xfs: clean up log tickets and record writes v2
This series follows up on conversions about relogging infrastructure and the way xfs_log_done() does two things but only one of several callers uses both of those functions. It also pointed out that xfs_trans_commit() never writes to the log anymore, so only checkpoints pass a ticket to xlog_write() with this flag set and no transaction makes multiple calls to xlog_write() calls on the same ticket. Hence there's no real need for XLOG_TIC_INITED to track whether a ticket has written a start record to the log anymore.
A lot of further cleanups fell out of this. Once we no longer use XLOG_TIC_INITED to carry state inside the write loop, the logic can be simplified in both xlog_write and xfs_log_done. xfs_log_done can be split up, and then the call chain can be flattened because xlog_write_done() and xlog_commit_record() are basically the same.
This then leads to cleanups writing both commit and unmount records.
Finally, to complete what started all this, the XLOG_TIC_INITED flag is removed.
A git tree is avaiblable here:
git://git.infradead.org/users/hch/xfs.git xlog-ticket-cleanup.2
Gitweb:
http://git.infradead.org/users/hch/xfs.git/shortlog/refs/heads/xlog-ticket-cleanup.2
Changes since v1:
- taking this over from Dave (for now) as he is still injured, an it interacts closely with my log error handling bits
- rebased on top of for-next + the "more log cleanups" series
- fix an accounting error in xlog_write
- use a bool for the ticket header in xlog_write
- add a new patch to split xlog_ticket_done
PATCH 1/9 xfs: don't try to write a start record into every iclog
From: Dave Chinner <[email protected]>
The xlog_write() function iterates over iclogs until it completes writing all the log vectors passed in. The ticket tracks whether a start record has been written or not, so only the first iclog gets a start record. We only ever pass single use tickets to xlog_write() we only ever need to write a start record once per xlog_write() call.
Hence we don't need to store whether we should write a start record in the ticket as the callers provide all the information we need to determine if a start record should be written. For the moment, we have to ensure that we clear the XLOG_TIC_INITED appropriately so the code in xfs_log_done() still works correctly for committing transactions.
--- fs/xfs/xfs_log.c | 63 ++++++++++++++++++++++++------------------------ 1 file changed, 32 insertions(+), 31 deletions(-)
fs/xfs/xfs_log.c +32 -31
@@ -2112,23 +2112,21 @@ xlog_print_trans( }/* -* Calculate the potential space needed by the log vector. Each region gets -* its own xlog_op_header_t and may need to be double word aligned. +* Calculate the potential space needed by the log vector. We may need a start +* record, and each region gets its own struct xlog_op_header and may need to be +* double word aligned. */ static int xlog_write_calc_vec_length (struct xlog_ticket *ticket, -struct xfs_log_vec *log_vector) +struct xfs_log_vec *log_vector, +bool need_start_rec) {struct xfs_log_vec *lv; -int headers =0 ; +int headers = need_start_rec ?1 :0 ;int len =0 ;int i; -/* acct for start rec of xact */ -if (ticket->t_flags &XLOG_TIC_INITED ) - headers++; -for (lv = log_vector; lv; lv = lv->lv_next) {/* we don't write ordered log vectors */ if (lv->lv_buf_len ==XFS_LOG_VEC_ORDERED )
@@ -2150,27 +2148,16 @@ xlog_write_calc_vec_length(return len; } -/* -* If first write for transaction, insert start record We can't be trying to -* commit if we are inited. We can't have any "partial_copy" if we are inited. -*/ -static int +static void xlog_write_start_rec (struct xlog_op_header *ophdr,struct xlog_ticket *ticket) { -if (!(ticket->t_flags &XLOG_TIC_INITED )) -return 0 ; - ophdr->oh_tid =cpu_to_be32 (ticket->t_tid); ophdr->oh_clientid = ticket->t_clientid; ophdr->oh_len =0 ; ophdr->oh_flags =XLOG_START_TRANS ; ophdr->oh_res2 =0 ; - - ticket->t_flags &= ~XLOG_TIC_INITED ; - -return sizeof (struct xlog_op_header ); } static xlog_op_header_t *
@@ -2372,25 +2359,29 @@ xlog_write(int record_cnt =0 ;int data_cnt =0 ;int error =0 ; +bool need_start_rec = true; *start_lsn =0 ; - len =xlog_write_calc_vec_length (ticket, log_vector);/* * Region headers and bytes are already accounted for. * We only need to take into account start records and * split regions in this function. */ -if (ticket->t_flags &XLOG_TIC_INITED ) - ticket->t_curr_res -=sizeof (xlog_op_header_t); +if (ticket->t_flags &XLOG_TIC_INITED ) { + ticket->t_curr_res -=sizeof (struct xlog_op_header ); + ticket->t_flags &= ~XLOG_TIC_INITED ; + }/* * Commit record headers need to be accounted for. These * come in as separate writes so are easy to detect. */ -if (flags & (XLOG_COMMIT_TRANS |XLOG_UNMOUNT_TRANS )) - ticket->t_curr_res -=sizeof (xlog_op_header_t); +if (flags & (XLOG_COMMIT_TRANS |XLOG_UNMOUNT_TRANS )) { + ticket->t_curr_res -=sizeof (struct xlog_op_header ); + need_start_rec = false; + }if (ticket->t_curr_res <0 ) { xfs_alert_tag(log->l_mp, XFS_PTAG_LOGRES,
@@ -2399,6 +2390,8 @@ xlog_write(xfs_force_shutdown (log->l_mp,SHUTDOWN_LOG_IO_ERROR ); } + len =xlog_write_calc_vec_length (ticket, log_vector, need_start_rec); + index =0 ; lv = log_vector; vecp = lv->lv_iovecp;
@@ -2425,7 +2418,6 @@ xlog_write(while (lv && (!lv->lv_niovecs || index < lv->lv_niovecs)) {struct xfs_log_iovec *reg;struct xlog_op_header *ophdr; -int start_rec_copy;int copy_len;int copy_off;bool ordered = false;
@@ -2441,11 +2433,15 @@ xlog_write(ASSERT (reg->i_len %sizeof (int32_t ) ==0 );ASSERT ((unsigned long )ptr %sizeof (int32_t ) ==0 ); - start_rec_copy =xlog_write_start_rec (ptr, ticket); -if (start_rec_copy) { - record_cnt++; +/* +* Before we start formatting log vectors, we need to +* write a start record. Only do this for the first +* iclog we write to. +*/ +if (need_start_rec) { +xlog_write_start_rec (ptr, ticket);xlog_write_adv_cnt (&ptr, &len, &log_offset, - start_rec_copy); +sizeof (struct xlog_op_header )); } ophdr =xlog_write_setup_ophdr (log, ptr, ticket, flags);
@@ -2477,8 +2473,13 @@ xlog_write(xlog_write_adv_cnt (&ptr, &len, &log_offset, copy_len); } - copy_len += start_rec_copy +sizeof (xlog_op_header_t); + copy_len +=sizeof (struct xlog_op_header ); record_cnt++; +if (need_start_rec) { + copy_len +=sizeof (struct xlog_op_header ); + record_cnt++; + need_start_rec = false; + } data_cnt += contwr ? copy_len :0 ; error = xlog_write_copy_finish(log, iclog, flags,
Signature · 1 line
2.25.1
PATCH 2/9 xfs: re-order initial space accounting checks in xlog_write
In reply to Christoph HellwigFrom: Dave Chinner <[email protected]>
Commit and unmount records records do not need start records to be written, so rearrange the logic in xlog_write() to remove the need to check for XLOG_TIC_INITED to determine if we should account for the space used by a start record.
