From eed0f953b90e86e765197a1dad06bb48aedc27fe Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Sun, 8 Dec 2019 13:12:49 +0000 Subject: [PATCH 1/9] gfs2: Another gfs2_find_jhead fix On filesystems with a block size smaller than the page size, gfs2_find_jhead can split a page across two bios (for example, when blocks are not allocated consecutively). When that happens, the first bio that completes will unlock the page in its bi_end_io handler even though the page hasn't been read completely yet. Fix that by using a chained bio for the rest of the page. While at it, clean up the sector calculation logic in gfs2_log_alloc_bio. In gfs2_find_jhead, simplify the disk block and offset calculation logic and fix a variable name. Fixes: f4686c26ecc3 ("gfs2: read journal in large chunks") Cc: stable@vger.kernel.org # v5.2+ Signed-off-by: Andreas Gruenbacher --- fs/gfs2/lops.c | 70 ++++++++++++++++++++++++++++++++------------------ 1 file changed, 45 insertions(+), 25 deletions(-) diff --git a/fs/gfs2/lops.c b/fs/gfs2/lops.c index 55fed7daf2b1..aa0359b7e2ac 100644 --- a/fs/gfs2/lops.c +++ b/fs/gfs2/lops.c @@ -259,7 +259,7 @@ static struct bio *gfs2_log_alloc_bio(struct gfs2_sbd *sdp, u64 blkno, struct super_block *sb = sdp->sd_vfs; struct bio *bio = bio_alloc(GFP_NOIO, BIO_MAX_PAGES); - bio->bi_iter.bi_sector = blkno * (sb->s_blocksize >> 9); + bio->bi_iter.bi_sector = blkno << (sb->s_blocksize_bits - 9); bio_set_dev(bio, sb->s_bdev); bio->bi_end_io = end_io; bio->bi_private = sdp; @@ -472,6 +472,20 @@ static void gfs2_jhead_process_page(struct gfs2_jdesc *jd, unsigned long index, put_page(page); /* Once more for find_or_create_page */ } +static struct bio *gfs2_chain_bio(struct bio *prev, unsigned int nr_iovecs) +{ + struct bio *new; + + new = bio_alloc(GFP_NOIO, nr_iovecs); + bio_copy_dev(new, prev); + new->bi_iter.bi_sector = bio_end_sector(prev); + new->bi_opf = prev->bi_opf; + new->bi_write_hint = prev->bi_write_hint; + bio_chain(new, prev); + submit_bio(prev); + return new; +} + /** * gfs2_find_jhead - find the head of a log * @jd: The journal descriptor @@ -488,15 +502,15 @@ int gfs2_find_jhead(struct gfs2_jdesc *jd, struct gfs2_log_header_host *head, struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode); struct address_space *mapping = jd->jd_inode->i_mapping; unsigned int block = 0, blocks_submitted = 0, blocks_read = 0; - unsigned int bsize = sdp->sd_sb.sb_bsize; + unsigned int bsize = sdp->sd_sb.sb_bsize, off; unsigned int bsize_shift = sdp->sd_sb.sb_bsize_shift; unsigned int shift = PAGE_SHIFT - bsize_shift; - unsigned int readhead_blocks = BIO_MAX_PAGES << shift; + unsigned int readahead_blocks = BIO_MAX_PAGES << shift; struct gfs2_journal_extent *je; int sz, ret = 0; struct bio *bio = NULL; struct page *page = NULL; - bool done = false; + bool bio_chained = false, done = false; errseq_t since; memset(head, 0, sizeof(*head)); @@ -505,9 +519,9 @@ int gfs2_find_jhead(struct gfs2_jdesc *jd, struct gfs2_log_header_host *head, since = filemap_sample_wb_err(mapping); list_for_each_entry(je, &jd->extent_list, list) { - for (; block < je->lblock + je->blocks; block++) { - u64 dblock; + u64 dblock = je->dblock; + for (; block < je->lblock + je->blocks; block++, dblock++) { if (!page) { page = find_or_create_page(mapping, block >> shift, GFP_NOFS); @@ -516,35 +530,41 @@ int gfs2_find_jhead(struct gfs2_jdesc *jd, struct gfs2_log_header_host *head, done = true; goto out; } + off = 0; + } + + if (!bio || (bio_chained && !off)) { + /* start new bio */ + } else { + sz = bio_add_page(bio, page, bsize, off); + if (sz == bsize) + goto block_added; + if (off) { + unsigned int blocks = + (PAGE_SIZE - off) >> bsize_shift; + + bio = gfs2_chain_bio(bio, blocks); + bio_chained = true; + goto add_block_to_new_bio; + } } if (bio) { - unsigned int off; - - off = (block << bsize_shift) & ~PAGE_MASK; - sz = bio_add_page(bio, page, bsize, off); - if (sz == bsize) { /* block added */ - if (off + bsize == PAGE_SIZE) { - page = NULL; - goto page_added; - } - continue; - } blocks_submitted = block + 1; submit_bio(bio); - bio = NULL; } - dblock = je->dblock + (block - je->lblock); bio = gfs2_log_alloc_bio(sdp, dblock, gfs2_end_log_read); bio->bi_opf = REQ_OP_READ; - sz = bio_add_page(bio, page, bsize, 0); - gfs2_assert_warn(sdp, sz == bsize); - if (bsize == PAGE_SIZE) + bio_chained = false; +add_block_to_new_bio: + sz = bio_add_page(bio, page, bsize, off); + BUG_ON(sz != bsize); +block_added: + off += bsize; + if (off == PAGE_SIZE) page = NULL; - -page_added: - if (blocks_submitted < blocks_read + readhead_blocks) { + if (blocks_submitted < blocks_read + readahead_blocks) { /* Keep at least one bio in flight */ continue; } From 2e9eeaa1175112ac494ba3c2ea4b71b1febeb6ab Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Fri, 13 Dec 2019 08:10:51 -0600 Subject: [PATCH 2/9] gfs2: eliminate ssize parameter from gfs2_struct2blk Every caller of function gfs2_struct2blk specified sizeof(u64). This patch eliminates the unnecessary parameter and replaces the size calculation with a new superblock variable that is computed to be the maximum number of block pointers we can fit inside a log descriptor, as is done for pointers per dinode and indirect block. Signed-off-by: Bob Peterson Reviewed-by: Andrew Price Signed-off-by: Andreas Gruenbacher --- fs/gfs2/glops.c | 2 +- fs/gfs2/incore.h | 1 + fs/gfs2/log.c | 12 ++++-------- fs/gfs2/log.h | 4 ++-- fs/gfs2/lops.c | 2 +- fs/gfs2/ops_fstype.c | 2 ++ fs/gfs2/trans.c | 3 +-- 7 files changed, 12 insertions(+), 14 deletions(-) diff --git a/fs/gfs2/glops.c b/fs/gfs2/glops.c index 4ede1f18de85..061d22e1ceb6 100644 --- a/fs/gfs2/glops.c +++ b/fs/gfs2/glops.c @@ -95,7 +95,7 @@ static void gfs2_ail_empty_gl(struct gfs2_glock *gl) /* A shortened, inline version of gfs2_trans_begin() * tr->alloced is not set since the transaction structure is * on the stack */ - tr.tr_reserved = 1 + gfs2_struct2blk(sdp, tr.tr_revokes, sizeof(u64)); + tr.tr_reserved = 1 + gfs2_struct2blk(sdp, tr.tr_revokes); tr.tr_ip = _RET_IP_; if (gfs2_log_reserve(sdp, tr.tr_reserved) < 0) return; diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h index 5f89c515f5bb..b5d9c11f4901 100644 --- a/fs/gfs2/incore.h +++ b/fs/gfs2/incore.h @@ -703,6 +703,7 @@ struct gfs2_sbd { u32 sd_fsb2bb_shift; u32 sd_diptrs; /* Number of pointers in a dinode */ u32 sd_inptrs; /* Number of pointers in a indirect block */ + u32 sd_ldptrs; /* Number of pointers in a log descriptor block */ u32 