staging/lustre: patch bomb 2
cover staging/lustre: patch bomb 2
Hi Greg,
This is the second patch bomb. All 39 patches are ported from Lustre tree. And it syncs kernel client with Lustre tree up to commit (LU-3581 osc: Lustre returns EINTR from writes when SA_RESTART is set).
Thanks,
Tao
Andreas Dilger (1):
staging/lustre/osd-ldiskfs: remove dependency on mdd module
Andriy Skulysh (2):
staging/lustre/ptlrpc: add rpc_cache staging/lustre/ldlm: ldlm_flock_deadlock() ASSERTION( req != lock )
failed
Aurelien Degremont (1):
staging/lustre/hsm: copy start error should set HP_FLAG_COMPLETED
Bob Glossman (1):
staging/lustre/libcfs: improve validity test for valid file
descriptors
Christopher J. Morrone (1):
staging/lustre/changelogs: Account for changelog_ext_rec in
CR_MAXSIZE
Dmitry Eremin (2):
staging/lustre/lnet: build Lustre with Intel OFED for Xeon Phi staging/lustre/build: build error with gcc 4.7.0 20110509
Fan Yong (1):
staging/lustre/llite: return compatible fsid for statfs
Gaurav Mahajan (1):
staging/lustre/llite: Delaying creation of client side proc entries.
Henri Doreau (1):
staging/lustre/utils: Posix copytool fixes.
JC Lafoucriere (3):
staging/lustre/mdt: CDT cleanup follow-on patch staging/lustre/lmv: update coding style staging/lustre/mdt: HSM EXIST event not triggered at last rm/mv
James Simmons (5):
staging/lustre/lvfs: remove llog_lvfs.c and other lvfs code from llog staging/lustre/autoconf: remove LC_LOCK_MAP_ACQUIRE test staging/lustre/autoconf: remove LC_FS_STRUCT_USE_PATH staging/lustre/autoconf: remove LC_SECURITY_PLUG test staging/lustre/osc: Lustre returns EINTR from writes when SA_RESTART
is set
Jinshan Xiong (4):
staging/lustre/hsm: small fixes for HSM staging/lustre/llite: reset writeback index in ll_writepages staging/lustre/llite: cancel open lock before closing file staging/lustre/hsm: Add support to drop all pages for ll_data_version
John L. Hammond (3):
staging/lustre/hsm: remove hsm_nl proc file staging/lustre/clio: honor O_NOATIME staging/lustre/obd: add md_stats to MDC and LMV devices
Li Xi (1):
staging/lustre/libcfs: Add relocation function to libcfs heap
Mikhail Pershin (1):
staging/lustre/obdclass: fix wrong device put in case of race
Ned Bass (1):
staging/lustre/mdc: document mdc_rpc_lock
Niu Yawei (1):
staging/lustre/llite: use 64bits flags in ll_lov_setea()
Sebastien Buisson (5):
staging/lustre/build: fix 'deadcode' errors staging/lustre/build: fix 'integer handling' issues staging/lustre/build: fix 'data race condition' issues staging/lustre/build: fix 'NULL pointer dereference' errors staging/lustre/build: fix 'no effect' errors
Swapnil Pimpale (1):
staging/lustre/lov: Get the correct address of lmm_objects
Vitaly Fertman (1):
staging/lustre/clio: wrong cl_lock usage
wang di (2):
staging/lustre/lmv: choose right MDT for open by FID staging/lustre/osd: remove fld lookup during configuration
.../lustre/include/linux/libcfs/libcfs_debug.h | 129 +-- .../lustre/include/linux/libcfs/libcfs_heap.h | 1 + .../include/linux/libcfs/libcfs_kernelcomm.h | 1 + .../staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h | 2 + drivers/staging/lustre/lnet/selftest/conrpc.c | 8 +- drivers/staging/lustre/lustre/include/cl_object.h | 6 +- .../lustre/lustre/include/linux/lustre_compat25.h | 14 - .../lustre/lustre/include/linux/lustre_fsfilt.h | 84 -- drivers/staging/lustre/lustre/include/linux/lvfs.h | 5 - .../lustre/lustre/include/lustre/lustre_idl.h | 8 +- .../lustre/lustre/include/lustre/lustre_user.h | 16 +- .../lustre/lustre/include/lustre/lustreapi.h | 7 +- .../staging/lustre/lustre/include/lustre_capa.h | 25 +- .../staging/lustre/lustre/include/lustre_export.h | 1 - drivers/staging/lustre/lustre/include/lustre_log.h | 4 - drivers/staging/lustre/lustre/include/lustre_mdc.h | 18 + .../staging/lustre/lustre/include/lustre_update.h | 3 +- drivers/staging/lustre/lustre/include/md_object.h | 20 - drivers/staging/lustre/lustre/include/obd.h | 139 ++-- drivers/staging/lustre/lustre/include/obd_class.h | 129 ++- drivers/staging/lustre/lustre/ldlm/ldlm_flock.c | 3 + drivers/staging/lustre/lustre/libcfs/debug.c | 131 +-- drivers/staging/lustre/lustre/libcfs/heap.c | 18 +- .../lustre/lustre/libcfs/kernel_user_comm.c | 10 +- .../lustre/lustre/libcfs/linux/linux-curproc.c | 4 +- drivers/staging/lustre/lustre/llite/dir.c | 30 +- drivers/staging/lustre/lustre/llite/file.c | 98 ++- .../staging/lustre/lustre/llite/llite_internal.h | 6 +- drivers/staging/lustre/lustre/llite/llite_lib.c | 25 +- drivers/staging/lustre/lustre/llite/llite_nfs.c | 16 - drivers/staging/lustre/lustre/llite/rw.c | 6 +- drivers/staging/lustre/lustre/llite/xattr.c | 11 +- drivers/staging/lustre/lustre/lmv/lmv_intent.c | 8 +- drivers/staging/lustre/lustre/lmv/lmv_obd.c | 115 +-- drivers/staging/lustre/lustre/lov/lov_io.c | 4 +- drivers/staging/lustre/lustre/lov/lov_merge.c | 13 +- drivers/staging/lustre/lustre/lov/lov_obd.c | 2 +- drivers/staging/lustre/lustre/lov/lov_pack.c | 3 +- drivers/staging/lustre/lustre/lvfs/fsfilt_ext3.c | 760 ------------------ drivers/staging/lustre/lustre/lvfs/lvfs_lib.c | 6 + drivers/staging/lustre/lustre/lvfs/lvfs_linux.c | 17 +- drivers/staging/lustre/lustre/mdc/lproc_mdc.c | 73 -- drivers/staging/lustre/lustre/mdc/mdc_lib.c | 2 +- drivers/staging/lustre/lustre/mdc/mdc_request.c | 10 +- drivers/staging/lustre/lustre/obdclass/capa.c | 41 + drivers/staging/lustre/lustre/obdclass/cl_lock.c | 2 +- drivers/staging/lustre/lustre/obdclass/class_obd.c | 5 + drivers/staging/lustre/lustre/obdclass/llog_cat.c | 34 +- .../staging/lustre/lustre/obdclass/llog_ioctl.c | 6 +- drivers/staging/lustre/lustre/obdclass/llog_lvfs.c | 847 -------------------- .../staging/lustre/lustre/obdclass/local_storage.c | 13 +- .../lustre/lustre/obdclass/lprocfs_status.c | 58 +- drivers/staging/lustre/lustre/osc/osc_cache.c | 43 +- drivers/staging/lustre/lustre/osc/osc_io.c | 21 +- drivers/staging/lustre/lustre/osc/osc_lock.c | 19 +- drivers/staging/lustre/lustre/ptlrpc/client.c | 42 +- drivers/staging/lustre/lustre/ptlrpc/events.c | 2 +- .../staging/lustre/lustre/ptlrpc/ptlrpc_internal.h | 4 + .../staging/lustre/lustre/ptlrpc/ptlrpc_module.c | 24 +- drivers/staging/lustre/lustre/ptlrpc/sec.c | 6 +- drivers/staging/lustre/lustre/ptlrpc/sec_bulk.c | 1 - drivers/staging/lustre/lustre/ptlrpc/sec_config.c | 2 +- drivers/staging/lustre/lustre/ptlrpc/service.c | 13 +- 63 files changed, 772 insertions(+), 2402 deletions(-) delete mode 100644 drivers/staging/lustre/lustre/lvfs/fsfilt_ext3.c delete mode 100644 drivers/staging/lustre/lustre/obdclass/llog_lvfs.c
Signature · 1 line
1.7.9.5
PATCH 1/39 staging/lustre/hsm: remove hsm_nl proc file
From: "John L. Hammond" <[email protected]>
Remove the file /proc/fs/lustre/mdc/*/hsm_nl which was introduced "for testing purposes."
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2489
Lustre-change: http://review.whamcloud.com/6656
--- drivers/staging/lustre/lustre/mdc/lproc_mdc.c | 73 ------------------------- 1 file changed, 73 deletions(-)
drivers/staging/lustre/lustre/mdc/lproc_mdc.c +0 -73
@@ -77,78 +77,6 @@ static ssize_t mdc_max_rpcs_in_flight_seq_write(struct file *file, }LPROC_SEQ_FOPS (mdc_max_rpcs_in_flight); -static int mdc_kuc_open (struct inode *inode,struct file *file) -{ -return single_open (file,NULL ,PDE_DATA (inode)); -} - -/* temporary for testing */ -static ssize_t mdc_kuc_write (struct file *file,const char *buffer, -size_t count,loff_t *off) -{ -struct obd_device *obd = ((struct seq_file *)file->private_data)->private; -struct kuc_hdr *lh; -struct hsm_action_list *hal; -struct hsm_action_item *hai; -int len; -int fd, rc; - - rc =lprocfs_write_helper (buffer, count, &fd); -if (rc) -return rc; - -if (fd <0 ) -return -ERANGE ; -CWARN ("message to fd %d\n" , fd); - - len =sizeof (*lh) +sizeof (*hal) +MTI_NAME_MAXLEN + -/* for mockup below */ 2 *cfs_size_round (sizeof (*hai)); - -OBD_ALLOC (lh, len); - - lh->kuc_magic =KUC_MAGIC ; - lh->kuc_transport =KUC_TRANSPORT_HSM ; - lh->kuc_msgtype =HMT_ACTION_LIST ; - lh->kuc_msglen = len; - - hal = (struct hsm_action_list *)(lh +1 ); - hal->hal_version =HAL_VERSION ; - hal->hal_archive_id =1 ; - hal->hal_flags =0 ; -obd_uuid2fsname (hal->hal_fsname, obd->obd_name,MTI_NAME_MAXLEN ); - -/* mock up an action list */ - hal->hal_count =2 ; - hai =hai_zero (hal); - hai->hai_action =HSMA_ARCHIVE ; - hai->hai_fid.f_oid =5 ; - hai->hai_len =sizeof (*hai); - hai =hai_next (hai); - hai->hai_action =HSMA_RESTORE ; - hai->hai_fid.f_oid =10 ; - hai->hai_len =sizeof (*hai); - -/* This works for either broadcast or unicast to a single fd */ -if (fd ==0 ) { - rc =libcfs_kkuc_group_put (KUC_GRP_HSM , lh); - }else { -struct file *fp =fget (fd); - - rc =libcfs_kkuc_msg_put (fp, lh); -fput (fp); - } -OBD_FREE (lh, len); -if (rc <0 ) -return rc; -return count; -} - -struct file_operations mdc_kuc_fops = { - .open = mdc_kuc_open, - .write = mdc_kuc_write, - .release = single_release, -}; -LPROC_SEQ_FOPS_WR_ONLY (mdc, ping);LPROC_SEQ_FOPS_RO_TYPE (mdc, uuid);
@@ -197,7 +125,6 @@ static struct lprocfs_vars lprocfs_mdc_obd_vars[] = { { "timeouts", &mdc_timeouts_fops, 0, 0 }, { "import", &mdc_import_fops, 0 }, { "state", &mdc_state_fops, 0, 0 }, - { "hsm_nl", &mdc_kuc_fops, 0, 0200 }, { "pinger_recov", &mdc_pinger_recov_fops, 0, 0 }, { 0 } };
Signature · 1 line
1.7.9.5
Greg Kroah-Hartman
maintainerRe: [PATCH 01/39] staging/lustre/hsm: remove hsm_nl proc file
Quoted 4 lines · Peng Tao
From: "John L. Hammond" <[email protected]>
Remove the file
/proc/fs/lustre/mdc/*/hsm_nlwhich was introduced "for testing purposes."
What does this mean? Why is this allowed? How do you know no one uses it? I need a much better changelog comment here, sorry.
greg k-h
Re: [PATCH 01/39] staging/lustre/hsm: remove hsm_nl proc file
Quoted 9 lines · Greg Kroah-Hartman
Quoted 4 lines · Peng Tao
From: "John L. Hammond" <[email protected]>
Remove the file
/proc/fs/lustre/mdc/*/hsm_nlwhich was introduced "for testing purposes."What does this mean? Why is this allowed? How do you know no one uses it? I need a much better changelog comment here, sorry.
Those proc entries were added for testing HSM
(http://en.wikipedia.org/wiki/Hierarchical_storage_management) during the time of developing Lustre HSM feature. Now the HSM feature is almost completed, those files no longer serve their purpose and can be removed so that no one would develop dependency upon it after Lustre HSM is announced released.
I'll add this to the commit message.
Thanks for reviewing,
Tao
PATCH 2/39 staging/lustre/clio: wrong cl_lock usage
In reply to Peng TaoFrom: Vitaly Fertman <[email protected]>
granted lock is moved to HELD state in cl_wait_try() only which is done after upcall for non-agl locks. as the result, lock unuse moves cl_lock not to CACHED state, but to NEW state. A parallel thread gets this lock and tries to enqueue it - instead of re-using a cached lock, cl_enqueue_try() initiates a new enqueue and gets to osc_lock_enqueue() where it asserts:
LASSERTF(ols->ols_state == OLS_NEW, ...);
the state of osc lock is RELEASED already - moved here by unuse.
Lustre-change: http://review.whamcloud.com/6709
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3433
--- drivers/staging/lustre/lustre/osc/osc_lock.c | 19 ++++++++++++------- 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/staging/lustre/lustre/osc/osc_lock.c +12 -7
@@ -554,15 +554,18 @@ static int osc_lock_upcall(void *cookie, int errcode) lock->cll_flags |=CLF_FROM_UPCALL ;cl_wait_try (env, lock); lock->cll_flags &= ~CLF_FROM_UPCALL ; -if (!olck->ols_glimpse) - olck->ols_agl =0 ; }cl_lock_signal (env, lock);/* del user for lock upcall cookie */ -cl_unuse_try (env, lock); +if (olck->ols_agl) { +if (!olck->ols_glimpse) + olck->ols_agl =0 ; +cl_unuse_try (env, lock); + } }else {/* del user for lock upcall cookie */ -cl_lock_user_del (env, lock); +if (olck->ols_agl) +cl_lock_user_del (env, lock);cl_lock_error (env, lock, rc); }
@@ -1128,8 +1131,9 @@ static int osc_lock_enqueue(const struct lu_env *env,/* lock will be passed as upcall cookie, * hold ref to prevent to be released. */ cl_lock_hold_add (env, lock,"upcall" , lock); -/* a user for lock also */ -cl_lock_user_add (env, lock); +/* a user for agl lock also */ +if (ols->ols_agl) +cl_lock_user_add (env, lock); ols->ols_state =OLS_ENQUEUED ;/*
@@ -1147,7 +1151,8 @@ static int osc_lock_enqueue(const struct lu_env *env, ols, einfo, &ols->ols_handle, PTLRPCD_SET, 1, ols->ols_agl);if (result !=0 ) { -cl_lock_user_del (env, lock); +if (ols->ols_agl) +cl_lock_user_del (env, lock);cl_lock_unhold (env, lock,"upcall" , lock);if (unlikely (result == -ECANCELED )) { ols->ols_state =OLS_NEW ;
Signature · 1 line
1.7.9.5
PATCH 3/39 staging/lustre/build: fix 'deadcode' errors
In reply to Peng TaoFrom: Sebastien Buisson <[email protected]>
Fix 'deadcode' issues found by Coverity version 6.5.1:
Logically dead code (DEADCODE)
Execution cannot reach this statement.
Lustre-change: http://review.whamcloud.com/7167
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3097
--- .../lustre/lustre/libcfs/linux/linux-curproc.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/staging/lustre/lustre/libcfs/linux/linux-curproc.c +1 -3
@@ -195,7 +195,7 @@ static int cfs_access_process_vm(struct task_struct *tsk, unsigned long addr,int cfs_get_environ (const char *key,char *value,int *val_len) {struct mm_struct *mm; -char *buffer, *tmp_buf =NULL ; +char *buffer;int buf_len =PAGE_CACHE_SIZE ;int key_len =strlen (key);unsigned long addr;
@@ -286,8 +286,6 @@ int cfs_get_environ(const char *key, char *value, int *val_len) out:mmput (mm);kfree ((void *)buffer); -if (tmp_buf) -kfree ((void *)tmp_buf);return rc; }EXPORT_SYMBOL (cfs_get_environ);
Signature · 1 line
1.7.9.5
PATCH 4/39 staging/lustre/build: fix 'integer handling' issues
In reply to Peng TaoFrom: Sebastien Buisson <[email protected]>
Fix 'program hangs' defects found by Coverity version 6.5.1:
Improper use of negative value (NEGATIVE_RETURNS) Variable is passed to a parameter that cannot be negative. Macro compares unsigned to 0 (NO_EFFECT)
This comparison is always true.
Operands don't affect result (CONSTANT_EXPRESSION_RESULT) Test is always false regardless of the values of its operands. Wrong operator used (CONSTANT_EXPRESSION_RESULT) Test is always true regardless of the values of its operand.