--- fs/xfs/xfs_log.c | 38 ++++++++++++-------------------------- 1 file changed, 12 insertions(+), 26 deletions(-)
fs/xfs/xfs_log.c +12 -26
@@ -2349,39 +2349,27 @@ xlog_write( uint flags) {struct xlog_in_core *iclog =NULL ; -struct xfs_log_iovec *vecp; -struct xfs_log_vec *lv; +struct xfs_log_vec *lv = log_vector; +struct xfs_log_iovec *vecp = lv->lv_iovecp; +int index =0 ;int len; -int index;int partial_copy =0 ;int partial_copy_len =0 ;int contwr =0 ;int record_cnt =0 ;int data_cnt =0 ;int error =0 ; -bool need_start_rec = true; - - *start_lsn =0 ; - +bool need_start_rec;/* -* Region headers and bytes are already accounted for. -* We only need to take into account start records and -* split regions in this function. +* If this is a commit or unmount transaction, we don't need a start +* record to be written. We do, however, have to account for the +* commit or unmount header that gets written. Hence we always have +* to account for an extra xlog_op_header here. */ -if (ticket->t_flags &XLOG_TIC_INITED ) { - ticket->t_curr_res -=sizeof (struct xlog_op_header ); + ticket->t_curr_res -=sizeof (struct xlog_op_header ); +if (ticket->t_flags &XLOG_TIC_INITED ) ticket->t_flags &= ~XLOG_TIC_INITED ; - } - -/* -* Commit record headers need to be accounted for. These -* come in as separate writes so are easy to detect. -*/ -if (flags & (XLOG_COMMIT_TRANS |XLOG_UNMOUNT_TRANS )) { - ticket->t_curr_res -=sizeof (struct xlog_op_header ); - need_start_rec = false; - }if (ticket->t_curr_res <0 ) { xfs_alert_tag(log->l_mp, XFS_PTAG_LOGRES,
@@ -2390,11 +2378,9 @@ xlog_write(xfs_force_shutdown (log->l_mp,SHUTDOWN_LOG_IO_ERROR ); } + need_start_rec = !(flags & (XLOG_COMMIT_TRANS |XLOG_UNMOUNT_TRANS )); len =xlog_write_calc_vec_length (ticket, log_vector, need_start_rec); - - index =0 ; - lv = log_vector; - vecp = lv->lv_iovecp; + *start_lsn =0 ;while (lv && (!lv->lv_niovecs || index < lv->lv_niovecs)) {void *ptr;int log_offset;
Signature · 1 line
2.25.1
PATCH 3/9 xfs: refactor and split xfs_log_done()
In reply to Christoph HellwigFrom: Dave Chinner <[email protected]>
xfs_log_done() does two separate things. Firstly, it triggers commit records to be written for permanent transactions, and secondly it releases or regrants transaction reservation space.
Since delayed logging was introduced, transactions no longer write directly to the log, hence they never have the XLOG_TIC_INITED flag cleared on them. Hence transactions never write commit records to the log and only need to modify reservation space.
Split up xfs_log_done into two parts, and only call the parts of the operation needed for the context xfs_log_done() is currently being called from.