sd_jbsize; /* Size of a journaled data block */ u32 sd_hash_bsize; /* sizeof(exhash block) */ u32 sd_hash_bsize_shift; diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c index eb3f2e7b8085..66189ae62c1d 100644 --- a/fs/gfs2/log.c +++ b/fs/gfs2/log.c @@ -37,7 +37,6 @@ static void gfs2_log_shutdown(struct gfs2_sbd *sdp); * gfs2_struct2blk - compute stuff * @sdp: the filesystem * @nstruct: the number of structures - * @ssize: the size of the structures * * Compute the number of log descriptor blocks needed to hold a certain number * of structures of a certain size. @@ -45,18 +44,16 @@ static void gfs2_log_shutdown(struct gfs2_sbd *sdp); * Returns: the number of blocks needed (minimum is always 1) */ -unsigned int gfs2_struct2blk(struct gfs2_sbd *sdp, unsigned int nstruct, - unsigned int ssize) +unsigned int gfs2_struct2blk(struct gfs2_sbd *sdp, unsigned int nstruct) { unsigned int blks; unsigned int first, second; blks = 1; - first = (sdp->sd_sb.sb_bsize - sizeof(struct gfs2_log_descriptor)) / ssize; + first = sdp->sd_ldptrs; if (nstruct > first) { - second = (sdp->sd_sb.sb_bsize - - sizeof(struct gfs2_meta_header)) / ssize; + second = sdp->sd_inptrs; blks += DIV_ROUND_UP(nstruct - first, second); } @@ -473,8 +470,7 @@ static unsigned int calc_reserved(struct gfs2_sbd *sdp) } if (sdp->sd_log_commited_revoke > 0) - reserved += gfs2_struct2blk(sdp, sdp->sd_log_commited_revoke, - sizeof(u64)); + reserved += gfs2_struct2blk(sdp, sdp->sd_log_commited_revoke); /* One for the overall header */ if (reserved) reserved++; diff --git a/fs/gfs2/log.h b/fs/gfs2/log.h index 2ff163a8dce1..c0a65e5a126b 100644 --- a/fs/gfs2/log.h +++ b/fs/gfs2/log.h @@ -60,9 +60,9 @@ static inline void gfs2_ordered_add_inode(struct gfs2_inode *ip) spin_unlock(&sdp->sd_ordered_lock); } } + extern void gfs2_ordered_del_inode(struct gfs2_inode *ip); -extern unsigned int gfs2_struct2blk(struct gfs2_sbd *sdp, unsigned int nstruct, - unsigned int ssize); +extern unsigned int gfs2_struct2blk(struct gfs2_sbd *sdp, unsigned int nstruct); extern void gfs2_log_release(struct gfs2_sbd *sdp, unsigned int blks); extern int gfs2_log_reserve(struct gfs2_sbd *sdp, unsigned int blks); diff --git a/fs/gfs2/lops.c b/fs/gfs2/lops.c index aa0359b7e2ac..d9431724b788 100644 --- a/fs/gfs2/lops.c +++ b/fs/gfs2/lops.c @@ -866,7 +866,7 @@ static void revoke_lo_before_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr) if (!sdp->sd_log_num_revoke) return; - length = gfs2_struct2blk(sdp, sdp->sd_log_num_revoke, sizeof(u64)); + length = gfs2_struct2blk(sdp, sdp->sd_log_num_revoke); page = gfs2_get_log_desc(sdp, GFS2_LOG_DESC_REVOKE, length, sdp->sd_log_num_revoke); offset = sizeof(struct gfs2_log_descriptor); diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c index e8b7b0ce8404..b3e904bcc02c 100644 --- a/fs/gfs2/ops_fstype.c +++ b/fs/gfs2/ops_fstype.c @@ -298,6 +298,8 @@ static int gfs2_read_sb(struct gfs2_sbd *sdp, int silent) sizeof(struct gfs2_dinode)) / sizeof(u64); sdp->sd_inptrs = (sdp->sd_sb.sb_bsize - sizeof(struct gfs2_meta_header)) / sizeof(u64); + sdp->sd_ldptrs = (sdp->sd_sb.sb_bsize - + sizeof(struct gfs2_log_descriptor)) / sizeof(u64); sdp->sd_jbsize = sdp->sd_sb.sb_bsize - sizeof(struct gfs2_meta_header); sdp->sd_hash_bsize = sdp->sd_sb.sb_bsize / 2; sdp->sd_hash_bsize_shift = sdp->sd_sb.sb_bsize_shift - 1; diff --git a/fs/gfs2/trans.c b/fs/gfs2/trans.c index 9d4227330de4..4d01fe19c125 100644 --- a/fs/gfs2/trans.c +++ b/fs/gfs2/trans.c @@ -49,8 +49,7 @@ int gfs2_trans_begin(struct gfs2_sbd *sdp, unsigned int blocks, if (blocks) tr->tr_reserved += 6 + blocks; if (revokes) - tr->tr_reserved += gfs2_struct2blk(sdp, revokes, - sizeof(u64)); + tr->tr_reserved += gfs2_struct2blk(sdp, revokes); INIT_LIST_HEAD(&tr->tr_databuf); INIT_LIST_HEAD(&tr->tr_buf); From e556280d3691651c38022feaab6d6075adaa03bf Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Tue, 10 Dec 2019 12:05:55 -0600 Subject: [PATCH 3/9] gfs2: minor cleanup: remove unneeded variable ret in gfs2_jdata_writepage This patch simply removes variable ret, which is used to store the return code of its call to __gfs2_jdata_writepage, in favor of just returning the result directly. Signed-off-by: Bob Peterson --- fs/gfs2/aops.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/fs/gfs2/aops.c b/fs/gfs2/aops.c index 9c6df721321a..ba83b49ce18c 100644 --- a/fs/gfs2/aops.c +++ b/fs/gfs2/aops.c @@ -183,14 +183,12 @@ static int gfs2_jdata_writepage(struct page *page, struct writeback_control *wbc struct inode *inode = page->mapping->host; struct gfs2_inode *ip = GFS2_I(inode); struct gfs2_sbd *sdp = GFS2_SB(inode); - int ret; if (gfs2_assert_withdraw(sdp, gfs2_glock_is_held_excl(ip->i_gl))) goto out; if (PageChecked(page) || current->journal_info) goto out_ignore; - ret = __gfs2_jdata_writepage(page, wbc); - return ret; + return __gfs2_jdata_writepage(page, wbc); out_ignore: redirty_page_for_writepage(wbc, page); From 2b0fb353c029de78f9fb9462290af3f2432749b9 Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Wed, 15 Jan 2020 06:26:00 +0100 Subject: [PATCH 4/9] gfs2: Avoid access time thrashing in gfs2_inode_lookup In gfs2_inode_lookup, we initialize inode->i_atime to the lowest possibly value after gfs2_inode_refresh may already have been called. This should be the other way around, but we didn't notice because usually the inode type is known from the directory entry and so gfs2_inode_lookup won't call gfs2_inode_refresh. In addition, only initialize ip->i_no_formal_ino from no_formal_ino when actually needed. Signed-off-by: Andreas Gruenbacher --- fs/gfs2/inode.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c index dafef10b91f1..2716d56ed0a0 100644 --- a/fs/gfs2/inode.c +++ b/fs/gfs2/inode.c @@ -136,7 +136,6 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type, if (inode->i_state & I_NEW) { struct gfs2_sbd *sdp = GFS2_SB(inode); - ip->i_no_formal_ino = no_formal_ino; error = gfs2_glock_get(sdp, no_addr, &gfs2_inode_glops, CREATE, &ip->i_gl); if (unlikely(error)) @@ -175,21 +174,22 @@ struct inode *gfs2_inode_lookup(struct super_block *sb, unsigned int type, gfs2_glock_put(io_gl); io_gl = NULL; + /* Lowest possible timestamp; will be overwritten in gfs2_dinode_in. */ + inode->i_atime.tv_sec = 1LL << (8 * sizeof(inode->i_atime.tv_sec) - 1); + inode->i_atime.tv_nsec = 0; + if (type == DT_UNKNOWN) { /* Inode glock must be locked already */ error = gfs2_inode_refresh(GFS2_I(inode)); if (error) goto fail_refresh; } else { + ip->i_no_formal_ino = no_formal_ino; inode->i_mode = DT2IF(type); } gfs2_set_iop(inode); - /* Lowest possible timestamp; will be overwritten in gfs2_dinode_in. */ - inode->i_atime.tv_sec = 1LL << (8 * sizeof(inode->i_atime.tv_sec) - 1); - inode->i_atime.tv_nsec = 0; - unlock_new_inode(inode); } From 5d43975859bf4fa5a27d568f37091f46a73a9725 Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Thu, 9 Jan 2020 13:54:36 +0100 Subject: [PATCH 5/9] gfs2: Fix incorrect variable name Rename sd_log_commited_revoke to sd_log_committed_revoke. Signed-off-by: Andreas Gruenbacher --- fs/gfs2/incore.h | 2 +- fs/gfs2/log.c | 10 +++++----- 2 files changed, 6 insertions(+), 6 deletions(-) diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h index b5d9c11f4901..e9254e2fbc20 100644 --- a/fs/gfs2/incore.h +++ b/fs/gfs2/incore.h @@ -804,7 +804,7 @@ struct gfs2_sbd { struct gfs2_trans *sd_log_tr; unsigned int sd_log_blks_reserved; - int sd_log_commited_revoke; + int sd_log_committed_revoke; atomic_t sd_log_pinned; unsigned int sd_log_num_revoke; diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c index 66189ae62c1d..ca8fcdb35822 100644 --- a/fs/gfs2/log.c +++ b/fs/gfs2/log.c @@ -469,8 +469,8 @@ static unsigned int calc_reserved(struct gfs2_sbd *sdp) reserved += DIV_ROUND_UP(dbuf, databuf_limit(sdp)); } - if (sdp->sd_log_commited_revoke > 0) - reserved += gfs2_struct2blk(sdp, sdp->sd_log_commited_revoke); + if (sdp->sd_log_committed_revoke > 0) + reserved += gfs2_struct2blk(sdp, sdp->sd_log_committed_revoke); /* One for the overall header */ if (reserved) reserved++; @@ -825,7 +825,7 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags) if (unlikely(state == SFS_FROZEN)) gfs2_assert_withdraw(sdp, !sdp->sd_log_num_revoke); gfs2_assert_withdraw(sdp, - sdp->sd_log_num_revoke == sdp->sd_log_commited_revoke); + sdp->sd_log_num_revoke == sdp->sd_log_committed_revoke); gfs2_ordered_write(sdp); lops_before_commit(sdp, tr); @@ -844,7 +844,7 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags) gfs2_log_lock(sdp); sdp->sd_log_head = sdp->sd_log_flush_head; sdp->sd_log_blks_reserved = 0; - sdp->sd_log_commited_revoke = 0; + sdp->sd_log_committed_revoke = 0; spin_lock(&sdp->sd_ail_lock); if (tr && !list_empty(&tr->tr_ail1_list)) { @@ -916,7 +916,7 @@ static void log_refund(struct gfs2_sbd *sdp, struct gfs2_trans *tr) set_bit(TR_ATTACHED, &tr->tr_flags); } - sdp->sd_log_commited_revoke += tr->tr_num_revoke; + sdp->sd_log_committed_revoke += tr->tr_num_revoke; reserved = calc_reserved(sdp); maxres = sdp->sd_log_blks_reserved + tr->tr_reserved; gfs2_assert_withdraw(sdp, maxres >= reserved); From f7be987b825434e2cac5027d47501ae1102881b4 Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Thu, 16 Jan 2020 00:25:32 +0100 Subject: [PATCH 6/9] gfs2: Remove GFS2_MIN_LVB_SIZE define The dlm lockspace is set up to have lock value blocks of GDLM_LVB_SIZE bytes, and dlm is the only lock manager we support, so there is no point in claiming that the lock value block could have any other size. Signed-off-by: Andreas Gruenbacher --- fs/gfs2/glock.c | 2 +- fs/gfs2/incore.h | 2 -- 2 files changed, 1 insertion(+), 3 deletions(-) diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c index b7123de7c180..d0eceaff3cea 