Lustre-change: http://review.whamcloud.com/5888
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3069
--- drivers/staging/lustre/lustre/llite/rw.c | 1 - .../staging/lustre/lustre/obdclass/llog_ioctl.c | 6 ++++-- 2 files changed, 4 insertions(+), 3 deletions(-)
drivers/staging/lustre/lustre/llite/rw.c +0 -1
@@ -803,7 +803,6 @@ int ll_readahead(const struct lu_env *env, struct cl_io *io, ret =ll_read_ahead_pages (env, io, queue, ria, &reserved, mapping, &ra_end); -LASSERTF (reserved >=0 ,"reserved %lu\n" , reserved);if (reserved !=0 )ll_ra_count_put (ll_i2sbi (inode), reserved);
drivers/staging/lustre/lustre/obdclass/llog_ioctl.c +4 -2
@@ -90,7 +90,8 @@ static int llog_check_cb(const struct lu_env *env, struct llog_handle *handle, struct llog_rec_hdr *rec, void *data) { struct obd_ioctl_data *ioc_data = (struct obd_ioctl_data *)data; -static int l, remains, from, to; +static int l, remains; +static long from, to;static char *out;char *endp;int cur_index, rc =0 ;
@@ -174,7 +175,8 @@ static int llog_print_cb(const struct lu_env *env, struct llog_handle *handle, struct llog_rec_hdr *rec, void *data) { struct obd_ioctl_data *ioc_data = (struct obd_ioctl_data *)data; -static int l, remains, from, to; +static int l, remains; +static long from, to;static char *out;char *endp;int cur_index;
Signature · 1 line
1.7.9.5
PATCH 5/39 staging/lustre/ptlrpc: add rpc_cache
In reply to Peng TaoFrom: Andriy Skulysh <[email protected]>
Add rpc_cache for allocating ptlrpc_requests.
Lustre-change: http://review.whamcloud.com/6874
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2424
--- drivers/staging/lustre/lustre/ptlrpc/client.c | 42 ++++++++++++++++---- drivers/staging/lustre/lustre/ptlrpc/events.c | 2 +- .../staging/lustre/lustre/ptlrpc/ptlrpc_internal.h | 4 ++ .../staging/lustre/lustre/ptlrpc/ptlrpc_module.c | 24 +++++++---- drivers/staging/lustre/lustre/ptlrpc/sec.c | 6 +-- drivers/staging/lustre/lustre/ptlrpc/service.c | 13 +++--- 6 files changed, 65 insertions(+), 26 deletions(-)
drivers/staging/lustre/lustre/ptlrpc/client.c +35 -7
@@ -378,6 +378,34 @@ static int ptlrpc_at_recv_early_reply(struct ptlrpc_request *req)return rc; } +struct kmem_cache *request_cache; + +int ptlrpc_request_cache_init () +{ + request_cache =kmem_cache_create ("ptlrpc_cache" , +sizeof (struct ptlrpc_request ), +0 ,SLAB_HWCACHE_ALIGN ,NULL ); +return request_cache ==NULL ? -ENOMEM :0 ; +} + +void ptlrpc_request_cache_fini () +{ +kmem_cache_destroy (request_cache); +} + +struct ptlrpc_request *ptlrpc_request_cache_alloc (int flags) +{ +struct ptlrpc_request *req; + +OBD_SLAB_ALLOC_PTR_GFP (req, request_cache, flags); +return req; +} + +void ptlrpc_request_cache_free (struct ptlrpc_request *req) +{ +OBD_SLAB_FREE_PTR (req, request_cache); +} +/** * Wind down request pool \a pool. * Frees all requests from the pool too
@@ -396,7 +424,7 @@ void ptlrpc_free_rq_pool(struct ptlrpc_request_pool *pool)LASSERT (req->rq_reqbuf);LASSERT (req->rq_reqbuf_len == pool->prp_rq_size);OBD_FREE_LARGE (req->rq_reqbuf, pool->prp_rq_size); -OBD_FREE (req,sizeof (*req)); +ptlrpc_request_cache_free (req); }spin_unlock (&pool->prp_lock);OBD_FREE (pool,sizeof (*pool));
@@ -426,12 +454,12 @@ void ptlrpc_add_rqs_to_pool(struct ptlrpc_request_pool *pool, int num_rq)struct lustre_msg *msg;spin_unlock (&pool->prp_lock); -OBD_ALLOC (req,sizeof (struct ptlrpc_request )); + req =ptlrpc_request_cache_alloc (__GFP_IO);if (!req)return ;OBD_ALLOC_LARGE (msg, size);if (!msg) { -OBD_FREE (req,sizeof (struct ptlrpc_request )); +ptlrpc_request_cache_free (req);return ; } req->rq_reqbuf = msg;
@@ -667,7 +695,7 @@ struct ptlrpc_request *__ptlrpc_request_alloc(struct obd_import *imp, request =ptlrpc_prep_req_from_pool (pool);if (!request) -OBD_ALLOC_PTR (request); + request =ptlrpc_request_cache_alloc (__GFP_IO);if (request) {LASSERTF ((unsigned long )imp >0x1000 ,"%p" , imp);
@@ -738,7 +766,7 @@ void ptlrpc_request_free(struct ptlrpc_request *request)if (request->rq_pool)__ptlrpc_free_req_to_pool (request);else -OBD_FREE_PTR (request); +ptlrpc_request_cache_free (request); }EXPORT_SYMBOL (ptlrpc_request_free);
@@ -2216,7 +2244,7 @@ static void __ptlrpc_free_req(struct ptlrpc_request *request, int locked)if (request->rq_pool)__ptlrpc_free_req_to_pool (request);else -OBD_FREE (request,sizeof (*request)); +ptlrpc_request_cache_free (request); }static int __ptlrpc_req_finished (struct ptlrpc_request *request,int locked);
@@ -2934,7 +2962,7 @@ void *ptlrpcd_alloc_work(struct obd_import *imp,return ERR_PTR (-EINVAL );/* copy some code from deprecated fakereq. */ -OBD_ALLOC_PTR (req); + req =ptlrpc_request_cache_alloc (__GFP_IO);if (req ==NULL ) {CERROR ("ptlrpc: run out of memory!\n" );return ERR_PTR (-ENOMEM );
drivers/staging/lustre/lustre/ptlrpc/events.c +1 -1
@@ -307,7 +307,7 @@ void request_in_callback(lnet_event_t *ev)/* We moaned above already... */ return ; } -OBD_ALLOC_GFP (req,sizeof (*req),ALLOC_ATOMIC_TRY ); + req =ptlrpc_request_cache_alloc (ALLOC_ATOMIC_TRY );if (req ==NULL ) { CERROR("Can't allocate incoming request descriptor: " "Dropping %s RPC from %s\n",
drivers/staging/lustre/lustre/ptlrpc/ptlrpc_internal.h +4 -0
@@ -55,6 +55,10 @@ int ptlrpcd_start(int index, int max, const char *name, struct ptlrpcd_ctl *pc);/* client.c */ struct ptlrpc_bulk_desc *ptlrpc_new_bulk (unsigned npages,unsigned max_brw,unsigned type,unsigned portal); +int ptlrpc_request_cache_init (void ); +void ptlrpc_request_cache_fini (void ); +struct ptlrpc_request *ptlrpc_request_cache_alloc (int flags); +void ptlrpc_request_cache_free (struct ptlrpc_request *req);void ptlrpc_init_xid (void );/* events.c */
drivers/staging/lustre/lustre/ptlrpc/ptlrpc_module.c +16 -8
@@ -73,29 +73,34 @@ __init int ptlrpc_init(void)return rc; cleanup_phase =1 ; + rc =ptlrpc_request_cache_init (); +if (rc) +GOTO (cleanup, rc); + cleanup_phase =2 ; rc =ptlrpc_init_portals ();if (rc)GOTO (cleanup, rc); - cleanup_phase =2 ; + + cleanup_phase =3 ; rc =ptlrpc_connection_init ();if (rc)GOTO (cleanup, rc); - cleanup_phase =3 ; + cleanup_phase =4 ; ptlrpc_put_connection_superhack = ptlrpc_connection_put; rc =ptlrpc_start_pinger ();if (rc)GOTO (cleanup, rc); - cleanup_phase =4 ; + cleanup_phase =5 ; rc =ldlm_init ();if (rc)GOTO (cleanup, rc); - cleanup_phase =5 ; + cleanup_phase =6 ; rc =sptlrpc_init ();if (rc)GOTO (cleanup, rc);
@@ -117,14 +122,16 @@ cleanup:ptlrpc_nrs_fini ();case 7 :sptlrpc_fini (); -case 5 : +case 6 :ldlm_exit (); -case 4 : +case 5 :ptlrpc_stop_pinger (); -case 3 : +case 4 :ptlrpc_connection_fini (); -case 2 : +case 3 :ptlrpc_exit_portals (); +case 2 : +ptlrpc_request_cache_fini ();case 1 :ptlrpc_hr_fini ();req_layout_fini ();
@@ -142,6 +149,7 @@ static void __exit ptlrpc_exit(void)ldlm_exit ();ptlrpc_stop_pinger ();ptlrpc_exit_portals (); +ptlrpc_request_cache_fini ();ptlrpc_hr_fini ();ptlrpc_connection_fini (); }
drivers/staging/lustre/lustre/ptlrpc/sec.c +3 -3
@@ -904,7 +904,7 @@ int sptlrpc_import_check_ctx(struct obd_import *imp)return -EACCES ; } -OBD_ALLOC_PTR (req); + req =ptlrpc_request_cache_alloc (__GFP_IO);if (!req)return -ENOMEM ;
@@ -920,7 +920,7 @@ int sptlrpc_import_check_ctx(struct obd_import *imp) rc =sptlrpc_req_refresh_ctx (req,0 );LASSERT (list_empty (&req->rq_ctx_chain));sptlrpc_cli_ctx_put (req->rq_cli_ctx,1 ); -OBD_FREE_PTR (req); +ptlrpc_request_cache_free (req);return rc; }
@@ -1177,7 +1177,7 @@ void sptlrpc_cli_finish_early_reply(struct ptlrpc_request *early_req)sptlrpc_cli_ctx_put (early_req->rq_cli_ctx,1 );OBD_FREE_LARGE (early_req->rq_repbuf, early_req->rq_repbuf_len); -OBD_FREE_PTR (early_req); +ptlrpc_request_cache_free (early_req); }/**************************************************
drivers/staging/lustre/lustre/ptlrpc/service.c +6 -7
@@ -842,7 +842,7 @@ static void ptlrpc_server_free_request(struct ptlrpc_request *req)/* NB request buffers use an embedded * req if the incoming req unlinked the * MD; this isn't one of them! */ -OBD_FREE (req,sizeof (*req)); +ptlrpc_request_cache_free (req); } }
@@ -1306,14 +1306,12 @@ static int ptlrpc_at_send_early_reply(struct ptlrpc_request *req) } newdl =cfs_time_current_sec () +at_get (&svcpt->scp_at_estimate); -OBD_ALLOC (reqcopy,sizeof (*reqcopy)); + reqcopy =ptlrpc_request_cache_alloc (__GFP_IO);if (reqcopy ==NULL )return -ENOMEM ;OBD_ALLOC_LARGE (reqmsg, req->rq_reqlen); -if (!reqmsg) { -OBD_FREE (reqcopy,sizeof (*reqcopy)); -return -ENOMEM ; - } +if (!reqmsg) +GOTO (out_free, rc = -ENOMEM ); *reqcopy = *req; reqcopy->rq_reply_state =NULL ;
@@ -1370,7 +1368,8 @@ out_put: out:sptlrpc_svc_ctx_decref (reqcopy);OBD_FREE_LARGE (reqmsg, req->rq_reqlen); -OBD_FREE (reqcopy,sizeof (*reqcopy)); +out_free: +ptlrpc_request_cache_free (reqcopy);return rc; }
Signature · 1 line
1.7.9.5
PATCH 6/39 staging/lustre/lnet: build Lustre with Intel OFED for Xeon Phi
In reply to Peng TaoFrom: Dmitry Eremin <[email protected]>
The compilation of Lustre with Intel OFED for Xeon Phi failed because of missing include of pci-dma.h header file. In Linux kernel includes this file included automatically from pci.h but in Intel Xeon Phi headers it's not true. So, just add this include explicitly.
Lustre-change: http://review.whamcloud.com/7115
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3637
--- .../staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h | 2 ++ 1 file changed, 2 insertions(+)
drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h +2 -0
@@ -38,6 +38,7 @@* Author: Eric Barton <[email protected]> */ +#include <linux/version.h> #include <linux/module.h> #include <linux/kernel.h> #include <linux/mm.h>
@@ -57,6 +58,7 @@#include <linux/kmod.h> #include <linux/sysctl.h> #include <linux/pci.h> +#include <linux/pci-dma.h> #include <net/sock.h> #include <linux/in.h>
Signature · 1 line
1.7.9.5
PATCH 7/39 staging/lustre/osd-ldiskfs: remove dependency on mdd module
In reply to Peng TaoFrom: Andreas Dilger <[email protected]>
Move the lu_capainfo_get() function (formerly named md_capainfo()) into obdclass/capa.c instead of mdd_device.c. Otherwise, the osd-ldiskfs.ko module depends on mdd.ko in order to load, and that doesn't really make sense.
Move the lu_capainfo structure (formerly named md_capainfo) into lustre_capa.h, and change the structure field prefix to match the new name. Fix up the users of this structure/function, and do some code style cleanup in those functions at the same time.
Move the mdt_set_capainfo() and mdt_dump_capainfo() structures from being static inline in mdt_internal.h into mdt_lib.c, since they are too big to be static inline functions. mdt_dump_capainfo() isn't used anywhere, so it is put under #ifdef DEBUG_CAPA. They should really be named mdt_capainfo_{set,dump}(), but that was too big a change for this patch.
Lustre-change: http://review.whamcloud.com/6945
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3556
--- .../staging/lustre/lustre/include/lustre_capa.h | 25 ++++++++++-- drivers/staging/lustre/lustre/include/md_object.h | 20 ---------- drivers/staging/lustre/lustre/obdclass/capa.c | 41 ++++++++++++++++++++ drivers/staging/lustre/lustre/obdclass/class_obd.c | 5 +++ 4 files changed, 67 insertions(+), 24 deletions(-)
drivers/staging/lustre/lustre/include/lustre_capa.h +21 -4
@@ -292,14 +292,31 @@ struct filter_capa_key { struct lustre_capa_key k_key; }; -enum { -LC_ID_NONE =0 , -LC_ID_PLAIN =1 , -LC_ID_CONVERT =2 +enum lc_auth_id { +LC_ID_NONE =0 , +LC_ID_PLAIN =1 , +LC_ID_CONVERT =2 };#define BYPASS_CAPA (struct lustre_capa *)ERR_PTR(-ENOENT) +enum { +LU_CAPAINFO_MAX =5 +}; + +/** there are at most 5 FIDs in one operation, see rename, +* NOTE the last one is a temporary one used for is_subdir() */ +struct lu_capainfo { +enum lc_auth_id lci_auth; +__u32 lci_padding; +struct lu_fid lci_fid[LU_CAPAINFO_MAX ]; +struct lustre_capa *lci_capa[LU_CAPAINFO_MAX ]; +}; + +int lu_capainfo_init (void ); +void lu_capainfo_fini (void ); +struct lu_capainfo *lu_capainfo_get (const struct lu_env *env); +/** @} capa */ #endif /* __LINUX_CAPA_H_ */
drivers/staging/lustre/lustre/include/md_object.h +0 -20
@@ -70,30 +70,10 @@ enum { UCRED_NEW = 2 }; -enum { -MD_CAPAINFO_MAX =5 -}; - -/** there are at most 5 fids in one operation, see rename, NOTE the last one -* is a temporary one used for is_subdir() */ -struct md_capainfo { -__u32 mc_auth; -__u32 mc_padding; -struct lu_fid mc_fid[MD_CAPAINFO_MAX ]; -struct lustre_capa *mc_capa[MD_CAPAINFO_MAX ]; -}; -struct md_quota {struct obd_export *mq_exp; }; -/** -* Implemented in mdd/mdd_handler.c. -* -* XXX should be moved into separate .h/.c together with all md security -* related definitions. -*/ -struct md_capainfo *md_capainfo (const struct lu_env *env);struct md_quota *md_quota (const struct lu_env *env);/** metadata attributes */
drivers/staging/lustre/lustre/obdclass/capa.c +41 -0
44 lines not on this page · raw message
drivers/staging/lustre/lustre/obdclass/class_obd.c +5 -0
@@ -572,6 +572,10 @@ static int __init init_obdclass(void)if (err)return err; + err =lu_capainfo_init (); +if (err) +return err; + err =cl_global_init ();if (err !=0 )return err;
@@ -651,6 +655,7 @@ static void cleanup_obdclass(void) }llog_info_fini ();cl_global_fini (); +lu_capainfo_fini ();lu_global_fini ();obd_cleanup_caches ();
Signature · 1 line
1.7.9.5
PATCH 8/39 staging/lustre/libcfs: improve validity test for valid file descriptors
In reply to Peng TaoFrom: Bob Glossman <[email protected]>
File descriptor members of struct lustre_kernelcomm aren't set to anything special at start time.
This leads to incorrect tests and skipping valid fd closes at stop time. Fixed by defining, setting, and testing against a specific LK_NOFD value.