--- fs/xfs/xfs_log.c | 65 +++++++++++++------------------------------ fs/xfs/xfs_log.h | 4 --- fs/xfs/xfs_log_cil.c | 13 +++++---- fs/xfs/xfs_log_priv.h | 16 +++++------ fs/xfs/xfs_trans.c | 24 ++++++++-------- 5 files changed, 47 insertions(+), 75 deletions(-)
fs/xfs/xfs_log.c +20 -45
@@ -487,62 +487,40 @@ xfs_log_reserve(*/ /* -* This routine is called when a user of a log manager ticket is done with -* the reservation. If the ticket was ever used, then a commit record for -* the associated transaction is written out as a log operation header with -* no data. The flag XLOG_TIC_INITED is set when the first write occurs with -* a given ticket. If the ticket was one with a permanent reservation, then -* a few operations are done differently. Permanent reservation tickets by -* default don't release the reservation. They just commit the current -* transaction with the belief that the reservation is still needed. A flag -* must be passed in before permanent reservations are actually released. -* When these type of tickets are not released, they need to be set into -* the inited state again. By doing this, a start record will be written -* out when the next write occurs. +* Write a commit record to the log to close off a running log write. */ -xfs_lsn_t -xfs_log_done ( -struct xfs_mount *mp, +int +xlog_write_done ( +struct xlog *log,struct xlog_ticket *ticket,struct xlog_in_core **iclog, -bool regrant) +xfs_lsn_t *lsn) { -struct xlog *log = mp->m_log; -xfs_lsn_t lsn =0 ; - -if (XLOG_FORCED_SHUTDOWN (log) || -/* -* If nothing was ever written, don't write out commit record. -* If we get an error, just continue and give back the log ticket. -*/ - (((ticket->t_flags &XLOG_TIC_INITED ) ==0 ) && - (xlog_commit_record (log, ticket, iclog, &lsn)))) { - lsn = (xfs_lsn_t) -1 ; - regrant = false; - } +if (XLOG_FORCED_SHUTDOWN (log)) +return -EIO ; +return xlog_commit_record (log, ticket, iclog, lsn); +} +/* +* Release or regrant the ticket reservation now the transaction is done with +* it depending on caller context. Rolling transactions need the ticket +* regranted, otherwise we release it completely. +*/ +void +xlog_ticket_done ( +struct xlog *log, +struct xlog_ticket *ticket, +bool regrant) +{if (!regrant) {trace_xfs_log_done_nonperm (log, ticket); - -/* -* Release ticket if not permanent reservation or a specific -* request has been made to release a permanent reservation. -*/ xlog_ungrant_log_space (log, ticket); }else {trace_xfs_log_done_perm (log, ticket); -xlog_regrant_reserve_log_space (log, ticket); -/* If this ticket was a permanent reservation and we aren't -* trying to release it, reset the inited flags; so next time -* we write, a start record will be written out. -*/ - ticket->t_flags |=XLOG_TIC_INITED ; } -xfs_log_ticket_put (ticket); -return lsn; } static bool
@@ -2368,9 +2346,6 @@ xlog_write(* to account for an extra xlog_op_header here. */ ticket->t_curr_res -=sizeof (struct xlog_op_header ); -if (ticket->t_flags &XLOG_TIC_INITED ) - ticket->t_flags &= ~XLOG_TIC_INITED ; -if (ticket->t_curr_res <0 ) {xfs_alert_tag (log->l_mp,XFS_PTAG_LOGRES ,"ctx ticket reservation ran out. Need to up reservation" );
fs/xfs/xfs_log.h +0 -4
@@ -105,10 +105,6 @@ struct xfs_log_item;struct xfs_item_ops ;struct xfs_trans ; -xfs_lsn_t xfs_log_done (struct xfs_mount *mp, -struct xlog_ticket *ticket, -struct xlog_in_core **iclog, -bool regrant);int xfs_log_force (struct xfs_mount *mp,uint flags);int xfs_log_force_lsn (struct xfs_mount *mp,xfs_lsn_t lsn,uint flags,int *log_forced);
fs/xfs/xfs_log_cil.c +7 -6
@@ -839,10 +839,11 @@ xlog_cil_push_work( }spin_unlock (&cil->xc_push_lock); -/* xfs_log_done always frees the ticket on error. */ - commit_lsn =xfs_log_done (log->l_mp, tic, &commit_iclog, false); -if (commit_lsn ==-1 ) - goto out_abort; + error =xlog_write_done (log, tic, &commit_iclog, &commit_lsn); +if (error) + goto out_abort_free_ticket; + +xlog_ticket_done (log, tic, false);spin_lock (&commit_iclog->ic_callback_lock);if (commit_iclog->ic_state ==XLOG_STATE_IOERROR ) {
@@ -875,7 +876,7 @@ xlog_cil_push_work(return ; out_abort_free_ticket: -xfs_log_ticket_put (tic); +xlog_ticket_done (log, tic, false); out_abort:ASSERT (XLOG_FORCED_SHUTDOWN (log));xlog_cil_committed (ctx);
@@ -1007,7 +1008,7 @@ xfs_log_commit_cil(if (commit_lsn) *commit_lsn = xc_commit_lsn; -xfs_log_done (mp, tp->t_ticket,NULL , regrant); +xlog_ticket_done (log, tp->t_ticket, regrant); tp->t_ticket =NULL ;xfs_trans_unreserve_and_mod_sb (tp);
fs/xfs/xfs_log_priv.h +8 -8
@@ -439,14 +439,14 @@ xlog_write_adv_cnt(void **ptr, int *len, int *off, size_t bytes)void xlog_print_tic_res (struct xfs_mount *mp,struct xlog_ticket *ticket);void xlog_print_trans (struct xfs_trans *); -int -xlog_write ( -struct xlog *log, -struct xfs_log_vec *log_vector, -struct xlog_ticket *tic, -xfs_lsn_t *start_lsn, -struct xlog_in_core **commit_iclog, -uint flags); + +int xlog_write (struct xlog *log,struct xfs_log_vec *log_vector, +struct xlog_ticket *tic,xfs_lsn_t *start_lsn, +struct xlog_in_core **commit_iclog,uint flags); +int xlog_write_done (struct xlog *log,struct xlog_ticket *ticket, +struct xlog_in_core **iclog,xfs_lsn_t *lsn); +void xlog_ticket_done (struct xlog *log,struct xlog_ticket *ticket, +bool regrant);/* * When we crack an atomic LSN, we sample it first so that the value will not
fs/xfs/xfs_trans.c +12 -12
@@ -9,6 +9,7 @@#include "xfs_shared.h" #include "xfs_format.h" #include "xfs_log_format.h" +#include "xfs_log_priv.h" #include "xfs_trans_resv.h" #include "xfs_mount.h" #include "xfs_extent_busy.h"
@@ -150,8 +151,9 @@ xfs_trans_reserve( uint blocks, uint rtextents) { - int error = 0; -bool rsvd = (tp->t_flags &XFS_TRANS_RESERVE ) !=0 ; +struct xfs_mount *mp = tp->t_mountp; +int error =0 ; +bool rsvd = (tp->t_flags &XFS_TRANS_RESERVE ) !=0 ;/* Mark this thread as being in a transaction */ current_set_flags_nested (&tp->t_pflags,PF_MEMALLOC_NOFS );
