = shdr->CreditCharge; + shdr->CreditRequest = + cpu_to_le16(le16_to_cpu(shdr->CreditCharge) + 1); spin_lock(&server->req_lock); server->credits += rdata->credits - le16_to_cpu(shdr->CreditCharge); spin_unlock(&server->req_lock); wake_up(&server->request_q); + rdata->credits = le16_to_cpu(shdr->CreditCharge); flags |= CIFS_HAS_CREDITS; } @@ -3555,12 +3566,14 @@ smb2_async_writev(struct cifs_writedata *wdata, if (wdata->credits) { shdr->CreditCharge = cpu_to_le16(DIV_ROUND_UP(wdata->bytes, SMB2_MAX_BUFFER_SIZE)); - shdr->CreditRequest = shdr->CreditCharge; + shdr->CreditRequest = + cpu_to_le16(le16_to_cpu(shdr->CreditCharge) + 1); spin_lock(&server->req_lock); server->credits += wdata->credits - le16_to_cpu(shdr->CreditCharge); spin_unlock(&server->req_lock); wake_up(&server->request_q); + wdata->credits = le16_to_cpu(shdr->CreditCharge); flags |= CIFS_HAS_CREDITS; } diff --git a/fs/cifs/transport.c b/fs/cifs/transport.c @@ -387,7 +387,7 @@ smbd_done: if (rc < 0 && rc != -EINTR) cifs_dbg(VFS, "Error %d sending data on socket to server\n", rc); - else + else if (rc > 0) rc = 0; return rc; @@ -783,8 +783,34 @@ cifs_setup_request(struct cifs_ses *ses, struct smb_rqst *rqst) } static void -cifs_noop_callback(struct mid_q_entry *mid) +cifs_compound_callback(struct mid_q_entry *mid) { + struct TCP_Server_Info *server = mid->server; + unsigned int optype = mid->optype; + unsigned int credits_received = 0; + + if (mid->mid_state == MID_RESPONSE_RECEIVED) { + if (mid->resp_buf) + credits_received = server->ops->get_credits(mid); + else + cifs_dbg(FYI, "Bad state for cancelled MID\n"); + } + + add_credits(server, credits_received, optype); +} + +static void +cifs_compound_last_callback(struct mid_q_entry *mid) +{ + cifs_compound_callback(mid); + cifs_wake_up_task(mid); +} + +static void +cifs_cancelled_callback(struct mid_q_entry *mid) +{ + cifs_compound_callback(mid); + DeleteMidQEntry(mid); } int @@ -795,7 +821,8 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, int i, j, rc = 0; int timeout, optype; struct mid_q_entry *midQ[MAX_COMPOUND]; - unsigned int credits = 0; + bool cancelled_mid[MAX_COMPOUND] = {false}; + unsigned int credits[MAX_COMPOUND] = {0}; char *buf; timeout = flags & CIFS_TIMEOUT_MASK; @@ -813,13 +840,31 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, return -ENOENT; /* - * Ensure that we do not send more than 50 overlapping requests - * to the same server. We may make this configurable later or - * use ses->maxReq. + * Ensure we obtain 1 credit per request in the compound chain. + * It can be optimized further by waiting for all the credits + * at once but this can wait long enough if we don't have enough + * credits due to some heavy operations in progress or the server + * not granting us much, so a fallback to the current approach is + * needed anyway. */ - rc = wait_for_free_request(ses->server, timeout, optype); - if (rc) - return rc; + for (i = 0; i < num_rqst; i++) { + rc = wait_for_free_request(ses->server, timeout, optype); + if (rc) { + /* + * We haven't sent an SMB packet to the server yet but + * we already obtained credits for i requests in the + * compound chain - need to return those credits back + * for future use. Note that we need to call add_credits + * multiple times to match the way we obtained credits + * in the first place and to account for in flight + * requests correctly. + */ + for (j = 0; j < i; j++) + add_credits(ses->server, 1, optype); + return rc; + } + credits[i] = 1; + } /* * Make sure that we sign in the same order that we send on this socket @@ -835,18 +880,24 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, for (j = 0; j < i; j++) cifs_delete_mid(midQ[j]); mutex_unlock(&ses->server->srv_mutex); + /* Update # of requests on wire to server */ - add_credits(ses->server, 1, optype); + for (j = 0; j < num_rqst; j++) + add_credits(ses->server, credits[j], optype); return PTR_ERR(midQ[i]); } midQ[i]->mid_state = MID_REQUEST_SUBMITTED; + midQ[i]->optype = optype; /* - * We don't invoke the callback compounds unless it is the last - * request. + * Invoke callback for every part of the compound chain + * to calculate credits properly. Wake up this thread only when + * the last element is received. */ if (i < num_rqst - 1) - midQ[i]->callback = cifs_noop_callback; + midQ[i]->callback = cifs_compound_callback; + else + midQ[i]->callback = cifs_compound_last_callback; } cifs_in_send_inc(ses->server); rc = smb_send_rqst(ses->server, num_rqst, rqst, flags); @@ -860,8 +911,20 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, mutex_unlock(&ses->server->srv_mutex); - if (rc < 0) + if (rc < 0) { + /* Sending failed for some reason - return credits back */ + for (i = 0; i < num_rqst; i++) + add_credits(ses->server, credits[i], optype); goto out; + } + + /* + * At this point the request is passed to the network stack - we assume + * that any credits taken from the server structure on the client have + * been spent and we can't return them back. Once we receive responses + * we will collect credits granted by the server in the mid callbacks + * and add those credits to the server structure. + */ /* * Compounding is never used during session establish. @@ -875,36 +938,34 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, for (i = 0; i < num_rqst; i++) { rc = wait_for_response(ses->server, midQ[i]); - if (rc != 0) { + if (rc != 0) + break; + } + if (rc != 0) { + for (; i < num_rqst; i++) { cifs_dbg(VFS, "Cancelling wait for mid %llu cmd: %d\n", midQ[i]->mid, le16_to_cpu(midQ[i]->command)); send_cancel(ses->server, &rqst[i], midQ[i]); spin_lock(&GlobalMid_Lock); if (midQ[i]->mid_state == MID_REQUEST_SUBMITTED) { midQ[i]->mid_flags |= MID_WAIT_CANCELLED; - midQ[i]->callback = DeleteMidQEntry; - spin_unlock(&GlobalMid_Lock); - add_credits(ses->server, 1, optype); - return rc; + midQ[i]->callback = cifs_cancelled_callback; + cancelled_mid[i] = true; + credits[i] = 0; } spin_unlock(&GlobalMid_Lock); } } - for (i = 0; i < num_rqst; i++) - if (midQ[i]->resp_buf) - credits += ses->server->ops->get_credits(midQ[i]); - if (!credits) - credits = 1; - for (i = 0; i < num_rqst; i++) { if (rc < 0) goto out; rc = cifs_sync_mid_result(midQ[i], ses->server); if (rc != 0) { - add_credits(ses->server, credits, optype); - return rc; + /* mark this mid as cancelled to not free it below */ + cancelled_mid[i] = true; + goto out; } if (!midQ[i]->resp_buf || @@ -951,9 +1012,10 @@ out: * This is prevented above by using a noop callback that will not * wake this thread except for the very last PDU. */ - for (i = 0; i < num_rqst; i++) - cifs_delete_mid(midQ[i]); - add_credits(ses->server, credits, optype); + for (i = 0; i < num_rqst; i++) { + if (!cancelled_mid[i]) + cifs_delete_mid(midQ[i]); + } return rc; }