Lustre-change: http://review.whamcloud.com/7275
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3683
--- .../include/linux/libcfs/libcfs_kernelcomm.h | 1 + .../lustre/lustre/libcfs/kernel_user_comm.c | 10 ++++++++-- 2 files changed, 9 insertions(+), 2 deletions(-)
drivers/staging/lustre/include/linux/libcfs/libcfs_kernelcomm.h +1 -0
@@ -97,6 +97,7 @@ extern int libcfs_kkuc_group_foreach(int group, libcfs_kkuc_cb_t cb_func, void *cb_arg);#define LK_FLG_STOP 0x01 +#define LK_NOFD -1U/* kernelcomm control structure, passed from userspace to kernel */ typedef struct lustre_kernelcomm {
drivers/staging/lustre/lustre/libcfs/kernel_user_comm.c +8 -2
@@ -56,6 +56,8 @@ int libcfs_ukuc_start(lustre_kernelcomm *link, int group) {int pfd[2 ]; + link->lk_rfd = link->lk_wfd =LK_NOFD ; +if (pipe (pfd) <0 )return -errno;
@@ -69,9 +71,13 @@ int libcfs_ukuc_start(lustre_kernelcomm *link, int group)int libcfs_ukuc_stop (lustre_kernelcomm *link) { -if (link->lk_wfd >0 ) +int rc; + +if (link->lk_wfd !=LK_NOFD )close (link->lk_wfd); -return close (link->lk_rfd); + rc =close (link->lk_rfd); + link->lk_rfd = link->lk_wfd =LK_NOFD ; +return rc; }#define lhsz sizeof(*kuch)
Signature · 1 line
1.7.9.5
PATCH 9/39 staging/lustre/llite: use 64bits flags in ll_lov_setea()
In reply to Peng TaoFrom: Niu Yawei <[email protected]>
In ll_lov_setea(), setting MDS_OPEN_HAS_OBJS to an int flags will result in the flags being overflowed.
Lustre-change: http://review.whamcloud.com/7312
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3744
--- drivers/staging/lustre/lustre/llite/file.c | 7 ++++--- .../staging/lustre/lustre/llite/llite_internal.h | 2 +- drivers/staging/lustre/lustre/llite/xattr.c | 6 +++--- 3 files changed, 8 insertions(+), 7 deletions(-)
drivers/staging/lustre/lustre/llite/file.c +4 -3
@@ -1401,7 +1401,8 @@ static int ll_lov_recreate_fid(struct inode *inode, unsigned long arg) }int ll_lov_setstripe_ea_info (struct inode *inode,struct file *file, -int flags,struct lov_user_md *lum,int lum_size) +__u64 flags,struct lov_user_md *lum, +int lum_size) {struct lov_stripe_md *lsm =NULL ;struct lookup_intent oit = {.it_op =IT_OPEN , .it_flags = flags};
@@ -1522,7 +1523,7 @@ out:static int ll_lov_setea (struct inode *inode,struct file *file,unsigned long arg) { -int flags =MDS_OPEN_HAS_OBJS |FMODE_WRITE ; +__u64 flags =MDS_OPEN_HAS_OBJS |FMODE_WRITE ;struct lov_user_md *lump;int lum_size =sizeof (struct lov_user_md ) +sizeof (struct lov_user_ost_data );
@@ -1554,7 +1555,7 @@ static int ll_lov_setstripe(struct inode *inode, struct file *file,struct lov_user_md_v1 *lumv1p = (struct lov_user_md_v1 *)arg;struct lov_user_md_v3 *lumv3p = (struct lov_user_md_v3 *)arg;int lum_size, rc; -int flags =FMODE_WRITE ; +__u64 flags =FMODE_WRITE ;/* first try with v1 which is smaller than v3 */ lum_size =sizeof (struct lov_user_md_v1 );
drivers/staging/lustre/lustre/llite/llite_internal.h +1 -1
@@ -790,7 +790,7 @@ struct posix_acl * ll_get_acl(struct inode *inode, int type);int ll_inode_permission (struct inode *inode,int mask);int ll_lov_setstripe_ea_info (struct inode *inode,struct file *file, -int flags,struct lov_user_md *lum, +__u64 flags,struct lov_user_md *lum,int lum_size); int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, struct lov_mds_md **lmm, int *lmm_size,
drivers/staging/lustre/lustre/llite/xattr.c +3 -3
@@ -236,13 +236,13 @@ int ll_setxattr(struct dentry *dentry, const char *name, lump->lmm_stripe_offset =-1 ;if (lump !=NULL &&S_ISREG (inode->i_mode)) { -struct file f; -int flags =FMODE_WRITE ; +struct file f; +__u64 it_flags =FMODE_WRITE ;int lum_size = (lump->lmm_magic ==LOV_USER_MAGIC_V1 ) ?sizeof (*lump) :sizeof (struct lov_user_md_v3 ); f.f_dentry = dentry; - rc =ll_lov_setstripe_ea_info (inode, &f, flags, lump, + rc =ll_lov_setstripe_ea_info (inode, &f, it_flags, lump, lum_size);/* b10667: rc always be 0 here for now */ rc =0 ;
Signature · 1 line
1.7.9.5
PATCH 10/39 staging/lustre/hsm: small fixes for HSM
In reply to Peng TaoFrom: Jinshan Xiong <[email protected]>
This patch includes the following fixes for HSM:
1. Redefine D_HSM. It was defined to D_TRACE;
2. Fix lu_ref for lease handle
3. User input verification to make sure the hur size is less than MDS_MAXREQSIZE;
Lustre-change: http://review.whamcloud.com/7243
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3647
--- .../lustre/include/linux/libcfs/libcfs_debug.h | 129 ++++++++++--------- drivers/staging/lustre/lustre/libcfs/debug.c | 131 ++------------------ drivers/staging/lustre/lustre/llite/dir.c | 5 + drivers/staging/lustre/lustre/llite/file.c | 4 +- drivers/staging/lustre/lustre/mdc/mdc_lib.c | 2 +- 5 files changed, 89 insertions(+), 182 deletions(-)
drivers/staging/lustre/include/linux/libcfs/libcfs_debug.h +69 -60
@@ -87,74 +87,83 @@ struct ptldebug_header {/* Debugging subsystems (32 bits, non-overlapping) */ /* keep these in sync with lnet/utils/debug.c and lnet/libcfs/debug.c */ -#define S_UNDEFINED 0x00000001 -#define S_MDC 0x00000002 -#define S_MDS 0x00000004 -#define S_OSC 0x00000008 -#define S_OST 0x00000010 -#define S_CLASS 0x00000020 -#define S_LOG 0x00000040 -#define S_LLITE 0x00000080 -#define S_RPC 0x00000100 -#define S_MGMT 0x00000200 -#define S_LNET 0x00000400 -#define S_LND 0x00000800/* ALL LNDs */ -#define S_PINGER 0x00001000 -#define S_FILTER 0x00002000 +#define S_UNDEFINED 0x00000001 +#define S_MDC 0x00000002 +#define S_MDS 0x00000004 +#define S_OSC 0x00000008 +#define S_OST 0x00000010 +#define S_CLASS 0x00000020 +#define S_LOG 0x00000040 +#define S_LLITE 0x00000080 +#define S_RPC 0x00000100 +#define S_MGMT 0x00000200 +#define S_LNET 0x00000400 +#define S_LND 0x00000800/* ALL LNDs */ +#define S_PINGER 0x00001000 +#define S_FILTER 0x00002000/* unused */ -#define S_ECHO 0x00008000 -#define S_LDLM 0x00010000 -#define S_LOV 0x00020000 -#define S_LQUOTA 0x00040000 +#define S_ECHO 0x00008000 +#define S_LDLM 0x00010000 +#define S_LOV 0x00020000 +#define S_LQUOTA 0x00040000#define S_OSD 0x00080000 +#define S_LFSCK 0x00100000/* unused */ /* unused */ +#define S_LMV 0x00800000/* b_new_cmd */ /* unused */ -#define S_LMV 0x00800000/* b_new_cmd */ +#define S_SEC 0x02000000/* upcall cache */ +#define S_GSS 0x04000000/* b_new_cmd */ /* unused */ -#define S_SEC 0x02000000/* upcall cache */ -#define S_GSS 0x04000000/* b_new_cmd */ -/* unused */ -#define S_MGC 0x10000000 -#define S_MGS 0x20000000 -#define S_FID 0x40000000/* b_new_cmd */ -#define S_FLD 0x80000000/* b_new_cmd */ -/* keep these in sync with lnet/utils/debug.c and lnet/libcfs/debug.c */ +#define S_MGC 0x10000000 +#define S_MGS 0x20000000 +#define S_FID 0x40000000/* b_new_cmd */ +#define S_FLD 0x80000000/* b_new_cmd */ + +#define LIBCFS_DEBUG_SUBSYS_NAMES { \ + "undefined", "mdc", "mds", "osc", "ost", "class", "log", \ + "llite", "rpc", "mgmt", "lnet", "lnd", "pinger", "filter", "", \ + "echo", "ldlm", "lov", "lquota", "osd", "lfsck", "", "", "lmv", \ + "", "sec", "gss", "", "mgc", "mgs", "fid", "fld", NULL }/* Debugging masks (32 bits, non-overlapping) */ -/* keep these in sync with lnet/utils/debug.c and lnet/libcfs/debug.c */ -#define D_TRACE 0x00000001/* ENTRY/EXIT markers */ -#define D_INODE 0x00000002 -#define D_SUPER 0x00000004 -#define D_EXT2 0x00000008/* anything from ext2_debug */ -#define D_MALLOC 0x00000010/* print malloc, free information */ -#define D_CACHE 0x00000020/* cache-related items */ -#define D_INFO 0x00000040/* general information */ -#define D_IOCTL 0x00000080/* ioctl related information */ -#define D_NETERROR 0x00000100/* network errors */ -#define D_NET 0x00000200/* network communications */ -#define D_WARNING 0x00000400/* CWARN(...) == CDEBUG (D_WARNING, ...) */ -#define D_BUFFS 0x00000800 -#define D_OTHER 0x00001000 -#define D_DENTRY 0x00002000 -#define D_NETTRACE 0x00004000 -#define D_PAGE 0x00008000/* bulk page handling */ -#define D_DLMTRACE 0x00010000 -#define D_ERROR 0x00020000/* CERROR(...) == CDEBUG (D_ERROR, ...) */ -#define D_EMERG 0x00040000/* CEMERG(...) == CDEBUG (D_EMERG, ...) */ -#define D_HA 0x00080000/* recovery and failover */ -#define D_RPCTRACE 0x00100000/* for distributed debugging */ -#define D_VFSTRACE 0x00200000 -#define D_READA 0x00400000/* read-ahead */ -#define D_MMAP 0x00800000 -#define D_CONFIG 0x01000000 -#define D_CONSOLE 0x02000000 -#define D_QUOTA 0x04000000 -#define D_SEC 0x08000000 -#define D_LFSCK 0x10000000/* For both OI scrub and LFSCK */ -/* keep these in sync with lnet/{utils,libcfs}/debug.c */ - -#define D_HSM D_TRACE +#define D_TRACE 0x00000001/* ENTRY/EXIT markers */ +#define D_INODE 0x00000002 +#define D_SUPER 0x00000004 +#define D_EXT2 0x00000008/* anything from ext2_debug */ +#define D_MALLOC 0x00000010/* print malloc, free information */ +#define D_CACHE 0x00000020/* cache-related items */ +#define D_INFO 0x00000040/* general information */ +#define D_IOCTL 0x00000080/* ioctl related information */ +#define D_NETERROR 0x00000100/* network errors */ +#define D_NET 0x00000200/* network communications */ +#define D_WARNING 0x00000400/* CWARN(...) == CDEBUG (D_WARNING, ...) */ +#define D_BUFFS 0x00000800 +#define D_OTHER 0x00001000 +#define D_DENTRY 0x00002000 +#define D_NETTRACE 0x00004000 +#define D_PAGE 0x00008000/* bulk page handling */ +#define D_DLMTRACE 0x00010000 +#define D_ERROR 0x00020000/* CERROR(...) == CDEBUG (D_ERROR, ...) */ +#define D_EMERG 0x00040000/* CEMERG(...) == CDEBUG (D_EMERG, ...) */ +#define D_HA 0x00080000/* recovery and failover */ +#define D_RPCTRACE 0x00100000/* for distributed debugging */ +#define D_VFSTRACE 0x00200000 +#define D_READA 0x00400000/* read-ahead */ +#define D_MMAP 0x00800000 +#define D_CONFIG 0x01000000 +#define D_CONSOLE 0x02000000 +#define D_QUOTA 0x04000000 +#define D_SEC 0x08000000 +#define D_LFSCK 0x10000000/* For both OI scrub and LFSCK */ +#define D_HSM 0x20000000 + +#define LIBCFS_DEBUG_MASKS_NAMES { \ + "trace", "inode", "super", "ext2", "malloc", "cache", "info", \ + "ioctl", "neterror", "net", "warning", "buffs", "other", \ + "dentry", "nettrace", "page", "dlmtrace", "error", "emerg", \ + "ha", "rpctrace", "vfstrace", "reada", "mmap", "config", \ + "console", "quota", "sec", "lfsck", "hsm", NULL }#define D_CANTMASK (D_ERROR | D_EMERG | D_WARNING | D_CONSOLE)
drivers/staging/lustre/lustre/libcfs/debug.c +12 -119
144 lines not on this page · raw message
drivers/staging/lustre/lustre/llite/dir.c +5 -0
@@ -1804,6 +1804,11 @@ out_rmdir:/* Compute the whole struct size */ totalsize =hur_len (hur);OBD_FREE_PTR (hur); + +/* Make sure the size is reasonable */ +if (totalsize >=MDS_MAXREQSIZE ) +return -E2BIG ; +OBD_ALLOC_LARGE (hur, totalsize);if (hur ==NULL )return -ENOMEM ;
drivers/staging/lustre/lustre/llite/file.c +2 -2
@@ -881,7 +881,7 @@ int ll_lease_close(struct obd_client_handle *och, struct inode *inode,lock_res_and_lock (lock); cancelled =ldlm_is_cancel (lock);unlock_res_and_lock (lock); -ldlm_lock_put (lock); +LDLM_LOCK_PUT (lock); } CDEBUG(D_INODE, "lease for "DFID" broken? %d\n",
@@ -2478,7 +2478,7 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) rc = och->och_flags & (FMODE_READ |FMODE_WRITE );unlock_res_and_lock (lock); -ldlm_lock_put (lock); +LDLM_LOCK_PUT (lock); } }mutex_unlock (&lli->lli_och_mutex);
drivers/staging/lustre/lustre/mdc/mdc_lib.c +1 -1
@@ -504,7 +504,7 @@ static void mdc_hsm_release_pack(struct ptlrpc_request *req, lock =ldlm_handle2lock (&op_data->op_lease_handle);if (lock !=NULL ) { data->cd_handle = lock->l_remote_handle; -ldlm_lock_put (lock); +LDLM_LOCK_PUT (lock); }ldlm_cli_cancel (&op_data->op_lease_handle,LCF_LOCAL );
Signature · 1 line
1.7.9.5
PATCH 11/39 staging/lustre/mdt: CDT cleanup follow-on patch
In reply to Peng TaoFrom: JC Lafoucriere <[email protected]>
This patch implements the comments made during the review after maloo passed test or after the landing.
rename in hai_zero() hai_first()
Use a better name for hai helper
replace use of cfs_helpers by native one in CDT
replace atomic_t and helpers
replace cfs_list and helpers
change list_for_each + list_entry by list_entry_for_each replace cfs_task_t
clean err msg in HSM agent interface
remove duplicate rc print
parenthesis code cleaning
remove useless () in conditionnal tests
Lustre-change: http://review.whamcloud.com/7254
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3709
--- .../lustre/lustre/include/lustre/lustre_user.h | 4 ++-- drivers/staging/lustre/lustre/mdc/mdc_request.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-)
drivers/staging/lustre/lustre/include/lustre/lustre_user.h +2 -2
@@ -1106,7 +1106,7 @@ static inline int cfs_size_round (int val) #endif /* Return pointer to first hai in action list */ -static inline struct hsm_action_item * hai_zero(struct hsm_action_list *hal) +static inline struct hsm_action_item *hai_first(struct hsm_action_list *hal) { return (struct hsm_action_item *)(hal->hal_fsname + cfs_size_round(strlen(hal-> \
@@ -1126,7 +1126,7 @@ static inline int hal_size(struct hsm_action_list *hal)struct hsm_action_item *hai; sz =sizeof (*hal) +cfs_size_round (strlen (hal->hal_fsname)); - hai =hai_zero (hal); + hai =hai_first (hal);for (i =0 ; i < hal->hal_count ; i++, hai =hai_next (hai)) sz +=cfs_size_round (hai->hai_len);
drivers/staging/lustre/lustre/mdc/mdc_request.c +1 -1
@@ -1957,7 +1957,7 @@ static void lustre_swab_hal(struct hsm_action_list *h)__swab32s (&h->hal_count);__swab32s (&h->hal_archive_id);__swab64s (&h->hal_flags); - hai =hai_zero (h); + hai =hai_first (h);for (i =0 ; i < h->hal_count; i++, hai =hai_next (hai))lustre_swab_hai (hai); }
Signature · 1 line
1.7.9.5
PATCH 12/39 staging/lustre/utils: Posix copytool fixes.
In reply to Peng TaoFrom: Henri Doreau <[email protected]>
Fixed several defects in the HSM posix copytool.
- Fixed two invalid format strings.
- Made --daemon optional.
- Remove .lov entries from the archive when removing the corresponding files.
- When restoring a file for which no LOV EA has been saved we must not create the volatile with O_LOV_DELAY_CREATE. Additionally, the caller should be able to specify the MDT index where to create the volatile file. Added parameters to llapi_hsm_action_begin() accordingly.