@@ -162,7 +164,7 @@ xfs_trans_reserve(* fail if the count would go below zero. */ if (blocks >0 ) { - error =xfs_mod_fdblocks (tp->t_mountp, -((int64_t )blocks), rsvd); + error =xfs_mod_fdblocks (mp, -((int64_t )blocks), rsvd);if (error !=0 ) {current_restore_flags_nested (&tp->t_pflags,PF_MEMALLOC_NOFS );return -ENOSPC ;
@@ -191,9 +193,9 @@ xfs_trans_reserve(if (tp->t_ticket !=NULL ) {ASSERT (resp->tr_logflags &XFS_TRANS_PERM_LOG_RES ); - error =xfs_log_regrant (tp->t_mountp, tp->t_ticket); + error =xfs_log_regrant (mp, tp->t_ticket); }else { - error = xfs_log_reserve(tp->t_mountp, + error = xfs_log_reserve(mp, resp->tr_logres, resp->tr_logcount, &tp->t_ticket, XFS_TRANSACTION,
@@ -213,7 +215,7 @@ xfs_trans_reserve(* fail if the count would go below zero. */ if (rtextents >0 ) { - error =xfs_mod_frextents (tp->t_mountp, -((int64_t )rtextents)); + error =xfs_mod_frextents (mp, -((int64_t )rtextents));if (error) { error = -ENOSPC ; goto undo_log;
@@ -229,7 +231,7 @@ xfs_trans_reserve(*/ undo_log:if (resp->tr_logres >0 ) { -xfs_log_done (tp->t_mountp, tp->t_ticket,NULL , false); +xlog_ticket_done (mp->m_log, tp->t_ticket, false); tp->t_ticket =NULL ; tp->t_log_res =0 ; tp->t_flags &= ~XFS_TRANS_PERM_LOG_RES ;
@@ -237,7 +239,7 @@ xfs_trans_reserve( undo_blocks:if (blocks >0 ) { -xfs_mod_fdblocks (tp->t_mountp, (int64_t )blocks, rsvd); +xfs_mod_fdblocks (mp, (int64_t )blocks, rsvd); tp->t_blk_res =0 ; }
@@ -999,9 +1001,7 @@ __xfs_trans_commit(*/ xfs_trans_unreserve_and_mod_dquots (tp);if (tp->t_ticket) { - commit_lsn =xfs_log_done (mp, tp->t_ticket,NULL , regrant); -if (commit_lsn ==-1 && !error) - error = -EIO ; +xlog_ticket_done (mp->m_log, tp->t_ticket, regrant); tp->t_ticket =NULL ; }current_restore_flags_nested (&tp->t_pflags,PF_MEMALLOC_NOFS );
@@ -1060,7 +1060,7 @@ xfs_trans_cancel(xfs_trans_unreserve_and_mod_dquots (tp);if (tp->t_ticket) { -xfs_log_done (mp, tp->t_ticket,NULL , false); +xlog_ticket_done (mp->m_log, tp->t_ticket, false); tp->t_ticket =NULL ; }
Signature · 1 line
2.25.1
PATCH 4/9 xfs: split xlog_ticket_done
In reply to Christoph HellwigSplit the regrant case out of xlog_ticket_done and into a new xlog_ticket_regrant helper. Merge both functions with the low-level functions implementing the actual functionality and adjust the tracepoints.
--- fs/xfs/xfs_log.c | 84 ++++++++++++++----------------------------- fs/xfs/xfs_log_cil.c | 9 +++-- fs/xfs/xfs_log_priv.h | 4 +-- fs/xfs/xfs_trace.h | 14 ++++---- fs/xfs/xfs_trans.c | 9 +++-- 5 files changed, 47 insertions(+), 73 deletions(-)
fs/xfs/xfs_log.c +27 -57
@@ -66,14 +66,6 @@ xlog_grant_push_ail(struct xlog *log, int need_bytes); STATIC void -xlog_regrant_reserve_log_space ( -struct xlog *log, -struct xlog_ticket *ticket); -STATIC void -xlog_ungrant_log_space ( -struct xlog *log, -struct xlog_ticket *ticket); -STATIC voidxlog_sync (struct xlog *log,struct xlog_in_core *iclog);
@@ -502,27 +494,6 @@ xlog_write_done(return xlog_commit_record (log, ticket, iclog, lsn); } -/* -* Release or regrant the ticket reservation now the transaction is done with -* it depending on caller context. Rolling transactions need the ticket -* regranted, otherwise we release it completely. -*/ -void -xlog_ticket_done ( -struct xlog *log, -struct xlog_ticket *ticket, -bool regrant) -{ -if (!regrant) { -trace_xfs_log_done_nonperm (log, ticket); -xlog_ungrant_log_space (log, ticket); - }else { -trace_xfs_log_done_perm (log, ticket); -xlog_regrant_reserve_log_space (log, ticket); - } -xfs_log_ticket_put (ticket); -} - static bool __xlog_state_release_iclog( struct xlog *log,
@@ -921,8 +892,7 @@ xfs_log_write_unmount_record(if (tic) {trace_xfs_log_umount_write (log, tic); -xlog_ungrant_log_space (log, tic); -xfs_log_ticket_put (tic); +xlog_ticket_done (log, tic); } }
@@ -2987,19 +2957,18 @@ xlog_state_get_iclog_space(return 0 ; }/* xlog_state_get_iclog_space */ -/* The first cnt-1 times through here we don't need to -* move the grant write head because the permanent -* reservation has reserved cnt times the unit amount. -* Release part of current permanent unit reservation and -* reset current reservation to be one units worth. Also -* move grant reservation head forward. +/* +* The first cnt-1 times through here we don't need to move the grant write head +* because the permanent reservation has reserved cnt times the unit amount. +* Release part of current permanent unit reservation and reset current +* reservation to be one units worth. Also move grant reservation head forward. */ -STATIC void -xlog_regrant_reserve_log_space ( +void +xlog_ticket_regrant (struct xlog *log,struct xlog_ticket *ticket) { -trace_xfs_log_regrant_reserve_enter (log, ticket); +trace_xfs_log_ticket_regrant (log, ticket);if (ticket->t_cnt >0 ) ticket->t_cnt--;
@@ -3011,21 +2980,20 @@ xlog_regrant_reserve_log_space( ticket->t_curr_res = ticket->t_unit_res;xlog_tic_reset_res (ticket); -trace_xfs_log_regrant_reserve_sub (log, ticket); +trace_xfs_log_ticket_regrant_sub (log, ticket);/* just return if we still have some of the pre-reserved space */ -if (ticket->t_cnt >0 ) -return ; +if (!ticket->t_cnt) { +xlog_grant_add_space (log, &log->l_reserve_head.grant, + ticket->t_unit_res); +trace_xfs_log_ticket_regrant_exit (log, ticket); -xlog_grant_add_space (log, &log->l_reserve_head.grant, - ticket->t_unit_res); - -trace_xfs_log_regrant_reserve_exit (log, ticket); - - ticket->t_curr_res = ticket->t_unit_res; -xlog_tic_reset_res (ticket); -}/* xlog_regrant_reserve_log_space */ + ticket->t_curr_res = ticket->t_unit_res; +xlog_tic_reset_res (ticket); + } +xfs_log_ticket_put (ticket); +}/* * Give back the space left from a reservation.
@@ -3041,18 +3009,19 @@ xlog_regrant_reserve_log_space(* space, the count will stay at zero and the only space remaining will be * in the current reservation field. */ -STATIC void -xlog_ungrant_log_space ( +void +xlog_ticket_done (struct xlog *log,struct xlog_ticket *ticket) { -int bytes; +int bytes; + +trace_xfs_log_ticket_done (log, ticket);if (ticket->t_cnt >0 ) ticket->t_cnt--; -trace_xfs_log_ungrant_enter (log, ticket); -trace_xfs_log_ungrant_sub (log, ticket); +trace_xfs_log_ticket_done_sub (log, ticket);/* * If this is a permanent reservation ticket, we may be able to free
@@ -3067,9 +3036,10 @@ xlog_ungrant_log_space(xlog_grant_sub_space (log, &log->l_reserve_head.grant, bytes);xlog_grant_sub_space (log, &log->l_write_head.grant, bytes); -trace_xfs_log_ungrant_exit (log, ticket); +trace_xfs_log_ticket_done_exit (log, ticket);xfs_log_space_wake (log->l_mp); +xfs_log_ticket_put (ticket); }/*
fs/xfs/xfs_log_cil.c +6 -3
@@ -843,7 +843,7 @@ xlog_cil_push_work(if (error) goto out_abort_free_ticket; -xlog_ticket_done (log, tic, false); +xlog_ticket_done (log, tic);spin_lock (&commit_iclog->ic_callback_lock);if (commit_iclog->ic_state ==XLOG_STATE_IOERROR ) {
@@ -876,7 +876,7 @@ xlog_cil_push_work(return ; out_abort_free_ticket: -xlog_ticket_done (log, tic, false); +xlog_ticket_done (log, tic); out_abort:ASSERT (XLOG_FORCED_SHUTDOWN (log));xlog_cil_committed (ctx);
@@ -1008,7 +1008,10 @@ xfs_log_commit_cil(if (commit_lsn) *commit_lsn = xc_commit_lsn; -xlog_ticket_done (log, tp->t_ticket, regrant); +if (regrant) +xlog_ticket_regrant (log, tp->t_ticket); +else +xlog_ticket_done (log, tp->t_ticket); tp->t_ticket =NULL ;xfs_trans_unreserve_and_mod_sb (tp);
fs/xfs/xfs_log_priv.h +2 -2
@@ -445,8 +445,8 @@ int xlog_write(struct xlog *log, struct xfs_log_vec *log_vector, struct xlog_in_core **commit_iclog, uint flags);int xlog_write_done (struct xlog *log,struct xlog_ticket *ticket,struct xlog_in_core **iclog,xfs_lsn_t *lsn); -void xlog_ticket_done (struct xlog *log,struct xlog_ticket *ticket, -bool regrant); +void xlog_ticket_done (struct xlog *log,struct xlog_ticket *ticket); +void xlog_ticket_regrant (struct xlog *log,struct xlog_ticket *ticket);/* * When we crack an atomic LSN, we sample it first so that the value will not
fs/xfs/xfs_trace.h +6 -8
@@ -1001,8 +1001,6 @@ DECLARE_EVENT_CLASS(xfs_loggrant_class,DEFINE_EVENT (xfs_loggrant_class, name, \ TP_PROTO(struct xlog *log, struct xlog_ticket *tic), \ TP_ARGS(log, tic)) -DEFINE_LOGGRANT_EVENT (xfs_log_done_nonperm); -DEFINE_LOGGRANT_EVENT (xfs_log_done_perm);DEFINE_LOGGRANT_EVENT (xfs_log_umount_write);DEFINE_LOGGRANT_EVENT (xfs_log_grant_sleep);DEFINE_LOGGRANT_EVENT (xfs_log_grant_wake);
@@ -1011,12 +1009,12 @@ DEFINE_LOGGRANT_EVENT(xfs_log_reserve);DEFINE_LOGGRANT_EVENT (xfs_log_reserve_exit);DEFINE_LOGGRANT_EVENT (xfs_log_regrant);DEFINE_LOGGRANT_EVENT (xfs_log_regrant_exit); -DEFINE_LOGGRANT_EVENT (xfs_log_regrant_reserve_enter); -DEFINE_LOGGRANT_EVENT (xfs_log_regrant_reserve_exit); -DEFINE_LOGGRANT_EVENT (xfs_log_regrant_reserve_sub); -DEFINE_LOGGRANT_EVENT (xfs_log_ungrant_enter); -DEFINE_LOGGRANT_EVENT (xfs_log_ungrant_exit); -DEFINE_LOGGRANT_EVENT (xfs_log_ungrant_sub); +DEFINE_LOGGRANT_EVENT (xfs_log_ticket_regrant); +DEFINE_LOGGRANT_EVENT (xfs_log_ticket_regrant_exit); +DEFINE_LOGGRANT_EVENT (xfs_log_ticket_regrant_sub); +DEFINE_LOGGRANT_EVENT (xfs_log_ticket_done); +DEFINE_LOGGRANT_EVENT (xfs_log_ticket_done_sub); +DEFINE_LOGGRANT_EVENT (xfs_log_ticket_done_exit); DECLARE_EVENT_CLASS(xfs_log_item_class, TP_PROTO(struct xfs_log_item *lip),
fs/xfs/xfs_trans.c +6 -3
@@ -231,7 +231,7 @@ xfs_trans_reserve(*/ undo_log:if (resp->tr_logres >0 ) { -xlog_ticket_done (mp->m_log, tp->t_ticket, false); +xlog_ticket_done (mp->m_log, tp->t_ticket); tp->t_ticket =NULL ; tp->t_log_res =0 ; tp->t_flags &= ~XFS_TRANS_PERM_LOG_RES ;