Lustre-change: http://review.whamcloud.com/7237
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3693
--- .../lustre/lustre/include/lustre/lustreapi.h | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/staging/lustre/lustre/include/lustre/lustreapi.h +4 -3
@@ -307,11 +307,12 @@ extern int llapi_hsm_copytool_unregister(struct hsm_copytool_private **priv);extern int llapi_hsm_copytool_recv (struct hsm_copytool_private *priv,struct hsm_action_list **hal,int *msgsize);extern void llapi_hsm_action_list_free (struct hsm_action_list **hal); -extern int llapi_hsm_action_begin (struct hsm_copyaction_private **hcp, -const struct hsm_copytool_private *ct_priv, +extern int llapi_hsm_action_begin (struct hsm_copyaction_private **phcp, +const struct hsm_copytool_private *ct,const struct hsm_action_item *hai, +int restore_mdt_index,int restore_open_flags,bool is_error); -extern int llapi_hsm_action_end (struct hsm_copyaction_private **hcp, +extern int llapi_hsm_action_end (struct hsm_copyaction_private **phcp,const struct hsm_extent *he,int flags,int errval); extern int llapi_hsm_action_progress(struct hsm_copyaction_private *hcp,
Signature · 1 line
1.7.9.5
PATCH 13/39 staging/lustre/lmv: update coding style
In reply to Peng TaoFrom: JC Lafoucriere <[email protected]>
in lmv_obd.c:
- Correct wrong coding styles introduced by HSM (and others)
- Remove useless include
Lustre-change: http://review.whamcloud.com/7252
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3709
--- drivers/staging/lustre/lustre/lmv/lmv_obd.c | 42 +++++++++++++-------------- 1 file changed, 21 insertions(+), 21 deletions(-)
drivers/staging/lustre/lustre/lmv/lmv_obd.c +21 -21
@@ -43,7 +43,6 @@#include <asm/div64.h> #include <linux/seq_file.h> #include <linux/namei.h> -#include <asm/uaccess.h> #include <lustre/lustre_idl.h> #include <obd_support.h>
@@ -802,10 +801,10 @@ static void lmv_hsm_req_build(struct lmv_obd *lmv, nr_out =0 ;for (i =0 ; i < hur_in->hur_request.hr_itemcount; i++) { curr_tgt =lmv_find_target (lmv, - &hur_in->hur_user_item[i].hui_fid); + &hur_in->hur_user_item[i].hui_fid);if (obd_uuid_equals (&curr_tgt->ltd_uuid, &tgt_mds->ltd_uuid)) { hur_out->hur_user_item[nr_out] = - hur_in->hur_user_item[i]; + hur_in->hur_user_item[i]; nr_out++; } }
@@ -851,17 +850,18 @@ static int lmv_hsm_ct_register(struct lmv_obd *lmv, unsigned int cmd, int len,if (err) {if (lmv->tgts[i]->ltd_active) {/* permanent error */ -CERROR ("error: iocontrol MDC %s on MDT" -"idx %d cmd %x: err = %d\n" , - lmv->tgts[i]->ltd_uuid.uuid, - i, cmd, err); +CERROR ("%s: iocontrol MDC %s on MDT" +" idx %d cmd %x: err = %d\n" , +class_exp2obd (lmv->exp)->obd_name, + lmv->tgts[i]->ltd_uuid.uuid, + i, cmd, err); rc = err; lk->lk_flags |=LK_FLG_STOP ;/* unregister from previous MDS */ for (j =0 ; j < i; j++)obd_iocontrol (cmd, - lmv->tgts[j]->ltd_exp, - len, lk, uarg); + lmv->tgts[j]->ltd_exp, + len, lk, uarg);return rc; }/* else: transient error.
@@ -924,8 +924,8 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp,/* copy UUID */ if (copy_to_user (data->ioc_pbuf2,obd2cli_tgt (mdc_obd), -min ((int ) data->ioc_plen2, - (int )sizeof (struct obd_uuid )))) +min ((int ) data->ioc_plen2, + (int )sizeof (struct obd_uuid ))))return -EFAULT ; rc = obd_statfs(NULL, lmv->tgts[index]->ltd_exp, &stat_buf,
@@ -934,8 +934,8 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp,if (rc)return rc;if (copy_to_user (data->ioc_pbuf1, &stat_buf, -min ((int ) data->ioc_plen1, - (int )sizeof (stat_buf)))) +min ((int ) data->ioc_plen1, + (int )sizeof (stat_buf))))return -EFAULT ;break ; }
@@ -1020,10 +1020,10 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp, tgt =lmv_find_target (lmv, &op_data->op_fid1);if (IS_ERR (tgt)) -return PTR_ERR (tgt); +return PTR_ERR (tgt);if (tgt->ltd_exp ==NULL ) -return -EINVAL ; +return -EINVAL ; rc =obd_iocontrol (cmd, tgt->ltd_exp, len, karg, uarg);break ;
@@ -1069,7 +1069,7 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp,/* build a request with fids for this MDS */ reqlen = offsetof(typeof(*hur), hur_user_item[nr]) - + hur->hur_request.hr_data_len; + + hur->hur_request.hr_data_len;OBD_ALLOC_LARGE (req, reqlen);if (req ==NULL )return -ENOMEM ;
@@ -1134,9 +1134,9 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp, }else if (err) {if (lmv->tgts[i]->ltd_active) {CERROR ("error: iocontrol MDC %s on MDT" -"idx %d cmd %x: err = %d\n" , - lmv->tgts[i]->ltd_uuid.uuid, - i, cmd, err); +" idx %d cmd %x: err = %d\n" , + lmv->tgts[i]->ltd_uuid.uuid, + i, cmd, err);if (!rc) rc = err; }
@@ -1342,8 +1342,8 @@ static int lmv_setup(struct obd_device *obd, struct lustre_cfg *lcfg) 0444, &lmv_proc_target_fops, obd);if (rc)CWARN ("%s: error adding LMV target_obd file: rc = %d\n" , - obd->obd_name, rc); - } + obd->obd_name, rc); + }#endif rc =fld_client_init (&lmv->lmv_fld, obd->obd_name,LUSTRE_CLI_FLD_HASH_DHT );
Signature · 1 line
1.7.9.5
PATCH 14/39 staging/lustre/hsm: copy start error should set HP_FLAG_COMPLETED
In reply to Peng TaoFrom: Aurelien Degremont <[email protected]>
If an error occurs when initializing a HSM request, in ll_ioc_copy_start(), the PROGRESS message, sent to coordinator, should carry the error code but also HP_FLAG_COMPLETED to mark the request as finished (with error). If not, the Coordinator will ignore this message and consider the request is still running.
Lustre-change: http://review.whamcloud.com/7265
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3685
--- drivers/staging/lustre/lustre/llite/dir.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/staging/lustre/lustre/llite/dir.c +3 -0
@@ -942,6 +942,9 @@ static int ll_ioc_copy_start(struct super_block *sb, struct hsm_copy *copy) } progress: +/* On error, the request should be considered as completed */ +if (hpk.hpk_errval >0 ) + hpk.hpk_flags |=HP_FLAG_COMPLETED ; rc =obd_iocontrol (LL_IOC_HSM_PROGRESS , sbi->ll_md_exp,sizeof (hpk), &hpk,NULL );
Signature · 1 line
1.7.9.5
PATCH 15/39 staging/lustre/lov: Get the correct address of lmm_objects
In reply to Peng TaoFrom: Swapnil Pimpale <[email protected]>
The introduction of lmm_layout_gen makes the assumption that lmm_objects is present after lmm_stripe_count incorrect. Fixed this to get the correct address of lmm_objects when lmmk is cast to lov_mds_md_v1.
Lustre-change: http://review.whamcloud.com/7258
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3162
--- drivers/staging/lustre/lustre/lov/lov_pack.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/staging/lustre/lustre/lov/lov_pack.c +1 -2
@@ -641,8 +641,7 @@ int lov_getstripe(struct obd_export *exp, struct lov_stripe_md *lsm,if (lum.lmm_magic ==LOV_USER_MAGIC ) {/* User request for v1, we need skip lmm_pool_name */ if (lmmk->lmm_magic ==LOV_MAGIC_V3 ) { -memmove ((char *)(&lmmk->lmm_stripe_count) + -sizeof (lmmk->lmm_stripe_count), +memmove (((struct lov_mds_md_v1 *)lmmk)->lmm_objects, ((struct lov_mds_md_v3 *)lmmk)->lmm_objects, lmmk->lmm_stripe_count *sizeof (struct lov_ost_data_v1 ));
Signature · 1 line
1.7.9.5
PATCH 16/39 staging/lustre/lvfs: remove llog_lvfs.c and other lvfs code from llog
In reply to Peng TaoFrom: James Simmons <[email protected]>
- remove obdfilter/llog_lvfs.c
- remove lgh_file from llog_header
Lustre-change: http://review.whamcloud.com/5331
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2158
--- .../lustre/lustre/include/linux/lustre_fsfilt.h | 84 -- drivers/staging/lustre/lustre/include/linux/lvfs.h | 5 - drivers/staging/lustre/lustre/include/lustre_log.h | 4 - drivers/staging/lustre/lustre/include/obd_class.h | 10 - drivers/staging/lustre/lustre/lvfs/fsfilt_ext3.c | 760 ------------------ drivers/staging/lustre/lustre/obdclass/llog_cat.c | 34 +- drivers/staging/lustre/lustre/obdclass/llog_lvfs.c | 847 -------------------- 7 files changed, 13 insertions(+), 1731 deletions(-) delete mode 100644 drivers/staging/lustre/lustre/lvfs/fsfilt_ext3.c delete mode 100644 drivers/staging/lustre/lustre/obdclass/llog_lvfs.c
drivers/staging/lustre/lustre/include/linux/lustre_fsfilt.h +0 -84
@@ -49,24 +49,13 @@#include <obd.h> #include <obd_class.h> -typedef void (*fsfilt_cb_t )(struct obd_device *obd,__u64 last_rcvd, -void *data,int error); -struct fsfilt_operations {struct list_head fs_list;struct module *fs_owner;char *fs_type; -char *(* fs_getlabel)(struct super_block *sb); -void *(* fs_start)(struct inode *inode,int op,void *desc_private, -int logs); -int (* fs_commit)(struct inode *inode,void *handle,int force_sync);int (* fs_map_inode_pages)(struct inode *inode,struct page **page,int pages,unsigned long *blocks,int create,struct mutex *sem); -int (* fs_write_record)(struct file *,void *,int size,loff_t *, -int force_sync); -int (* fs_read_record)(struct file *,void *,int size,loff_t *); -int (* fs_setup)(struct super_block *sb); };extern int fsfilt_register_ops (struct fsfilt_operations *fs_ops);
@@ -74,20 +63,6 @@ extern void fsfilt_unregister_ops(struct fsfilt_operations *fs_ops);extern struct fsfilt_operations *fsfilt_get_ops (const char *type);extern void fsfilt_put_ops (struct fsfilt_operations *fs_ops); -static inline char *fsfilt_get_label (struct obd_device *obd, -struct super_block *sb) -{ -if (obd->obd_fsops->fs_getlabel ==NULL ) -return NULL ; -if (obd->obd_fsops->fs_getlabel (sb)[0 ] =='\0' ) -return NULL ; - -return obd->obd_fsops->fs_getlabel (sb); -} - -#define FSFILT_OP_UNLINK 1 -#define FSFILT_OP_CANCEL_UNLINK 10 -#define __fsfilt_check_slow (obd, start, msg) \ do { \ if (cfs_time_before(jiffies, start + 15 * HZ)) \
@@ -109,63 +84,4 @@ do { \ start = jiffies; \ } while (0) -static inline void *fsfilt_start_log(struct obd_device *obd, -struct inode *inode,int op, -struct obd_trans_info *oti,int logs) -{ -unsigned long now = jiffies; -void *parent_handle = oti ? oti->oti_handle :NULL ; -void *handle; - - handle = obd->obd_fsops->fs_start (inode, op, parent_handle, logs); -CDEBUG (D_INFO ,"started handle %p (%p)\n" , handle, parent_handle); - -if (oti !=NULL ) { -if (parent_handle ==NULL ) { - oti->oti_handle = handle; - }else if (handle != parent_handle) { -CERROR ("mismatch: parent %p, handle %p, oti %p\n" , - parent_handle, handle, oti); -LBUG (); - } - } -fsfilt_check_slow (obd, now,"journal start" ); -return handle; -} - -static inline int fsfilt_commit (struct obd_device *obd,struct inode *inode, -void *handle,int force_sync) -{ -unsigned long now = jiffies; -int rc = obd->obd_fsops->fs_commit (inode, handle, force_sync); -CDEBUG (D_INFO ,"committing handle %p\n" , handle); - -fsfilt_check_slow (obd, now,"journal start" ); - -return rc; -} - -static inline int fsfilt_read_record (struct obd_device *obd,struct file *file, -void *buf,loff_t size,loff_t *offs) -{ -return obd->obd_fsops->fs_read_record (file, buf, size, offs); -} - -static inline int fsfilt_write_record (struct obd_device *obd,struct file *file, -void *buf,loff_t size,loff_t *offs, -int force_sync) -{ -return obd->obd_fsops->fs_write_record (file, buf, size,offs,force_sync); -} - -static inline int fsfilt_setup (struct obd_device *obd,struct super_block *fs) -{ -if (obd->obd_fsops->fs_setup) -return obd->obd_fsops->fs_setup (fs); -return 0 ; -} - - - -#endif
drivers/staging/lustre/lustre/include/linux/lvfs.h +0 -5
@@ -64,10 +64,6 @@ struct lvfs_ucred {struct md_identity *luc_identity; }; -struct lvfs_callback_ops { -struct dentry *(*l_fid2dentry)(__u64 id_ino,__u32 gen,__u64 gr,void *data); -}; -#define OBD_RUN_CTXT_MAGIC 0xC0FFEEAA#define OBD_CTXT_DEBUG /* development-only debugging */ struct lvfs_run_ctxt {
@@ -76,7 +72,6 @@ struct lvfs_run_ctxt {mm_segment_t fs;struct lvfs_ucred luc;int ngroups; -struct lvfs_callback_ops cb_ops;struct group_info *group_info;struct dt_device *dt;#ifdef OBD_CTXT_DEBUG
drivers/staging/lustre/lustre/include/lustre_log.h +0 -4
@@ -309,7 +309,6 @@ struct llog_handle {spinlock_t lgh_hdr_lock;/* protect lgh_hdr data */ struct llog_logid lgh_id;/* id of this log */ struct llog_log_hdr *lgh_hdr; -struct file *lgh_file;struct dt_object *lgh_obj;int lgh_last_idx;int lgh_cur_idx;/* used during llog_process */
@@ -325,9 +324,6 @@ struct llog_handle {atomic_t lgh_refcount; }; -/* llog_lvfs.c */ -extern struct llog_operations llog_lvfs_ops; -/* llog_osd.c */ extern struct llog_operations llog_osd_ops; int llog_osd_get_cat_list(const struct lu_env *env, struct dt_device *d,
drivers/staging/lustre/lustre/include/obd_class.h +0 -10
@@ -1141,16 +1141,6 @@ static inline int obd_extent_calc(struct obd_export *exp,return rc; } -static inline struct dentry * -obd_lvfs_fid2dentry (struct obd_export *exp,struct ost_id *oi,__u32 gen) -{ -struct lvfs_run_ctxt *ctxt = &exp->exp_obd->obd_lvfs_ctxt; -LASSERT (exp->exp_obd); - -return ctxt->cb_ops.l_fid2dentry (ostid_id (oi), gen,ostid_seq (oi), - exp->exp_obd); -} -/* @max_age is the oldest time in jiffies that we accept using a cached data. * If the cache is older than @max_age we will get a new value from the * target. Use a value of "cfs_time_current() + HZ" to guarantee freshness. */
drivers/staging/lustre/lustre/lvfs/fsfilt_ext3.c deleted +0 -760
760 lines not on this page · raw message
drivers/staging/lustre/lustre/obdclass/llog_cat.c +13 -21
@@ -444,31 +444,23 @@ int llog_cat_add(const struct lu_env *env, struct llog_handle *cathandle,LASSERT (ctxt);LASSERT (ctxt->loc_exp); -if (cathandle->lgh_obj !=NULL ) { - dt = ctxt->loc_exp->exp_obd->obd_lvfs_ctxt.dt; -LASSERT (dt); +LASSERT (cathandle->lgh_obj !=NULL ); + dt =lu2dt_dev (cathandle->lgh_obj->do_lu.lo_dev); - th =dt_trans_create (env, dt); -if (IS_ERR (th)) -return PTR_ERR (th); + th =dt_trans_create (env, dt); +if (IS_ERR (th)) +return PTR_ERR (th); - rc =llog_cat_declare_add_rec (env, cathandle, rec, th); -if (rc) -GOTO (out_trans, rc); + rc =llog_cat_declare_add_rec (env, cathandle, rec, th); +if (rc) +GOTO (out_trans, rc); - rc =dt_trans_start_local (env, dt, th); -if (rc) -GOTO (out_trans, rc); - rc =llog_cat_add_rec (env, cathandle, rec, reccookie, buf, th); + rc =dt_trans_start_local (env, dt, th); +if (rc) +GOTO (out_trans, rc); + rc =llog_cat_add_rec (env, cathandle, rec, reccookie, buf, th); out_trans: -dt_trans_stop (env, dt, th); - }else {/* lvfs compat code */ -LASSERT (cathandle->lgh_file !=NULL ); - rc =llog_cat_declare_add_rec (env, cathandle, rec, th); -if (rc ==0 ) - rc =llog_cat_add_rec (env, cathandle, rec, reccookie, - buf, th); - } +dt_trans_stop (env, dt, th);return rc; }EXPORT_SYMBOL (llog_cat_add);
drivers/staging/lustre/lustre/obdclass/llog_lvfs.c deleted +0 -847
847 lines not on this page · raw message
Signature · 1 line
1.7.9.5
PATCH 17/39 staging/lustre/llite: reset writeback index in ll_writepages
In reply to Peng TaoFrom: Jinshan Xiong <[email protected]>
Otherwise after one round the writeback index will become beyond the file size and ->writepages() turns into an empty operation.
Also, a safety guard is added to limit the wait time for grant to 10 minutes(take recovery into consideration) at maximum in the osc_enter_cache() function. Otherwise, EDQUOT will be returned to the applications to start sync write.
Lustre-change: http://review.whamcloud.com/6554
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3416
--- drivers/staging/lustre/lustre/llite/rw.c | 5 ++-- drivers/staging/lustre/lustre/osc/osc_cache.c | 32 +++++++++++++++++-------- 2 files changed, 25 insertions(+), 12 deletions(-)
drivers/staging/lustre/lustre/llite/rw.c +3 -2
@@ -1262,8 +1262,9 @@ int ll_writepages(struct address_space *mapping, struct writeback_control *wbc)if (wbc->range_cyclic || (range_whole && wbc->nr_to_write >0 )) {if (end ==OBD_OBJECT_EOF ) - end =i_size_read (inode); - mapping->writeback_index = (end >>PAGE_CACHE_SHIFT ) +1 ; + mapping->writeback_index =0 ; +else + mapping->writeback_index = (end >>PAGE_CACHE_SHIFT ) +1 ; }return result; }
drivers/staging/lustre/lustre/osc/osc_cache.c +22 -10
@@ -1296,15 +1296,19 @@ static int osc_completion(const struct lu_env *env, struct osc_async_page *oap,return 0 ; } -#define OSC_DUMP_GRANT(cli, fmt, args...) do { \ +#define OSC_DUMP_GRANT(lvl, cli, fmt, args...) do { \ struct client_obd *__tmp = (cli); \ - CDEBUG(D_CACHE, "%s: { dirty: %ld/%ld dirty_pages: %d/%d " \ - "dropped: %ld avail: %ld, reserved: %ld, flight: %d } " fmt, \ + CDEBUG(lvl, "%s: grant { dirty: %ld/%ld dirty_pages: %d/%d " \ + "dropped: %ld avail: %ld, reserved: %ld, flight: %d } " \ + "lru {in list: %d, left: %d, waiters: %d }" fmt, \ __tmp->cl_import->imp_obd->obd_name, \ __tmp->cl_dirty, __tmp->cl_dirty_max, \ atomic_read(&obd_dirty_pages), obd_max_dirty_pages, \ __tmp->cl_lost_grant, __tmp->cl_avail_grant, \ - __tmp->cl_reserved_grant, __tmp->cl_w_in_flight, ##args); \ + __tmp->cl_reserved_grant, __tmp->cl_w_in_flight, \ + atomic_read(&__tmp->cl_lru_in_list), \ + atomic_read(&__tmp->cl_lru_busy), \ + atomic_read(&__tmp->cl_lru_shrinkers), ##args); \ } while (0)/* caller must hold loi_list_lock */
@@ -1441,7 +1445,7 @@ static int osc_enter_cache_try(struct client_obd *cli, {int rc; -OSC_DUMP_GRANT (cli,"need:%d.\n" , bytes); +OSC_DUMP_GRANT (D_CACHE , cli,"need:%d.\n" , bytes); rc =osc_reserve_grant (cli, bytes);if (rc <0 )
@@ -1485,10 +1489,11 @@ static int osc_enter_cache(const struct lu_env *env, struct client_obd *cli,struct osc_object *osc = oap->oap_obj;struct lov_oinfo *loi = osc->oo_oinfo;struct osc_cache_waiter ocw; -struct l_wait_info lwi =LWI_INTR (LWI_ON_SIGNAL_NOOP ,NULL ); +struct l_wait_info lwi =LWI_TIMEOUT_INTR (cfs_time_seconds (600 ),NULL , +LWI_ON_SIGNAL_NOOP ,NULL );int rc = -EDQUOT ; -OSC_DUMP_GRANT (cli,"need:%d.\n" , bytes); +OSC_DUMP_GRANT (D_CACHE , cli,"need:%d.\n" , bytes);client_obd_list_lock (&cli->cl_loi_list_lock);
@@ -1526,8 +1531,15 @@ static int osc_enter_cache(const struct lu_env *env, struct client_obd *cli,client_obd_list_lock (&cli->cl_loi_list_lock); -/* l_wait_event is interrupted by signal */ +/* l_wait_event is interrupted by signal, or timed out */ if (rc <0 ) { +if (rc == -ETIMEDOUT ) { +OSC_DUMP_GRANT (D_ERROR , cli, +"try to reserve %d.\n" , bytes); +osc_extent_tree_dump (D_ERROR , osc); + rc = -EDQUOT ; + } +list_del_init (&ocw.ocw_entry);GOTO (out, rc); }
@@ -1542,7 +1554,7 @@ static int osc_enter_cache(const struct lu_env *env, struct client_obd *cli, } out:client_obd_list_unlock (&cli->cl_loi_list_lock); -OSC_DUMP_GRANT (cli,"returned %d.\n" , rc); +OSC_DUMP_GRANT (D_CACHE , cli,"returned %d.\n" , rc);return rc; }
@@ -2868,7 +2880,7 @@ int osc_cache_writeback_range(const struct lu_env *env, struct osc_object *obj, result = rc; } -OSC_IO_DEBUG (obj,"cache page out.\n" ); +OSC_IO_DEBUG (obj,"pageout [%lu, %lu], %d.\n" , start, end, result);return result; }
Signature · 1 line
1.7.9.5
PATCH 18/39 staging/lustre/llite: Delaying creation of client side proc entries.
In reply to Peng TaoFrom: Gaurav Mahajan <[email protected]>
In client_common_fill_super() proc entries are created before before cl_sb_init() and therefore lu_site is not allocated resulting in client crash when tried reading lu_site stats.
Delaying creation of proc entries after creation of all required data structures fixed the problem.
Lustre-change: http://review.whamcloud.com/6852
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2941
--- drivers/staging/lustre/lustre/llite/llite_lib.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-)
drivers/staging/lustre/lustre/llite/llite_lib.c +9 -8
@@ -186,13 +186,6 @@ static int client_common_fill_super(struct super_block *sb, char *md, char *dt,return -ENOMEM ; } -if (proc_lustre_fs_root) { - err =lprocfs_register_mountpoint (proc_lustre_fs_root, sb, - dt, md); -if (err <0 ) -CERROR ("could not register mount in /proc/fs/lustre\n" ); - } -/* indicate the features supported by this client */ data->ocd_connect_flags =OBD_CONNECT_IBITS |OBD_CONNECT_NODEVOH | OBD_CONNECT_ATTRFID |
@@ -590,6 +583,15 @@ static int client_common_fill_super(struct super_block *sb, char *md, char *dt,OBD_FREE_PTR (data);if (osfs !=NULL )OBD_FREE_PTR (osfs); +if (proc_lustre_fs_root !=NULL ) { + err =lprocfs_register_mountpoint (proc_lustre_fs_root, sb, + dt, md); +if (err <0 ) { +CERROR ("%s: could not register mount in lprocfs: " +"rc = %d\n" ,ll_get_fsname (sb,NULL ,0 ), err); + err =0 ; + } + }return err; out_root:
@@ -612,7 +614,6 @@ out:OBD_FREE_PTR (data);if (osfs !=NULL )OBD_FREE_PTR (osfs); -lprocfs_unregister_mountpoint (sbi);return err; }
Signature · 1 line
1.7.9.5
PATCH 19/39 staging/lustre/libcfs: Add relocation function to libcfs heap
In reply to Peng TaoFrom: Li Xi <[email protected]>
When changing the values of the node in the heap which might affect its ranking, we have to remove the node from the heap, change the values and then insert the node into the heap again. The process of inserting into the heap might fail because of memory insufficiency. Sometimes, it is not convienient and efficient for a caller to handle such kind of failure.
This patch adds a new function to do the job.
Lustre-change: http://review.whamcloud.com/6743
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3494
--- .../lustre/include/linux/libcfs/libcfs_heap.h | 1 + drivers/staging/lustre/lustre/libcfs/heap.c | 18 ++++++++++++++++-- 2 files changed, 17 insertions(+), 2 deletions(-)
drivers/staging/lustre/include/linux/libcfs/libcfs_heap.h +1 -0
@@ -166,6 +166,7 @@ cfs_binheap_t *cfs_binheap_create(cfs_binheap_ops_t *ops, unsigned int flags,cfs_binheap_node_t *cfs_binheap_find (cfs_binheap_t *h,unsigned int idx);int cfs_binheap_insert (cfs_binheap_t *h,cfs_binheap_node_t *e);void cfs_binheap_remove (cfs_binheap_t *h,cfs_binheap_node_t *e); +void cfs_binheap_relocate (cfs_binheap_t *h,cfs_binheap_node_t *e);static inline int cfs_binheap_size (cfs_binheap_t *h)
drivers/staging/lustre/lustre/libcfs/heap.c +16 -2
@@ -463,8 +463,7 @@ cfs_binheap_remove(cfs_binheap_t *h, cfs_binheap_node_t *e) last->chn_index = cur_idx; *cur_ptr = last; -if (!cfs_binheap_bubble (h, *cur_ptr)) -cfs_binheap_sink (h, *cur_ptr); +cfs_binheap_relocate (h, *cur_ptr); e->chn_index =CBH_POISON ;if (h->cbh_ops->hop_exit)
@@ -472,4 +471,19 @@ cfs_binheap_remove(cfs_binheap_t *h, cfs_binheap_node_t *e) }EXPORT_SYMBOL (cfs_binheap_remove); +/** +* Relocate a node in the binary heap. +* Should be called whenever a node's values +* which affects its ranking are changed. +* +* \param[in] h The heap +* \param[in] e The node +*/ +void +cfs_binheap_relocate (cfs_binheap_t *h,cfs_binheap_node_t *e) +{ +if (!cfs_binheap_bubble (h, e)) +cfs_binheap_sink (h, e); +} +EXPORT_SYMBOL (cfs_binheap_relocate);/** @} heap */
Signature · 1 line
1.7.9.5
PATCH 20/39 staging/lustre/build: fix 'data race condition' issues
In reply to Peng TaoFrom: Sebastien Buisson <[email protected]>
Fix 'data race condition' defects found by Coverity version 6.5.0:
Data race condition (MISSING_LOCK)
Accessing variable without holding lock. Elsewhere, this variable is accessed with lock held.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2744
Lustre-change: http://review.whamcloud.com/6567
--- drivers/staging/lustre/lnet/selftest/conrpc.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/staging/lustre/lnet/selftest/conrpc.c +6 -2
@@ -938,8 +938,12 @@ lstcon_sesnew_stat_reply(lstcon_rpc_trans_t *trans,return status;if (!trans->tas_feats_updated) { - trans->tas_feats_updated =1 ; - trans->tas_features = reply->msg_ses_feats; +spin_lock (&console_session.ses_rpc_lock); +if (!trans->tas_feats_updated) {/* recheck with lock */ + trans->tas_feats_updated =1 ; + trans->tas_features = reply->msg_ses_feats; + } +spin_unlock (&console_session.ses_rpc_lock); }if (reply->msg_ses_feats != trans->tas_features) {
Signature · 1 line
1.7.9.5
PATCH 21/39 staging/lustre/autoconf: remove LC_LOCK_MAP_ACQUIRE test
In reply to Peng TaoFrom: James Simmons <[email protected]>
lock_map_acquire has been in the upstream kernel since v2.6.27. We can remove the test and dead code.
Lustre-change: http://review.whamcloud.com/5369
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2800
--- drivers/staging/lustre/lustre/obdclass/cl_lock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/lustre/lustre/obdclass/cl_lock.c +1 -1
@@ -162,7 +162,7 @@ static void cl_lock_lockdep_release(const struct lu_env *env, struct cl_lock *lock) {cl_lock_counters (env, lock)->ctc_nr_locks_acquired--; -lock_release (&lock->dep_map,0 ,RETIP ); +lock_map_release (&lock->dep_map); }#else /* !CONFIG_LOCKDEP */
Signature · 1 line
1.7.9.5
PATCH 22/39 staging/lustre/mdc: document mdc_rpc_lock
In reply to Peng TaoFrom: Ned Bass <[email protected]>
As this lock can be a bottleneck, clarifying why it is needed may be helpful to those working on client performance.
Lustre-change: http://review.whamcloud.com/6593
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3443
--- drivers/staging/lustre/lustre/include/lustre_mdc.h | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+)
drivers/staging/lustre/lustre/include/lustre_mdc.h +18 -0
@@ -67,9 +67,27 @@ struct obd_export;struct ptlrpc_request ;struct obd_device ; +/** +* Serializes in-flight MDT-modifying RPC requests to preserve idempotency. +* +* This mutex is used to implement execute-once semantics on the MDT. +* The MDT stores the last transaction ID and result for every client in +* its last_rcvd file. If the client doesn't get a reply, it can safely +* resend the request and the MDT will reconstruct the reply being aware +* that the request has already been executed. Without this lock, +* execution status of concurrent in-flight requests would be +* overwritten. +* +* This design limits the extent to which we can keep a full pipeline of +* in-flight requests from a single client. This limitation could be +* overcome by allowing multiple slots per client in the last_rcvd file. +*/ struct mdc_rpc_lock { +/** Lock protecting in-flight RPC concurrency. */ struct mutex rpcl_mutex; +/** Intent associated with currently executing request. */ struct lookup_intent *rpcl_it; +/** Used for MDS/RPC load testing purposes. */ int rpcl_fakes; };
Signature · 1 line
1.7.9.5
PATCH 23/39 staging/lustre/autoconf: remove LC_FS_STRUCT_USE_PATH
In reply to Peng TaoFrom: James Simmons <[email protected]>
struct fs_struct has been using struct path for working dirs since v2.6.26.
We can remove the test and the dead code.
Lustre-change: http://review.whamcloud.com/5359
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2800
--- .../lustre/lustre/include/linux/lustre_compat25.h | 5 ----- drivers/staging/lustre/lustre/lvfs/lvfs_linux.c | 14 +++++++------- drivers/staging/lustre/lustre/ptlrpc/sec_config.c | 2 +- 3 files changed, 8 insertions(+), 13 deletions(-)
drivers/staging/lustre/lustre/include/linux/lustre_compat25.h +0 -5
@@ -111,11 +111,6 @@ static inline void ll_set_fs_pwd(struct fs_struct *fs, struct vfsmount *mnt,#define cfs_bio_io_error (a,b) bio_io_error((a))#define cfs_bio_endio (a,b,c) bio_endio((a),(c)) -#define cfs_fs_pwd (fs) ((fs)->pwd.dentry) -#define cfs_fs_mnt (fs) ((fs)->pwd.mnt) -#define cfs_path_put (nd) path_put(&(nd)->path) - -#ifndef SLAB_DESTROY_BY_RCU #define SLAB_DESTROY_BY_RCU 0#endif
drivers/staging/lustre/lustre/lvfs/lvfs_linux.c +7 -7
@@ -119,10 +119,10 @@ void push_ctxt(struct lvfs_run_ctxt *save, struct lvfs_run_ctxt *new_ctx,OBD_SET_CTXT_MAGIC (save); save->fs =get_fs (); -LASSERT (d_count (cfs_fs_pwd (current->fs))); +LASSERT (d_count (current->fs->pwd.dentry));LASSERT (d_count (new_ctx->pwd)); - save->pwd =dget (cfs_fs_pwd (current->fs)); - save->pwdmnt =mntget (cfs_fs_mnt (current->fs)); + save->pwd =dget (current->fs->pwd.dentry); + save->pwdmnt =mntget (current->fs->pwd.mnt); save->luc.luc_umask =current_umask (); save->ngroups =current_cred ()->group_info->ngroups;
@@ -170,10 +170,10 @@ void pop_ctxt(struct lvfs_run_ctxt *saved, struct lvfs_run_ctxt *new_ctx,ASSERT_CTXT_MAGIC (saved->magic);ASSERT_KERNEL_CTXT ("popping non-kernel context!\n" ); -LASSERTF (cfs_fs_pwd (current->fs) == new_ctx->pwd,"%p != %p\n" , -cfs_fs_pwd (current->fs), new_ctx->pwd); -LASSERTF (cfs_fs_mnt (current->fs) == new_ctx->pwdmnt,"%p != %p\n" , -cfs_fs_mnt (current->fs), new_ctx->pwdmnt); +LASSERTF (current->fs->pwd.dentry == new_ctx->pwd,"%p != %p\n" , + current->fs->pwd.dentry, new_ctx->pwd); +LASSERTF (current->fs->pwd.mnt == new_ctx->pwdmnt,"%p != %p\n" , + current->fs->pwd.mnt, new_ctx->pwdmnt);set_fs (saved->fs);ll_set_fs_pwd (current->fs, saved->pwdmnt, saved->pwd);
drivers/staging/lustre/lustre/ptlrpc/sec_config.c +1 -1
@@ -1015,7 +1015,7 @@ int sptlrpc_target_local_copy_conf(struct obd_device *obd,push_ctxt (&saved, &obd->obd_lvfs_ctxt,NULL ); - dentry =ll_lookup_one_len (MOUNT_CONFIGS_DIR ,cfs_fs_pwd (current->fs), + dentry =ll_lookup_one_len (MOUNT_CONFIGS_DIR , current->fs->pwd.dentry,strlen (MOUNT_CONFIGS_DIR ));if (IS_ERR (dentry)) { rc =PTR_ERR (dentry);
Signature · 1 line
1.7.9.5
PATCH 24/39 staging/lustre/obdclass: fix wrong device put in case of race
In reply to Peng TaoFrom: Mikhail Pershin <[email protected]>
The ls_device_put() might be called from local_oid_storage_fini() wrongly if mutex-protected check failed due to race
Lustre-change: http://review.whamcloud.com/6334
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3316
--- .../staging/lustre/lustre/obdclass/local_storage.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)
drivers/staging/lustre/lustre/obdclass/local_storage.c +8 -5
@@ -882,12 +882,15 @@ void local_oid_storage_fini(const struct lu_env *env, ls =dt2ls_dev (los->los_dev);mutex_lock (&ls->ls_los_mutex); -if (atomic_read (&los->los_refcount) ==0 ) { -if (los->los_obj) -lu_object_put_nocache (env, &los->los_obj->do_lu); -list_del (&los->los_list); -OBD_FREE_PTR (los); +if (atomic_read (&los->los_refcount) >0 ) { +mutex_unlock (&ls->ls_los_mutex); +return ; } + +if (los->los_obj) +lu_object_put_nocache (env, &los->los_obj->do_lu); +list_del (&los->los_list); +OBD_FREE_PTR (los);mutex_unlock (&ls->ls_los_mutex);ls_device_put (env, ls); }
Signature · 1 line
1.7.9.5
PATCH 25/39 staging/lustre/lmv: choose right MDT for open by FID
In reply to Peng TaoFrom: wang di <[email protected]>
If the open is by FID, LMV should locate MDT by the FID of the file/directory, instead of its parent.