@@ -1001,7 +1001,10 @@ __xfs_trans_commit(*/ xfs_trans_unreserve_and_mod_dquots (tp);if (tp->t_ticket) { -xlog_ticket_done (mp->m_log, tp->t_ticket, regrant); +if (regrant) +xlog_ticket_regrant (mp->m_log, tp->t_ticket); +else +xlog_ticket_done (mp->m_log, tp->t_ticket); tp->t_ticket =NULL ; }current_restore_flags_nested (&tp->t_pflags,PF_MEMALLOC_NOFS );
@@ -1060,7 +1063,7 @@ xfs_trans_cancel(xfs_trans_unreserve_and_mod_dquots (tp);if (tp->t_ticket) { -xlog_ticket_done (mp->m_log, tp->t_ticket, false); +xlog_ticket_done (mp->m_log, tp->t_ticket); tp->t_ticket =NULL ; }
Signature · 1 line
2.25.1
Re: [PATCH 4/9] xfs: split xlog_ticket_done
Quoted 5 lines · Christoph Hellwig
Split the regrant case out of
xlog_ticket_doneand into a newxlog_ticket_regranthelper. Merge both functions with the low-level functions implementing the actual functionality and adjust the tracepoints.
Quoted 8 lines
Signed-off-by Christoph Hellwig <[email protected]>--- fs/xfs/xfs_log.c | 84 ++++++++++++++----------------------------- fs/xfs/xfs_log_cil.c | 9 +++-- fs/xfs/xfs_log_priv.h | 4 +-- fs/xfs/xfs_trace.h | 14 ++++---- fs/xfs/xfs_trans.c | 9 +++-- 5 files changed, 47 insertions(+), 73 deletions(-)
Nice. That unwinds it quite well.
Signature · 2 lines
Dave Chinner [email protected]
PATCH 5/9 xfs: merge xlog_commit_record with xlog_write_done()
In reply to Christoph HellwigFrom: Dave Chinner <[email protected]>
xlog_write_done() is just a thin wrapper around xlog_commit_record(), so they can be merged together easily. Convert all the xlog_commit_record() callers to use xlog_write_done() and merge the implementations.
--- fs/xfs/xfs_log.c | 44 +++++++++++--------------------------------- 1 file changed, 11 insertions(+), 33 deletions(-)
fs/xfs/xfs_log.c +11 -33
@@ -24,13 +24,6 @@kmem_zone_t *xfs_log_ticket_zone;/* Local miscellaneous function prototypes */ -STATIC int -xlog_commit_record ( -struct xlog *log, -struct xlog_ticket *ticket, -struct xlog_in_core **iclog, -xfs_lsn_t *commitlsnp); - STATIC struct xlog * xlog_alloc_log( struct xfs_mount *mp,
@@ -478,22 +471,6 @@ xfs_log_reserve(* marked as with WANT_SYNC. */ -/* -* Write a commit record to the log to close off a running log write. -*/ -int -xlog_write_done ( -struct xlog *log, -struct xlog_ticket *ticket, -struct xlog_in_core **iclog, -xfs_lsn_t *lsn) -{ -if (XLOG_FORCED_SHUTDOWN (log)) -return -EIO ; - -return xlog_commit_record (log, ticket, iclog, lsn); -} - static bool __xlog_state_release_iclog( struct xlog *log,
@@ -1463,20 +1440,17 @@ xlog_alloc_log(return ERR_PTR (error); }/* xlog_alloc_log */ -/* * Write out the commit record of a transaction associated with the given -* ticket. Return the lsn of the commit record. +* ticket to close off a running log write. Return the lsn of the commit record. */ -STATIC int -xlog_commit_record ( +int +xlog_write_done (struct xlog *log,struct xlog_ticket *ticket,struct xlog_in_core **iclog, -xfs_lsn_t *commitlsnp) +xfs_lsn_t *lsn) { -struct xfs_mount *mp = log->l_mp; -int error;struct xfs_log_iovec reg = { .i_addr =NULL , .i_len =0 ,
@@ -1486,12 +1460,16 @@ xlog_commit_record( .lv_niovecs = 1, .lv_iovecp = ®, }; +int error;ASSERT_ALWAYS (iclog); - error =xlog_write (log, &vec, ticket, commitlsnp, iclog, -XLOG_COMMIT_TRANS ); + +if (XLOG_FORCED_SHUTDOWN (log)) +return -EIO ; + + error =xlog_write (log, &vec, ticket, lsn, iclog,XLOG_COMMIT_TRANS );if (error) -xfs_force_shutdown (mp,SHUTDOWN_LOG_IO_ERROR ); +xfs_force_shutdown (log->l_mp,SHUTDOWN_LOG_IO_ERROR );return error; }
Signature · 1 line
2.25.1
PATCH 6/9 xfs: factor out unmount record writing
In reply to Christoph HellwigFrom: Dave Chinner <[email protected]>
Separate out the unmount record writing from the rest of the ticket and log state futzing necessary to make it work. This is a no-op, just makes the code cleaner and places the unmount record formatting and writing alongside the commit record formatting and writing code.