Lustre-change: http://review.whamcloud.com/7253
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3712
--- drivers/staging/lustre/lustre/lmv/lmv_intent.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/staging/lustre/lustre/lmv/lmv_intent.c +7 -1
@@ -167,7 +167,13 @@ int lmv_intent_open(struct obd_export *exp, struct md_op_data *op_data,struct mdt_body *body;int rc; - tgt =lmv_locate_mds (lmv, op_data, &op_data->op_fid1); +/* Note: client might open with some random flags(sanity 33b), so we can +* not make sure op_fid2 is being initialized with BY_FID flag */ +if (it->it_flags &MDS_OPEN_BY_FID &&fid_is_sane (&op_data->op_fid2)) + tgt =lmv_locate_mds (lmv, op_data, &op_data->op_fid2); +else + tgt =lmv_locate_mds (lmv, op_data, &op_data->op_fid1); +if (IS_ERR (tgt))return PTR_ERR (tgt);
Signature · 1 line
1.7.9.5
PATCH 26/39 staging/lustre/osd: remove fld lookup during configuration
In reply to Peng TaoFrom: wang di <[email protected]>
Remove fld lookup during configuration, so to avoid accessing FLDB on MDT0 before MDT0 is setup.
1. add od_is_ost to check whether the FID is on OST, instead of lookup in FLDB.
2. add oic_device in OI cache, so in OI cache lookup, it will try to match device to avoid the ino of agent inode is being retrieved from OI cache. Then osd_remote_fid, which will trigger fld lookup, can be removed from osd index lookup.
Lustre-change: http://review.whamcloud.com/7266
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3126
--- .../lustre/lustre/include/lustre/lustre_idl.h | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/staging/lustre/lustre/include/lustre/lustre_idl.h +6 -1
@@ -611,10 +611,15 @@ static inline obd_id fid_idif_id(obd_seq seq, __u32 oid, __u32 ver)return ((__u64 )ver <<48 ) | ((seq &0xffff ) <<32 ) | oid; } +static inline __u32 idif_ost_idx (obd_seq seq) +{ +return (seq >>16 ) &0xffff ; +} +/* extract ost index from IDIF FID */ static inline __u32 fid_idif_ost_idx (const struct lu_fid *fid) { -return (fid_seq (fid) >>16 ) &0xffff ; +return idif_ost_idx (fid_seq (fid)); }/* extract OST sequence (group) from a wire ost_id (id/seq) pair */
Signature · 1 line
1.7.9.5
PATCH 27/39 staging/lustre/mdt: HSM EXIST event not triggered at last rm/mv
In reply to Peng TaoFrom: JC Lafoucriere <[email protected]>
After a rm or mv, if the target file is destroyed (last unlink) and if there is an archive in the backend a flags must be set in changelog event
Lustre-change: http://review.whamcloud.com/7260
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3718
--- .../lustre/lustre/include/lustre/lustre_user.h | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/staging/lustre/lustre/include/lustre/lustre_user.h +4 -1
@@ -692,7 +692,10 @@ static inline const char *changelog_type2str(int type) {#define CLF_UNLINK_HSM_EXISTS 0x0002/* File has something in HSM */ /* HSM cleaning needed */ /* Flags for rename */ -#define CLF_RENAME_LAST 0x0001/* rename unlink last hardlink of target */ +#define CLF_RENAME_LAST 0x0001/* rename unlink last hardlink +* of target */ +#define CLF_RENAME_LAST_EXISTS 0x0002/* rename unlink last hardlink of target +* has an archive in backend */ /* Flags for HSM */ /* 12b used (from high weight to low weight):
Signature · 1 line
1.7.9.5
PATCH 28/39 staging/lustre/ldlm: ldlm_flock_deadlock() ASSERTION( req != lock ) failed
In reply to Peng TaoFrom: Andriy Skulysh <[email protected]>
Abort processing of flock blockers list
on reaching disconnected export
Lustre-change: http://review.whamcloud.com/7392
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3783
--- drivers/staging/lustre/lustre/ldlm/ldlm_flock.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/staging/lustre/lustre/ldlm/ldlm_flock.c +3 -0
@@ -195,6 +195,9 @@ ldlm_flock_deadlock(struct ldlm_lock *req, struct ldlm_lock *bl_lock)cfs_hash_put (bl_exp->exp_flock_hash, &lock->l_exp_flock_hash); bl_exp = bl_exp_new; +if (bl_exp->exp_failed) +break ; +if (bl_owner == req_owner && bl_exp == req_exp) {class_export_put (bl_exp);return 1 ;
Signature · 1 line
1.7.9.5
PATCH 29/39 staging/lustre/changelogs: Account for changelog_ext_rec in CR_MAXSIZE
In reply to Peng TaoFrom: "Christopher J. Morrone" <[email protected]>
CR_MAXSIZE needs to account for an llog_changelog_rec that actually contains a changelog_ext_rec structure rather than a changelog_rec. With out doing so, a file size approaching the Linux kernel NAME_MAX length that is renamed to a size also close to, or at, NAME_MAX will exceed CR_MAXSIZE and trip an assertion.
Lustre-change: http://review.whamcloud.com/6993
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3587
--- .../lustre/lustre/include/lustre/lustre_user.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/staging/lustre/lustre/include/lustre/lustre_user.h +2 -1
@@ -766,7 +766,8 @@ static inline void hsm_set_cl_error(int *flags, int error) *flags |= (error <<CLF_HSM_ERR_L ); } -#define CR_MAXSIZE cfs_size_round(2*NAME_MAX + 1 + sizeof(struct changelog_rec)) +#define CR_MAXSIZE cfs_size_round(2*NAME_MAX + 1 + \ + sizeof(struct changelog_ext_rec))struct changelog_rec {__u16 cr_namelen;
Signature · 1 line
1.7.9.5
PATCH 30/39 staging/lustre/build: build error with gcc 4.7.0 20110509
In reply to Peng TaoFrom: Dmitry Eremin <[email protected]>
Fixed the error: 'tgt' and 'nf' may be used uninitialized in this function. Also fixed comparison signed and unsigned indexes.
Lustre-change: http://review.whamcloud.com/7382
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3775
--- drivers/staging/lustre/lustre/lmv/lmv_obd.c | 50 +++++++++++++++------------ 1 file changed, 27 insertions(+), 23 deletions(-)
drivers/staging/lustre/lustre/lmv/lmv_obd.c +27 -23
@@ -257,9 +257,9 @@ static int lmv_connect(const struct lu_env *env,static void lmv_set_timeouts (struct obd_device *obd) { -struct lmv_tgt_desc *tgt; -struct lmv_obd *lmv; -int i; +struct lmv_tgt_desc *tgt; +struct lmv_obd *lmv; +__u32 i; lmv = &obd->u.lmv;if (lmv->server_timeout ==0 )
@@ -283,7 +283,7 @@ static int lmv_init_ea_size(struct obd_export *exp, int easize, {struct obd_device *obd = exp->exp_obd;struct lmv_obd *lmv = &obd->u.lmv; -int i; +__u32 i;int rc =0 ;int change =0 ;
@@ -540,7 +540,7 @@ int lmv_check_connect(struct obd_device *obd) {struct lmv_obd *lmv = &obd->u.lmv;struct lmv_tgt_desc *tgt; -int i; +__u32 i;int rc;int easize;
@@ -652,7 +652,7 @@ static int lmv_disconnect(struct obd_export *exp)struct obd_device *obd =class_exp2obd (exp);struct lmv_obd *lmv = &obd->u.lmv;int rc; -int i; +__u32 i;if (!lmv->tgts) goto out_local;
@@ -776,8 +776,9 @@ static int lmv_hsm_req_count(struct lmv_obd *lmv,const struct hsm_user_request *hur, const struct lmv_tgt_desc *tgt_mds) { - int i, nr = 0; -struct lmv_tgt_desc *curr_tgt; + __u32 i; +int nr =0 ; +struct lmv_tgt_desc *curr_tgt;/* count how many requests must be sent to the given target */ for (i =0 ; i < hur->hur_request.hr_itemcount; i++) {
@@ -793,8 +794,8 @@ static void lmv_hsm_req_build(struct lmv_obd *lmv,const struct lmv_tgt_desc *tgt_mds, struct hsm_user_request *hur_out) { - int i, nr_out; -struct lmv_tgt_desc *curr_tgt; + __u32 i, nr_out; +struct lmv_tgt_desc *curr_tgt;/* build the hsm_user_request for the given target */ hur_out->hur_request = hur_in->hur_request;
@@ -816,7 +817,8 @@ static void lmv_hsm_req_build(struct lmv_obd *lmv,static int lmv_hsm_ct_unregister (struct lmv_obd *lmv,unsigned int cmd,int len,struct lustre_kernelcomm *lk,void *uarg) { -int i, rc =0 ; +__u32 i; +int rc;/* unregister request (call from llapi_hsm_copytool_fini) */ for (i =0 ; i < lmv->desc.ld_tgt_count; i++) {
@@ -837,8 +839,8 @@ static int lmv_hsm_ct_register(struct lmv_obd *lmv, unsigned int cmd, int len, struct lustre_kernelcomm *lk, void *uarg) { struct file *filp; -int i, j, err; -int rc =0 ; +__u32 i, j; +int err, rc;bool any_set = false;/* All or nothing: try to register to all MDS.
@@ -895,10 +897,10 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp, {struct obd_device *obddev =class_exp2obd (exp);struct lmv_obd *lmv = &obddev->u.lmv; -int i =0 ; +__u32 i =0 ;int rc =0 ;int set =0 ; -int count = lmv->desc.ld_tgt_count; +__u32 count = lmv->desc.ld_tgt_count;if (count ==0 )return -ENOTTY ;
@@ -1414,7 +1416,7 @@ static int lmv_statfs(const struct lu_env *env, struct obd_export *exp,struct lmv_obd *lmv = &obd->u.lmv;struct obd_statfs *temp;int rc =0 ; -int i; +__u32 i; rc =lmv_check_connect (obd);if (rc)
@@ -1584,7 +1586,7 @@ static int lmv_find_cbdata(struct obd_export *exp, const struct lu_fid *fid, {struct obd_device *obd = exp->exp_obd;struct lmv_obd *lmv = &obd->u.lmv; -int i; +__u32 i;int rc; rc =lmv_check_connect (obd);
@@ -2410,7 +2412,7 @@ int lmv_packmd(struct obd_export *exp, struct lov_mds_md **lmmp,struct lmv_stripe_md *meap;struct lmv_stripe_md *lsmp;int mea_size; -int i; +__u32 i; mea_size =lmv_get_easize (lmv);if (!lmmp)
@@ -2458,7 +2460,7 @@ int lmv_unpackmd(struct obd_export *exp, struct lov_stripe_md **lsmp,struct lmv_stripe_md *mea = (struct lmv_stripe_md *)lmm;struct lmv_obd *lmv = &obd->u.lmv;int mea_size; -int i; +__u32 i;__u32 magic; mea_size =lmv_get_easize (lmv);
@@ -2512,7 +2514,7 @@ static int lmv_cancel_unused(struct obd_export *exp, const struct lu_fid *fid,struct lmv_obd *lmv = &obd->u.lmv;int rc =0 ;int err; -int i; +__u32 i;LASSERT (fid !=NULL );
@@ -2547,7 +2549,7 @@ ldlm_mode_t lmv_lock_match(struct obd_export *exp, __u64 flags,struct obd_device *obd = exp->exp_obd;struct lmv_obd *lmv = &obd->u.lmv;ldlm_mode_t rc; -int i; +__u32 i;CDEBUG (D_INODE ,"Lock match for " DFID "\n" ,PFID (fid));
@@ -2723,7 +2725,8 @@ int lmv_quotactl(struct obd_device *unused, struct obd_export *exp,struct obd_device *obd =class_exp2obd (exp);struct lmv_obd *lmv = &obd->u.lmv;struct lmv_tgt_desc *tgt = lmv->tgts[0 ]; -int rc =0 , i; +int rc =0 ; +__u32 i;__u64 curspace, curinodes;if (!lmv->desc.ld_tgt_count || !tgt->ltd_active) {
@@ -2770,7 +2773,8 @@ int lmv_quotacheck(struct obd_device *unused, struct obd_export *exp,struct obd_device *obd =class_exp2obd (exp);struct lmv_obd *lmv = &obd->u.lmv;struct lmv_tgt_desc *tgt; -int i, rc =0 ; +int rc =0 ; +__u32 i;for (i =0 ; i < lmv->desc.ld_tgt_count; i++) {int err;
Signature · 1 line
1.7.9.5
PATCH 31/39 staging/lustre/build: fix 'NULL pointer dereference' errors
In reply to Peng TaoFrom: Sebastien Buisson <[email protected]>
Fix 'NULL pointer dereference' defects found by Coverity version 6.5.0:
Dereference after null check (FORWARD_NULL)
For instance, Passing null pointer to a function which dereferences it.
Dereference before null check (REVERSE_INULL) Null-checking variable suggests that it may be null, but it has already been dereferenced on all paths leading to the check. Dereference null return value (NULL_RETURNS)
Lustre-change: http://review.whamcloud.com/5868
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3042
--- .../staging/lustre/lustre/include/lustre_update.h | 3 ++- drivers/staging/lustre/lustre/llite/dir.c | 17 ++++++++++++----- drivers/staging/lustre/lustre/lov/lov_io.c | 3 ++- 3 files changed, 16 insertions(+), 7 deletions(-)
drivers/staging/lustre/lustre/include/lustre_update.h +2 -1
@@ -165,12 +165,13 @@ static inline int update_get_reply_buf(struct update_reply *reply, void **buf,int result; ptr =update_get_buf_internal (reply, index, &size); +LASSERT (ptr !=NULL ); result = *(int *)ptr;if (result <0 )return result; -LASSERT ((ptr !=NULL && size >=sizeof (int ))); +LASSERT (size >=sizeof (int )); *buf = ptr +sizeof (int );return size -sizeof (int ); }
drivers/staging/lustre/lustre/llite/dir.c +12 -5
@@ -597,23 +597,30 @@ static int ll_readdir(struct file *filp, struct dir_context *ctx)struct inode *inode = filp->f_dentry->d_inode;struct ll_file_data *lfd =LUSTRE_FPRIVATE (filp);struct ll_sb_info *sbi =ll_i2sbi (inode); +__u64 pos;int hash64 = sbi->ll_flags &LL_SBI_64BIT_HASH ;int api32 =ll_need_32bit_api (sbi);int rc; -CDEBUG (D_VFSTRACE ,"VFS Op:inode=%lu/%u(%p) pos %lu/%llu " +if (lfd !=NULL ) + pos = lfd->lfd_pos; +else + pos =0 ; + +CDEBUG (D_VFSTRACE ,"VFS Op:inode=%lu/%u(%p) pos %llu/%llu " " 32bit_api %d\n" , inode->i_ino, inode->i_generation, - inode, (unsigned long )lfd->lfd_pos,i_size_read (inode), api32); + inode, pos,i_size_read (inode), api32); -if (lfd->lfd_pos ==MDS_DIR_END_OFF ) +if (pos ==MDS_DIR_END_OFF )/* * end-of-file. */ GOTO (out, rc =0 ); - ctx->pos = lfd->lfd_pos; + ctx->pos = pos; rc =ll_dir_read (inode, ctx); - lfd->lfd_pos = ctx->pos; +if (lfd !=NULL ) + lfd->lfd_pos = ctx->pos;if (ctx->pos ==MDS_DIR_END_OFF ) {if (api32) ctx->pos =LL_DIR_END_OFF_32BIT ;
drivers/staging/lustre/lustre/lov/lov_io.c +2 -1
@@ -272,10 +272,11 @@ struct lov_io_sub *lov_page_subio(const struct lu_env *env, struct lov_io *lio,static int lov_io_subio_init (const struct lu_env *env,struct lov_io *lio,struct cl_io *io) { -struct lov_stripe_md *lsm = lio->lis_object->lo_lsm; +struct lov_stripe_md *lsm;int result;LASSERT (lio->lis_object !=NULL ); + lsm = lio->lis_object->lo_lsm;/* * Need to be optimized, we can't afford to allocate a piece of memory
Signature · 1 line
1.7.9.5
PATCH 32/39 staging/lustre/clio: honor O_NOATIME
In reply to Peng TaoFrom: "John L. Hammond" <[email protected]>
Add a ci_noatime bit to struct cl_io. In ll_io_init() set this bit if O_NOATIME is set in f_flags. Ensure that this bit is propagated down to lower layers. In osc_io_read_start() don't update atime if this bit is set. Add sanity test 39n to check that passing O_NOATIME to open() is honored.
Lustre-change: http://review.whamcloud.com/7442
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3832
--- drivers/staging/lustre/lustre/include/cl_object.h | 6 ++++- drivers/staging/lustre/lustre/llite/file.c | 29 +++++++++++++++++++++ drivers/staging/lustre/lustre/lov/lov_io.c | 1 + drivers/staging/lustre/lustre/osc/osc_io.c | 21 +++++++-------- 4 files changed, 44 insertions(+), 13 deletions(-)
drivers/staging/lustre/lustre/include/cl_object.h +5 -1
@@ -2392,7 +2392,11 @@ struct cl_io {/** * file is released, restore has to to be triggered by vvp layer */ - ci_restore_needed:1 ; + ci_restore_needed:1 , +/** +* O_NOATIME +*/ + ci_noatime:1;/** * Number of pages owned by this IO. For invariant checking. */
drivers/staging/lustre/lustre/llite/file.c +29 -0
@@ -1032,6 +1032,33 @@ int ll_glimpse_ioctl(struct ll_sb_info *sbi, struct lov_stripe_md *lsm,return rc; } +static bool file_is_noatime (const struct file *file) +{ +const struct vfsmount *mnt = file->f_path.mnt; +const struct inode *inode = file->f_path.dentry->d_inode; + +/* Adapted from file_accessed() and touch_atime().*/ +if (file->f_flags &O_NOATIME ) +return true; + +if (inode->i_flags &S_NOATIME ) +return true; + +if (IS_NOATIME (inode)) +return true; + +if (mnt->mnt_flags & (MNT_NOATIME |MNT_READONLY )) +return true; + +if ((mnt->mnt_flags &MNT_NODIRATIME ) &&S_ISDIR (inode->i_mode)) +return true; + +if ((inode->i_sb->s_flags &MS_NODIRATIME ) &&S_ISDIR (inode->i_mode)) +return true; + +return false; +} +void ll_io_init (struct cl_io *io,const struct file *file,int write) {struct inode *inode = file->f_dentry->d_inode;
@@ -1051,6 +1078,8 @@ void ll_io_init(struct cl_io *io, const struct file *file, int write) }else if (file->f_flags &O_APPEND ) { io->ci_lockreq =CILR_MANDATORY ; } + + io->ci_noatime =file_is_noatime (file); } static ssize_t
drivers/staging/lustre/lustre/lov/lov_io.c +1 -0
@@ -194,6 +194,7 @@ static int lov_io_sub_init(const struct lu_env *env, struct lov_io *lio, sub_io->ci_lockreq = io->ci_lockreq; sub_io->ci_type = io->ci_type; sub_io->ci_no_srvlock = io->ci_no_srvlock; + sub_io->ci_noatime = io->ci_noatime;lov_sub_enter (sub); result = cl_io_sub_init(sub->sub_env, sub_io,
drivers/staging/lustre/lustre/osc/osc_io.c +9 -12
@@ -509,22 +509,19 @@ static void osc_io_setattr_end(const struct lu_env *env,static int osc_io_read_start (const struct lu_env *env,const struct cl_io_slice *slice) { -struct osc_io *oio =cl2osc_io (env, slice); -struct cl_object *obj = slice->cis_obj; -struct cl_attr *attr = &osc_env_info (env)->oti_attr; -int result =0 ; +struct osc_io *oio =cl2osc_io (env, slice); +struct cl_object *obj = slice->cis_obj; +struct cl_attr *attr = &osc_env_info (env)->oti_attr; +int rc =0 ; -if (oio->oi_lockless ==0 ) { +if (oio->oi_lockless ==0 && !slice->cis_io->ci_noatime) {cl_object_attr_lock (obj); - result =cl_object_attr_get (env, obj, attr); -if (result ==0 ) { - attr->cat_atime =LTIME_S (CURRENT_TIME ); - result =cl_object_attr_set (env, obj, attr, -CAT_ATIME ); - } + attr->cat_atime =LTIME_S (CURRENT_TIME ); + rc =cl_object_attr_set (env, obj, attr,CAT_ATIME );cl_object_attr_unlock (obj); } -return result; + +return rc; } static int osc_io_write_start(const struct lu_env *env,
Signature · 1 line
1.7.9.5
PATCH 33/39 staging/lustre/llite: return compatible fsid for statfs
In reply to Peng TaoFrom: Fan Yong <[email protected]>
Lustre uses 64-bits inode number to identify object on client side. When re-export Lustre via NFS, NFS will detect whether support fsid via statfs(). For the non-support case, it will only recognizes and packs low 32-bits inode number in nfs handle. Such handle cannot be used to locate the object properly.
To avoid patch linux kernel, Lustre client should generate fsid and return it via statfs() to up layer.
To be compatible with old Lustre client (NFS server), the fsid will be generated from super_block::s_dev.
Lustre-change: http://review.whamcloud.com/7434
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2904
--- .../staging/lustre/lustre/llite/llite_internal.h | 2 -- drivers/staging/lustre/lustre/llite/llite_lib.c | 8 ++++---- drivers/staging/lustre/lustre/llite/llite_nfs.c | 16 ---------------- 3 files changed, 4 insertions(+), 22 deletions(-)
drivers/staging/lustre/lustre/llite/llite_internal.h +0 -2
@@ -527,7 +527,6 @@ struct ll_sb_info {* clustred nfs */ struct rmtacl_ctl_table ll_rct;struct eacl_table ll_et; -__kernel_fsid_t ll_fsid; };#define LL_DEFAULT_MAX_RW_CHUNK (32 * 1024 * 1024)
@@ -872,7 +871,6 @@ char *ll_get_fsname(struct super_block *sb, char *buf, int buflen);/* llite/llite_nfs.c */ extern struct export_operations lustre_export_operations;__u32 get_uuid2int (const char *name,int len); -void get_uuid2fsid (const char *name,int len,__kernel_fsid_t *fsid);struct inode *search_inode_for_lustre (struct super_block *sb,const struct lu_fid *fid);
drivers/staging/lustre/lustre/llite/llite_lib.c +4 -4
@@ -574,10 +574,8 @@ static int client_common_fill_super(struct super_block *sb, char *md, char *dt,/* s_dev is also used in lt_compare() to compare two fs, but that is * only a node-local comparison. */ uuid =obd_get_uuid (sbi->ll_md_exp); -if (uuid !=NULL ) { +if (uuid !=NULL ) sb->s_dev =get_uuid2int (uuid->uuid,strlen (uuid->uuid)); -get_uuid2fsid (uuid->uuid,strlen (uuid->uuid), &sbi->ll_fsid); - }if (data !=NULL )OBD_FREE_PTR (data);
@@ -1612,6 +1610,7 @@ int ll_statfs(struct dentry *de, struct kstatfs *sfs) {struct super_block *sb = de->d_sb;struct obd_statfs osfs; +__u64 fsid =huge_encode_dev (sb->s_dev);int rc;CDEBUG (D_VFSTRACE ,"VFS Op: at " LPU64 " jiffies\n" ,get_jiffies_64 ());
@@ -1643,7 +1642,8 @@ int ll_statfs(struct dentry *de, struct kstatfs *sfs) sfs->f_blocks = osfs.os_blocks; sfs->f_bfree = osfs.os_bfree; sfs->f_bavail = osfs.os_bavail; - sfs->f_fsid =ll_s2sbi (sb)->ll_fsid; + sfs->f_fsid.val[0 ] = (__u32 )fsid; + sfs->f_fsid.val[1 ] = (__u32)(fsid >>32 );return 0 ; }
drivers/staging/lustre/lustre/llite/llite_nfs.c +0 -16
@@ -58,22 +58,6 @@ __u32 get_uuid2int(const char *name, int len)return (key0 <<1 ); } -void get_uuid2fsid (const char *name,int len,__kernel_fsid_t *fsid) -{ -__u64 key =0 , key0 =0x12a3fe2d , key1 =0x37abe8f9 ; - -while (len--) { - key = key1 + (key0 ^ (*name++ *7152373 )); -if (key &0x8000000000000000ULL ) - key -=0x7fffffffffffffffULL ; - key1 = key0; - key0 = key; - } - - fsid->val[0 ] = key; - fsid->val[1 ] = key >>32 ; -} -static int ll_nfs_test_inode (struct inode *inode,void *opaque) { return lu_fid_eq(&ll_i2info(inode)->lli_fid,
Signature · 1 line
1.7.9.5
PATCH 34/39 staging/lustre/llite: cancel open lock before closing file
In reply to Peng TaoFrom: Jinshan Xiong <[email protected]>
In error handling path of ll_lease_open(), och has already been freed in ll_close_inode_openhandle() so the sequence of cancel open lock and close open handle need adjusting.
Lustre-change: http://review.whamcloud.com/7346
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3647
--- drivers/staging/lustre/lustre/llite/file.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)
drivers/staging/lustre/lustre/llite/file.c +11 -7
@@ -847,20 +847,24 @@ struct obd_client_handle *ll_lease_open(struct inode *inode, struct file *file,return och; out_close: - rc2 =ll_close_inode_openhandle (sbi->ll_md_exp, inode, och,NULL ); -if (rc2) -CERROR ("Close openhandle returned %d\n" , rc2); - -/* cancel open lock */ +/* Cancel open lock */ if (it.d.lustre.it_lock_mode !=0 ) {ldlm_lock_decref_and_cancel (&och->och_lease_handle, - it.d.lustre.it_lock_mode); + it.d.lustre.it_lock_mode); it.d.lustre.it_lock_mode =0 ; + och->och_lease_handle.cookie =0ULL ; } + rc2 =ll_close_inode_openhandle (sbi->ll_md_exp, inode, och,NULL ); +if (rc2 <0 ) +CERROR ("%s: error closing file " DFID ": %d\n" , +ll_get_fsname (inode->i_sb,NULL ,0 ), +PFID (&ll_i2info (inode)->lli_fid), rc2); + och =NULL ;/* och has been freed in ll_close_inode_openhandle() */ out_release_it:ll_intent_release (&it); out: -OBD_FREE_PTR (och); +if (och !=NULL ) +OBD_FREE_PTR (och);return ERR_PTR (rc); }EXPORT_SYMBOL (ll_lease_open);
Signature · 1 line
1.7.9.5
PATCH 35/39 staging/lustre/hsm: Add support to drop all pages for ll_data_version
In reply to Peng TaoFrom: Jinshan Xiong <[email protected]>
This will be used by HSM release to get data version and drop all caching pages from all clients, before sending IT_RELEASE close REQ to MDT.
Lustre-change: http://review.whamcloud.com/6794
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3647
--- .../lustre/lustre/include/lustre/lustre_idl.h | 1 + .../lustre/lustre/include/lustre/lustre_user.h | 4 +-- drivers/staging/lustre/lustre/llite/dir.c | 5 ++- drivers/staging/lustre/lustre/llite/file.c | 36 ++++++++++++-------- .../staging/lustre/lustre/llite/llite_internal.h | 2 +- drivers/staging/lustre/lustre/lov/lov_merge.c | 13 ++++++- 6 files changed, 40 insertions(+), 21 deletions(-)
drivers/staging/lustre/lustre/include/lustre/lustre_idl.h +1 -0
@@ -1506,6 +1506,7 @@ enum obdo_flags {* clients prior than 2.2 */ OBD_FL_RECOV_RESEND =0x00080000 ,/* recoverable resent */ OBD_FL_NOSPC_BLK =0x00100000 ,/* no more block space on OST */ + OBD_FL_FLUSH = 0x00200000, /* flush pages on the OST *//* Note that while these checksum values are currently separate bits, * in 2.x we can actually allow all values from 1-31 if we wanted. */
drivers/staging/lustre/lustre/include/lustre/lustre_user.h +2 -2
@@ -849,8 +849,8 @@ struct ioc_data_version {__u64 idv_version;__u64 idv_flags;/* See LL_DV_xxx */ }; -#define LL_DV_NOFLUSH 0x01/* Do not take READ EXTENT LOCK before sampling -version. Dirty caches are left unchanged. */ +#define LL_DV_RD_FLUSH (1 << 0)/* Flush dirty pages from clients */ +#define LL_DV_WR_FLUSH (1 << 1)/* Flush all caching pages from clients */ #ifndef offsetof# define offsetof (typ,memb) ((unsigned long)((char *)&(((typ *)0)->memb)))
drivers/staging/lustre/lustre/llite/dir.c +2 -3
@@ -929,7 +929,7 @@ static int ll_ioc_copy_start(struct super_block *sb, struct hsm_copy *copy) }/* Read current file data version */ - rc =ll_data_version (inode, &data_version,1 ); + rc =ll_data_version (inode, &data_version,LL_DV_RD_FLUSH );iput (inode);if (rc !=0 ) { CDEBUG(D_HSM, "Could not read file data version of "
@@ -1012,8 +1012,7 @@ static int ll_ioc_copy_end(struct super_block *sb, struct hsm_copy *copy)GOTO (progress, rc =PTR_ERR (inode)); } - rc =ll_data_version (inode, &data_version, - copy->hc_hai.hai_action ==HSMA_ARCHIVE ); + rc =ll_data_version (inode, &data_version,LL_DV_RD_FLUSH );iput (inode);if (rc) { CDEBUG(D_HSM, "Could not read file data version. "
drivers/staging/lustre/lustre/llite/file.c +22 -14
@@ -905,7 +905,7 @@ EXPORT_SYMBOL(ll_lease_close);/* Fills the obdo with the attributes for the lsm */ static int ll_lsm_getattr (struct lov_stripe_md *lsm,struct obd_export *exp,struct obd_capa *capa,struct obdo *obdo, -__u64 ioepoch,int sync) +__u64 ioepoch,int dv_flags) {struct ptlrpc_request_set *set;struct obd_info oinfo = { { {0 } } };
@@ -925,9 +925,11 @@ static int ll_lsm_getattr(struct lov_stripe_md *lsm, struct obd_export *exp,OBD_MD_FLGROUP |OBD_MD_FLEPOCH |OBD_MD_FLDATAVERSION ; oinfo.oi_capa = capa; -if (sync) { +if (dv_flags & (LL_DV_WR_FLUSH |LL_DV_RD_FLUSH )) { oinfo.oi_oa->o_valid |=OBD_MD_FLFLAGS ; oinfo.oi_oa->o_flags |=OBD_FL_SRVLOCK ; +if (dv_flags &LL_DV_WR_FLUSH ) + oinfo.oi_oa->o_flags |=OBD_FL_FLUSH ; } set =ptlrpc_prep_set ();
@@ -940,11 +942,16 @@ static int ll_lsm_getattr(struct lov_stripe_md *lsm, struct obd_export *exp, rc =ptlrpc_set_wait (set);ptlrpc_set_destroy (set); } -if (rc ==0 ) +if (rc ==0 ) { oinfo.oi_oa->o_valid &= (OBD_MD_FLBLOCKS |OBD_MD_FLBLKSZ |OBD_MD_FLATIME |OBD_MD_FLMTIME |OBD_MD_FLCTIME |OBD_MD_FLSIZE | -OBD_MD_FLDATAVERSION ); +OBD_MD_FLDATAVERSION |OBD_MD_FLFLAGS ); +if (dv_flags &LL_DV_WR_FLUSH && + !(oinfo.oi_oa->o_valid &OBD_MD_FLFLAGS && + oinfo.oi_oa->o_flags &OBD_FL_FLUSH )) +return -ENOTSUPP ; + }return rc; }
@@ -961,7 +968,7 @@ int ll_inode_getattr(struct inode *inode, struct obdo *obdo, lsm =ccc_inode_lsm_get (inode); rc =ll_lsm_getattr (lsm,ll_i2dtexp (inode), - capa, obdo, ioepoch, sync); + capa, obdo, ioepoch, sync ?LL_DV_RD_FLUSH :0 );capa_put (capa);if (rc ==0 ) {struct ost_id *oi = lsm ? &lsm->lsm_oi : &obdo->o_oi;
@@ -1905,11 +1912,12 @@ error:* This value is computed using stripe object version on OST. * Version is computed using server side locking. * -* @param extent_lock Take extent lock. Not needed if a process is already -* holding the OST object group locks. +* @param sync if do sync on the OST side; +* 0: no sync +* LL_DV_RD_FLUSH: flush dirty pages, LCK_PR on OSTs +* LL_DV_WR_FLUSH: drop all caching pages, LCK_PW on OSTs */ -int ll_data_version (struct inode *inode,__u64 *data_version, -int extent_lock) +int ll_data_version (struct inode *inode,__u64 *data_version,int flags) {struct lov_stripe_md *lsm =NULL ;struct ll_sb_info *sbi =ll_i2sbi (inode);
@@ -1928,7 +1936,7 @@ int ll_data_version(struct inode *inode, __u64 *data_version,if (obdo ==NULL )GOTO (out, rc = -ENOMEM ); - rc =ll_lsm_getattr (lsm, sbi->ll_dt_exp,NULL , obdo,0 , extent_lock); + rc =ll_lsm_getattr (lsm, sbi->ll_dt_exp,NULL , obdo,0 , flags);if (rc ==0 ) {if (!(obdo->o_valid &OBD_MD_FLDATAVERSION )) rc = -EOPNOTSUPP ;
@@ -1963,7 +1971,7 @@ int ll_hsm_release(struct inode *inode)GOTO (out, rc =PTR_ERR (och));/* Grab latest data_version and [am]time values */ - rc =ll_data_version (inode, &data_version,1 ); + rc =ll_data_version (inode, &data_version,LL_DV_WR_FLUSH );if (rc !=0 )GOTO (out, rc);
@@ -2335,13 +2343,13 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg)return ll_fid2path (inode, (void *)arg);case LL_IOC_DATA_VERSION : {struct ioc_data_version idv; -int rc; +int rc;if (copy_from_user (&idv, (char *)arg,sizeof (idv)))return -EFAULT ; - rc =ll_data_version (inode, &idv.idv_version, - !(idv.idv_flags &LL_DV_NOFLUSH )); + idv.idv_flags &=LL_DV_RD_FLUSH |LL_DV_WR_FLUSH ; + rc =ll_data_version (inode, &idv.idv_version, idv.idv_flags);if (rc ==0 &©_to_user ((char *) arg, &idv,sizeof (idv)))return -EFAULT ;
drivers/staging/lustre/lustre/llite/llite_internal.h +1 -1
@@ -805,7 +805,7 @@ int ll_merge_lvb(const struct lu_env *env, struct inode *inode);int ll_get_grouplock (struct inode *inode,struct file *file,unsigned long arg);int ll_put_grouplock (struct inode *inode,struct file *file,unsigned long arg);int ll_fid2path (struct inode *inode,void *arg); -int ll_data_version (struct inode *inode,__u64 *data_version,int extent_lock); +int ll_data_version (struct inode *inode,__u64 *data_version,int flags);int ll_hsm_release (struct inode *inode); struct obd_client_handle *ll_lease_open(struct inode *inode, struct file *file,
drivers/staging/lustre/lustre/lov/lov_merge.c +12 -1
@@ -181,7 +181,8 @@ void lov_merge_attrs(struct obdo *tgt, struct obdo *src, obd_valid valid, { valid &= src->o_valid; -if (*set) { +if (*set !=0 ) { + tgt->o_valid &= valid;if (valid &OBD_MD_FLSIZE ) {/* this handles sparse files properly */ obd_size lov_size;
@@ -200,12 +201,22 @@ void lov_merge_attrs(struct obdo *tgt, struct obdo *src, obd_valid valid, tgt->o_mtime = src->o_mtime;if (valid &OBD_MD_FLDATAVERSION ) tgt->o_data_version += src->o_data_version; + +/* handle flags */ +if (valid &OBD_MD_FLFLAGS ) + tgt->o_flags &= src->o_flags; +else + tgt->o_flags =0 ; }else {memcpy (tgt, src,sizeof (*tgt)); tgt->o_oi = lsm->lsm_oi; + tgt->o_valid = valid;if (valid &OBD_MD_FLSIZE ) tgt->o_size =lov_stripe_size (lsm, src->o_size, stripeno); + tgt->o_flags =0 ; +if (valid &OBD_MD_FLFLAGS ) + tgt->o_flags = src->o_flags; }/* data_version needs to be valid on all stripes to be correct! */
Signature · 1 line
1.7.9.5
PATCH 36/39 staging/lustre/build: fix 'no effect' errors
In reply to Peng TaoFrom: Sebastien Buisson <[email protected]>
Fix 'no effect' issues found by Coverity version 6.5.1:
Unsigned compared against 0 (NO_EFFECT)
This greater-than-or-equal-to-zero comparison of an unsigned value is always true.