We can also get rid of the ticket flag clearing before the xlog_write() call because it no longer cares about the state of XLOG_TIC_INITED.
--- fs/xfs/xfs_log.c | 59 ++++++++++++++++++++++++++++-------------------- 1 file changed, 35 insertions(+), 24 deletions(-)
fs/xfs/xfs_log.c +35 -24
@@ -471,6 +471,38 @@ xfs_log_reserve(* marked as with WANT_SYNC. */ +/* +* Write out an unmount record using the ticket provided. We have to account for +* the data space used in the unmount ticket as this write is not done from a +* transaction context that has already done the accounting for us. +*/ +static int +xlog_write_unmount ( +struct xlog *log, +struct xlog_ticket *ticket, +xfs_lsn_t *lsn, +uint flags) +{ +/* the data section must be 32 bit size aligned */ +struct xfs_unmount_log_format magic = { + .magic =XLOG_UNMOUNT_TYPE , + }; +struct xfs_log_iovec reg = { + .i_addr = &magic, + .i_len =sizeof (magic), + .i_type =XLOG_REG_TYPE_UNMOUNT , + }; +struct xfs_log_vec vec = { + .lv_niovecs =1 , + .lv_iovecp = ®, + }; + +/* account for space used by record data */ + ticket->t_curr_res -=sizeof (magic); + +return xlog_write (log, &vec, ticket, lsn,NULL , flags); +} + static bool __xlog_state_release_iclog( struct xlog *log,
@@ -795,31 +827,13 @@ xlog_wait_on_iclog( }/* -* Final log writes as part of unmount. -* -* Mark the filesystem clean as unmount happens. Note that during relocation -* this routine needs to be executed as part of source-bag while the -* deallocation must not be done until source-end. +* Mark the filesystem clean by writing an unmount record to the head of the +* log. */ - -/* Actually write the unmount record to disk. */ static void xfs_log_write_unmount_record (struct xfs_mount *mp) { -/* the data section must be 32 bit size aligned */ -struct xfs_unmount_log_format magic = { - .magic =XLOG_UNMOUNT_TYPE , - }; -struct xfs_log_iovec reg = { - .i_addr = &magic, - .i_len =sizeof (magic), - .i_type =XLOG_REG_TYPE_UNMOUNT , - }; -struct xfs_log_vec vec = { - .lv_niovecs =1 , - .lv_iovecp = ®, - };struct xlog *log = mp->m_log;struct xlog_in_core *iclog;struct xlog_ticket *tic =NULL ;
@@ -844,10 +858,7 @@ xfs_log_write_unmount_record( flags &= ~XLOG_UNMOUNT_TRANS ; } -/* remove inited flag, and account for space used */ - tic->t_flags =0 ; - tic->t_curr_res -=sizeof (magic); - error =xlog_write (log, &vec, tic, &lsn,NULL , flags); + error =xlog_write_unmount (log, tic, &lsn, flags);/* * At this point, we're umounting anyway, so there's no point in * transitioning log state to IOERROR. Just continue...
Signature · 1 line
2.25.1
PATCH 7/9 xfs: rename the log unmount writing functions.
In reply to Christoph HellwigFrom: Dave Chinner <[email protected]>
The naming and calling conventions are a bit of a mess. Clean it up so the call chain looks like:
xfs_log_unmount_write(mp)
xlog_unmount_write(log)
xlog_write_unmount_record(log, ticket)
--- fs/xfs/xfs_log.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
fs/xfs/xfs_log.c +6 -6
@@ -477,7 +477,7 @@ xfs_log_reserve(* transaction context that has already done the accounting for us. */ static int -xlog_write_unmount( +xlog_write_unmount_record( struct xlog *log, struct xlog_ticket *ticket, xfs_lsn_t *lsn,
@@ -831,10 +831,10 @@ xlog_wait_on_iclog(* log. */ static void -xfs_log_write_unmount_record ( -struct xfs_mount *mp) +xlog_unmount_write ( +struct xlog *log) { -struct xlog *log = mp->m_log; +struct xfs_mount *mp = log->l_mp;struct xlog_in_core *iclog;struct xlog_ticket *tic =NULL ;xfs_lsn_t lsn;
@@ -858,7 +858,7 @@ xfs_log_write_unmount_record( flags &= ~XLOG_UNMOUNT_TRANS ; } - error =xlog_write_unmount (log, tic, &lsn, flags); + error =xlog_write_unmount_record (log, tic, &lsn, flags);/* * At this point, we're umounting anyway, so there's no point in * transitioning log state to IOERROR. Just continue...
@@ -924,7 +924,7 @@ xfs_log_unmount_write(if (XLOG_FORCED_SHUTDOWN (log))return ;xfs_log_unmount_verify_iclog (log); -xfs_log_write_unmount_record (mp); +xlog_unmount_write (log); }/*
Signature · 1 line
2.25.1
PATCH 8/9 xfs: remove some stale comments from the log code
In reply to Christoph HellwigFrom: Dave Chinner <[email protected]>
--- fs/xfs/xfs_log.c | 59 +++++++++++------------------------------------- 1 file changed, 13 insertions(+), 46 deletions(-)
fs/xfs/xfs_log.c +13 -46
@@ -463,14 +463,6 @@ xfs_log_reserve(return error; } - -/* -* NOTES: -* -* 1. currblock field gets updated at startup and after in-core logs -* marked as with WANT_SYNC. -*/ -/* * Write out an unmount record using the ticket provided. We have to account for * the data space used in the unmount ticket as this write is not done from a
@@ -1912,7 +1904,7 @@ xlog_dealloc_log( log->l_mp->m_log =NULL ;destroy_workqueue (log->l_ioend_workqueue);kmem_free (log); -}/* xlog_dealloc_log */ +}/* * Update counters atomically now that memcpy is done.