Lustre-change: http://review.whamcloud.com/7166
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3097
--- drivers/staging/lustre/lustre/llite/xattr.c | 5 ++--- drivers/staging/lustre/lustre/lmv/lmv_obd.c | 2 +- drivers/staging/lustre/lustre/lov/lov_obd.c | 2 +- drivers/staging/lustre/lustre/ptlrpc/sec_bulk.c | 1 - 4 files changed, 4 insertions(+), 6 deletions(-)
drivers/staging/lustre/lustre/llite/xattr.c +2 -3
@@ -410,11 +410,10 @@ getxattr_nocache: }#ifdef CONFIG_FS_POSIX_ACL -if (rce && rce->rce_ops ==RMT_LSETFACL ) { +if (rce !=NULL && rce->rce_ops ==RMT_LSETFACL ) {ext_acl_xattr_header *acl; - acl =lustre_posix_acl_xattr_2ext ((posix_acl_xattr_header *)buffer, - rc); + acl =lustre_posix_acl_xattr_2ext (buffer, rc);if (IS_ERR (acl))GOTO (out, rc =PTR_ERR (acl));
drivers/staging/lustre/lustre/lmv/lmv_obd.c +1 -1
@@ -947,7 +947,7 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp,struct obd_quotactl *oqctl;if (qctl->qc_valid ==QC_MDTIDX ) { -if (qctl->qc_idx <0 || count <= qctl->qc_idx) +if (count <= qctl->qc_idx)return -EINVAL ; tgt = lmv->tgts[qctl->qc_idx];
drivers/staging/lustre/lustre/lov/lov_obd.c +1 -1
@@ -1982,7 +1982,7 @@ static int lov_iocontrol(unsigned int cmd, struct obd_export *exp, int len,struct obd_quotactl *oqctl;if (qctl->qc_valid ==QC_OSTIDX ) { -if (qctl->qc_idx <0 || count <= qctl->qc_idx) +if (count <= qctl->qc_idx)return -EINVAL ; tgt = lov->lov_tgts[qctl->qc_idx];
drivers/staging/lustre/lustre/ptlrpc/sec_bulk.c +0 -1
@@ -451,7 +451,6 @@ out:static inline void enc_pools_wakeup (void ) {LASSERT (spin_is_locked (&page_pools.epp_lock)); -LASSERT (page_pools.epp_waitqlen >=0 );if (unlikely (page_pools.epp_waitqlen)) {LASSERT (waitqueue_active (&page_pools.epp_waitq));
Signature · 1 line
1.7.9.5
PATCH 37/39 staging/lustre/obd: add md_stats to MDC and LMV devices
In reply to Peng TaoFrom: "John L. Hammond" <[email protected]>
Enable md_stats for MDC and LMV devices and fix the definition of EXP_MD_COUNTER_INCREMENT() so that they will be tallied. These stats track usage of the md_ops methods (from the OBD layer, not to be confused with md_device methods) and are exported through the files /proc/fs/lustre/{lmv,mdc}/*/md_stats.
Rename m_sync to m_fsync making the counter name (fsync) more intuitive. Prune the minimally useful operations from set of md_ops to be counted. The operations counted are close, create, enqueue, getattr, intent_lock, link, rename, setattr, fsync, readpage, unlink, setxattr, egtxattr, intent_getattr_async, and revalidate_lock.
Add assertions to lprocfs_counter_{add,sub}() to ensure that the counter index lies in the appropriate range for stats. Remove the corresponding assertions from OBD_COUNTER_INCREMENT(), EXP_COUNTER_INCREMENT(), and EXP_MD_COUNTER_INCREMENT(). Add macros NUM_OBD_STATS and NUM_MD_STATS for use when allocating stats.
In mdt_export_stats_init() allocate and initialize the nid_stats member of struct exp_nid_stats using only the MDT private counters. This does not affect the contents of /proc/fs/lustre/mdt/*/exports/*/stats but does save memory.
Add a flag to struct obd_device to indicate if EXP_COUNTER_INCREMENT() should increment stats in the nid_stats member (as in the case of OFD) or not (MDT).
Remove the unused member exp_md_stats from struct obd_export.
Lustre-change: http://review.whamcloud.com/4827
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2484
--- .../staging/lustre/lustre/include/lustre_export.h | 1 - drivers/staging/lustre/lustre/include/obd.h | 139 ++++++++++++-------- drivers/staging/lustre/lustre/include/obd_class.h | 119 +++++++---------- drivers/staging/lustre/lustre/llite/file.c | 4 +- drivers/staging/lustre/lustre/lmv/lmv_obd.c | 21 +-- drivers/staging/lustre/lustre/lvfs/lvfs_lib.c | 6 + drivers/staging/lustre/lustre/mdc/mdc_request.c | 8 +- .../lustre/lustre/obdclass/lprocfs_status.c | 58 +++++--- 8 files changed, 198 insertions(+), 158 deletions(-)
drivers/staging/lustre/lustre/include/lustre_export.h +0 -1
@@ -191,7 +191,6 @@ struct obd_export {*/ struct obd_import *exp_imp_reverse;struct nid_stat *exp_nid_stats; -struct lprocfs_stats *exp_md_stats;/** Active connetion */ struct ptlrpc_connection *exp_connection;/** Connection count value from last succesful reconnect rpc */
drivers/staging/lustre/lustre/include/obd.h +83 -56
224 lines not on this page · raw message
drivers/staging/lustre/lustre/include/obd_class.h +51 -68
145 lines not on this page · raw message
drivers/staging/lustre/lustre/llite/file.c +2 -2
@@ -2692,8 +2692,8 @@ int ll_fsync(struct file *file, loff_t start, loff_t end, int datasync) } oc =ll_mdscapa_get (inode); - err =md_sync (ll_i2sbi (inode)->ll_md_exp,ll_inode2fid (inode), oc, - &req); + err =md_fsync (ll_i2sbi (inode)->ll_md_exp,ll_inode2fid (inode), oc, + &req);capa_put (oc);if (!rc) rc = err;
drivers/staging/lustre/lustre/lmv/lmv_obd.c +12 -9
@@ -1338,6 +1338,7 @@ static int lmv_setup(struct obd_device *obd, struct lustre_cfg *lcfg)lprocfs_lmv_init_vars (&lvars);lprocfs_obd_setup (obd, lvars.obd_vars); +lprocfs_alloc_md_stats (obd,0 );#ifdef LPROCFS { rc = lprocfs_seq_create(obd->obd_proc_entry, "target_obd",
@@ -2035,23 +2036,24 @@ static int lmv_setattr(struct obd_export *exp, struct md_op_data *op_data,return rc; } -static int lmv_sync (struct obd_export *exp,const struct lu_fid *fid, -struct obd_capa *oc,struct ptlrpc_request **request) +static int lmv_fsync (struct obd_export *exp,const struct lu_fid *fid, +struct obd_capa *oc,struct ptlrpc_request **request) { -struct obd_device *obd = exp->exp_obd; -struct lmv_obd *lmv = &obd->u.lmv; -struct lmv_tgt_desc *tgt; -int rc; +struct obd_device *obd = exp->exp_obd; +struct lmv_obd *lmv = &obd->u.lmv; +struct lmv_tgt_desc *tgt; +int rc; + rc =lmv_check_connect (obd); -if (rc) +if (rc !=0 )return rc; tgt =lmv_find_target (lmv, fid);if (IS_ERR (tgt))return PTR_ERR (tgt); - rc =md_sync (tgt->ltd_exp, fid, oc, request); + rc =md_fsync (tgt->ltd_exp, fid, oc, request);return rc; }
@@ -2298,6 +2300,7 @@ static int lmv_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)case OBD_CLEANUP_EXPORTS :fld_client_proc_fini (&lmv->lmv_fld);lprocfs_obd_cleanup (obd); +lprocfs_free_md_stats (obd);break ;default :break ;
@@ -2828,7 +2831,7 @@ struct md_ops lmv_md_ops = { .m_rename = lmv_rename, .m_setattr = lmv_setattr, .m_setxattr = lmv_setxattr, - .m_sync = lmv_sync, + .m_fsync = lmv_fsync, .m_readpage = lmv_readpage, .m_unlink = lmv_unlink, .m_init_ea_size = lmv_init_ea_size,
drivers/staging/lustre/lustre/lvfs/lvfs_lib.c +6 -0
@@ -54,6 +54,9 @@ void lprocfs_counter_add(struct lprocfs_stats *stats, int idx, long amount)if (stats ==NULL )return ; +LASSERTF (0 <= idx && idx < stats->ls_num, +"idx %d, ls_num %hu\n" , idx, stats->ls_num); +/* With per-client stats, statistics are allocated only for * single CPU area, so the smp_id should be 0 always. */ smp_id =lprocfs_stats_lock (stats,LPROCFS_GET_SMP_ID , &flags);
@@ -102,6 +105,9 @@ void lprocfs_counter_sub(struct lprocfs_stats *stats, int idx, long amount)if (stats ==NULL )return ; +LASSERTF (0 <= idx && idx < stats->ls_num, +"idx %d, ls_num %hu\n" , idx, stats->ls_num); +/* With per-client stats, statistics are allocated only for * single CPU area, so the smp_id should be 0 always. */ smp_id =lprocfs_stats_lock (stats,LPROCFS_GET_SMP_ID , &flags);
drivers/staging/lustre/lustre/mdc/mdc_request.c +5 -3
@@ -2233,8 +2233,8 @@ static int mdc_unpin(struct obd_export *exp, struct obd_client_handle *handle,return rc; } -int mdc_sync (struct obd_export *exp,const struct lu_fid *fid, -struct obd_capa *oc,struct ptlrpc_request **request) +int mdc_fsync (struct obd_export *exp,const struct lu_fid *fid, +struct obd_capa *oc,struct ptlrpc_request **request) {struct ptlrpc_request *req;int rc;
@@ -2385,6 +2385,7 @@ static int mdc_setup(struct obd_device *obd, struct lustre_cfg *cfg)GOTO (err_close_lock, rc);lprocfs_mdc_init_vars (&lvars);lprocfs_obd_setup (obd, lvars.obd_vars); +lprocfs_alloc_md_stats (obd,0 );sptlrpc_lprocfs_cliobd_attach (obd);ptlrpc_lprocfs_register_obd (obd);
@@ -2445,6 +2446,7 @@ static int mdc_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)obd_cleanup_client_import (obd);ptlrpc_lprocfs_unregister_obd (obd);lprocfs_obd_cleanup (obd); +lprocfs_free_md_stats (obd); rc =obd_llog_finish (obd,0 );if (rc !=0 )
@@ -2651,7 +2653,7 @@ struct md_ops mdc_md_ops = { .m_setattr = mdc_setattr, .m_setxattr = mdc_setxattr, .m_getxattr = mdc_getxattr, - .m_sync = mdc_sync, + .m_fsync = mdc_fsync, .m_readpage = mdc_readpage, .m_unlink = mdc_unlink, .m_cancel_unused = mdc_cancel_unused,
drivers/staging/lustre/lustre/obdclass/lprocfs_status.c +39 -19
114 lines not on this page · raw message
Signature · 1 line
1.7.9.5
PATCH 38/39 staging/lustre/autoconf: remove LC_SECURITY_PLUG test
In reply to Peng TaoFrom: James Simmons <[email protected]>
Only the SLES 10 kernel took a vfsmount for various vfs operations to support AppArmor before the path variants were available. We can remove the test and the dead code.
Lustre-change: http://review.whamcloud.com/5360
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2800
--- .../lustre/lustre/include/linux/lustre_compat25.h | 9 --------- drivers/staging/lustre/lustre/lvfs/lvfs_linux.c | 3 +-- 2 files changed, 1 insertion(+), 11 deletions(-)
drivers/staging/lustre/lustre/include/linux/lustre_compat25.h +0 -9
@@ -99,15 +99,6 @@ static inline void ll_set_fs_pwd(struct fs_struct *fs, struct vfsmount *mnt,#define FS_HAS_FIEMAP (0)#endif -#define ll_vfs_rmdir (dir,entry,mnt) vfs_rmdir(dir,entry) -#define ll_vfs_mkdir (inode,dir,mnt,mode) vfs_mkdir(inode,dir,mode) -#define ll_vfs_link (old,mnt,dir,new,mnt1) vfs_link(old,dir,new) -#define ll_vfs_unlink (inode,entry,mnt) vfs_unlink(inode,entry) -#define ll_vfs_mknod (dir,entry,mnt,mode,dev) vfs_mknod(dir,entry,mode,dev) -#define ll_security_inode_unlink (dir,entry,mnt) security_inode_unlink(dir,entry) -#define ll_vfs_rename (old,old_dir,mnt,new,new_dir,mnt1) \ - vfs_rename(old,old_dir,new,new_dir) -#define cfs_bio_io_error (a,b) bio_io_error((a))#define cfs_bio_endio (a,b,c) bio_endio((a),(c))
drivers/staging/lustre/lustre/lvfs/lvfs_linux.c +1 -2
@@ -223,8 +223,7 @@ int lustre_rename(struct dentry *dir, struct vfsmount *mnt,if (IS_ERR (dchild_new))GOTO (put_old, err =PTR_ERR (dchild_new)); - err =ll_vfs_rename (dir->d_inode, dchild_old, mnt, - dir->d_inode, dchild_new, mnt); + err =vfs_rename (dir->d_inode, dchild_old, dir->d_inode, dchild_new);dput (dchild_new); put_old:
Signature · 1 line
1.7.9.5
PATCH 39/39 staging/lustre/osc: Lustre returns EINTR from writes when SA_RESTART is set
In reply to Peng TaoFrom: James Simmons <[email protected]>
When Lustre is in a read or write system call and receives a SIGALRM, it will return EINTR if interrupted in osc_enter_cache. This prevents the system call from being restarted if SA_RESTART is set in the handler.
This patch changes behavior in this location to return ERESTARTSYS when a signal arrives during the call to l_wait_event.
Lustre-change: http://review.whamcloud.com/7002
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3581
--- drivers/staging/lustre/lustre/osc/osc_cache.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-)
drivers/staging/lustre/lustre/osc/osc_cache.c +13 -2
@@ -1533,13 +1533,24 @@ static int osc_enter_cache(const struct lu_env *env, struct client_obd *cli,/* l_wait_event is interrupted by signal, or timed out */ if (rc <0 ) { -if (rc == -ETIMEDOUT ) { +switch (rc) { +case -ETIMEDOUT :OSC_DUMP_GRANT (D_ERROR , cli,"try to reserve %d.\n" , bytes);osc_extent_tree_dump (D_ERROR , osc); rc = -EDQUOT ; +break ; +case -EINTR : +/* Ensures restartability - LU-3581 */ + rc = -ERESTARTSYS ; +break ; +default : +CDEBUG (D_CACHE ,"%s: event for cache space @" +" %p never arrived due to %d\n" , + cli->cl_import->imp_obd->obd_name, + &ocw, rc); +break ; } -list_del_init (&ocw.ocw_entry);GOTO (out, rc); }
Signature · 1 line
1.7.9.5