@@ -2456,14 +2448,6 @@ xlog_write(return error; } - -/***************************************************************************** -* -* State Machine functions -* -***************************************************************************** -*/ - static void xlog_state_activate_iclog( struct xlog_in_core *iclog,
@@ -2824,7 +2808,7 @@ xlog_state_done_syncing(*/ wake_up_all (&iclog->ic_write_wait);spin_unlock (&log->l_icloglock); -xlog_state_do_callback (log);/* also cleans log */ +xlog_state_do_callback (log); }/*
@@ -2944,13 +2928,14 @@ xlog_state_get_iclog_space( *logoffsetp = log_offset;return 0 ; -}/* xlog_state_get_iclog_space */ +}/* -* The first cnt-1 times through here we don't need to move the grant write head -* because the permanent reservation has reserved cnt times the unit amount. -* Release part of current permanent unit reservation and reset current -* reservation to be one units worth. Also move grant reservation head forward. +* The first cnt-1 times a ticket goes through here we don't need to move the +* grant write head because the permanent reservation has reserved cnt times the +* unit amount. Release part of current permanent unit reservation and reset +* current reservation to be one units worth. Also move grant reservation head +* forward. */ void xlog_ticket_regrant(
@@ -3032,12 +3017,8 @@ xlog_ticket_done( }/* -* Mark the current iclog in the ring as WANT_SYNC and move the current iclog -* pointer to the next iclog in the ring. -* -* When called from xlog_state_get_iclog_space(), the exact size of the iclog -* has not yet been determined, all we know is that we have run out of space in -* the current iclog. +* This routine will mark the current iclog in the ring as WANT_SYNC and move +* the current iclog pointer to the next iclog in the ring. */ STATIC void xlog_state_switch_iclogs(
@@ -3082,7 +3063,7 @@ xlog_state_switch_iclogs( }ASSERT (iclog == log->l_iclog); log->l_iclog = iclog->ic_next; -}/* xlog_state_switch_iclogs */ +}/* * Write out all data in the in-core log as of this exact moment in time.
@@ -3289,13 +3270,6 @@ xfs_log_force_lsn(return ret; } -/***************************************************************************** -* -* TICKET functions -* -***************************************************************************** -*/ -/* * Free a used ticket when its refcount falls to zero. */
@@ -3453,13 +3427,6 @@ xlog_ticket_alloc(return tic; } - -/****************************************************************************** -* -* Log debug routines -* -****************************************************************************** -*/ #if defined(DEBUG )/* * Make sure that the destination ptr is within the valid data region of
@@ -3545,7 +3512,7 @@ xlog_verify_tail_lsn(if (blocks <BTOBB (iclog->ic_offset) +1 )xfs_emerg (log->l_mp,"%s: ran out of log space" , __func__); } -}/* xlog_verify_tail_lsn */ +}/* * Perform a number of checks on the iclog before writing to disk.
@@ -3648,7 +3615,7 @@ xlog_verify_iclog( } ptr +=sizeof (xlog_op_header_t) + op_len; } -}/* xlog_verify_iclog */ +}#endif /*
Signature · 1 line
2.25.1
PATCH 9/9 xfs: kill XLOG_TIC_INITED
In reply to Christoph HellwigFrom: Dave Chinner <[email protected]>
It is not longer used or checked by anything, so remove the last traces from the log ticket code.
--- fs/xfs/xfs_log.c | 1 - fs/xfs/xfs_log_priv.h | 6 ++---- 2 files changed, 2 insertions(+), 5 deletions(-)
fs/xfs/xfs_log.c +0 -1
@@ -3418,7 +3418,6 @@ xlog_ticket_alloc( tic->t_ocnt = cnt; tic->t_tid =prandom_u32 (); tic->t_clientid = client; - tic->t_flags =XLOG_TIC_INITED ;if (permanent) tic->t_flags |=XLOG_TIC_PERM_RESERV ;
fs/xfs/xfs_log_priv.h +2 -4
@@ -51,13 +51,11 @@ enum xlog_iclog_state { };/* -* Flags to log ticket +* Log ticket flags */ -#define XLOG_TIC_INITED 0x1/* has been initialized */ -#define XLOG_TIC_PERM_RESERV 0x2/* permanent reservation */ +#define XLOG_TIC_PERM_RESERV 0x1/* permanent reservation */ #define XLOG_TIC_FLAGS \ - { XLOG_TIC_INITED, "XLOG_TIC_INITED" }, \ { XLOG_TIC_PERM_RESERV, "XLOG_TIC_PERM_RESERV" }/*
Signature · 1 line
2.25.1
Quoted 29 lines · Christoph Hellwig
This series follows up on conversions about relogging infrastructure and the way
xfs_log_done()does two things but only one of several callers uses both of those functions. It also pointed out thatxfs_trans_commit()never writes to the log anymore, so only checkpoints pass a ticket toxlog_write()with this flag set and no transaction makes multiple calls toxlog_write()calls on the same ticket. Hence there's no real need forXLOG_TIC_INITEDto track whether a ticket has written a start record to the log anymore.A lot of further cleanups fell out of this. Once we no longer use
XLOG_TIC_INITEDto carry state inside the write loop, the logic can be simplified in bothxlog_writeandxfs_log_done.xfs_log_donecan be split up, and then the call chain can be flattened becausexlog_write_done()andxlog_commit_record()are basically the same.This then leads to cleanups writing both commit and unmount records.
Finally, to complete what started all this, the
XLOG_TIC_INITEDflag is removed.A git tree is avaiblable here:
git://git.infradead.org/users/hch/xfs.gitxlog-ticket-cleanup.2Gitweb:
http://git.infradead.org/users/hch/xfs.git/shortlog/refs/heads/xlog-ticket-cleanup.2
Quoted 8 lines
Changes since v1:
- taking this over from Dave (for now) as he is still injured, an it interacts closely with my log error handling bits
- rebased on top of for-next + the "more log cleanups" series
- fix an accounting error inxlog_write
- use a bool for the ticket header inxlog_write
- add a new patch to splitxlog_ticket_done
This seems to ignore various bits of (trivial) feedback from v1 as well as drops all reviews...
Brian
Or, I'll send another one with those incorporated.