100644 --- a/fs/gfs2/glock.c +++ b/fs/gfs2/glock.c @@ -826,7 +826,7 @@ int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number, memset(&gl->gl_lksb, 0, sizeof(struct dlm_lksb)); if (glops->go_flags & GLOF_LVB) { - gl->gl_lksb.sb_lvbptr = kzalloc(GFS2_MIN_LVB_SIZE, GFP_NOFS); + gl->gl_lksb.sb_lvbptr = kzalloc(GDLM_LVB_SIZE, GFP_NOFS); if (!gl->gl_lksb.sb_lvbptr) { kmem_cache_free(cachep, gl); return -ENOMEM; diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h index e9254e2fbc20..68d67d38a09d 100644 --- a/fs/gfs2/incore.h +++ b/fs/gfs2/incore.h @@ -387,8 +387,6 @@ struct gfs2_glock { struct rhash_head gl_node; }; -#define GFS2_MIN_LVB_SIZE 32 /* Min size of LVB that gfs2 supports */ - enum { GIF_INVALID = 0, GIF_QD_LOCKED = 1, From b3ca4e447dba6c30c7867a1353712e3507dc5110 Mon Sep 17 00:00:00 2001 From: Alex Shi Date: Tue, 21 Jan 2020 16:49:12 +0800 Subject: [PATCH 7/9] fs/gfs2: remove unused IS_DINODE and IS_LEAF macros Since commit 1579343a73e3 ("GFS2: Remove dirent_first() function"), these macros aren't used any more, so remove them. Signed-off-by: Alex Shi Signed-off-by: Andreas Gruenbacher --- fs/gfs2/dir.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/fs/gfs2/dir.c b/fs/gfs2/dir.c index eb9c0578978f..c8b62577e2f2 100644 --- a/fs/gfs2/dir.c +++ b/fs/gfs2/dir.c @@ -73,9 +73,6 @@ #include "bmap.h" #include "util.h" -#define IS_LEAF 1 /* Hashed (leaf) directory */ -#define IS_DINODE 2 /* Linear (stuffed dinode block) directory */ - #define MAX_RA_BLOCKS 32 /* max read-ahead blocks */ #define gfs2_disk_hash2offset(h) (((u64)(h)) >> 1) From c04f2e0dd5309607dbc425f02b5ac076b395f19d Mon Sep 17 00:00:00 2001 From: Alex Shi Date: Tue, 21 Jan 2020 16:49:17 +0800 Subject: [PATCH 8/9] gfs2: remove unused LBIT macros Since commit 223b2b889f37 ("GFS2: Fix alignment issue and tidy gfs2_bitfit"), these 3 macros aren't used anymore, so remove them. Signed-off-by: Alex Shi Signed-off-by: Andreas Gruenbacher --- fs/gfs2/rgrp.c | 10 ---------- 1 file changed, 10 deletions(-) diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c index 2466bb44a23c..e7bf91ec231c 100644 --- a/fs/gfs2/rgrp.c +++ b/fs/gfs2/rgrp.c @@ -36,16 +36,6 @@ #define BFITNOENT ((u32)~0) #define NO_BLOCK ((u64)~0) -#if BITS_PER_LONG == 32 -#define LBITMASK (0x55555555UL) -#define LBITSKIP55 (0x55555555UL) -#define LBITSKIP00 (0x00000000UL) -#else -#define LBITMASK (0x5555555555555555UL) -#define LBITSKIP55 (0x5555555555555555UL) -#define LBITSKIP00 (0x0000000000000000UL) -#endif - /* * These routines are used by the resource group routines (rgrp.c) * to keep track of block allocation. Each block is represented by two From a31b4ec539e966515f1f97f4000d0e2a399930ce Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Mon, 20 Jan 2020 15:49:28 +0100 Subject: [PATCH 9/9] Revert "gfs2: eliminate tr_num_revoke_rm" This reverts commit e955537e3262de8e56f070b13817f525f472fa00. Before patch e955537e32, tr_num_revoke tracked the number of revokes added to the transaction, and tr_num_revoke_rm tracked how many revokes were removed. But since revokes are queued off the sdp (superblock) pointer, some transactions could remove more revokes than they added. (e.g. revokes added by a different process). Commit e955537e32 eliminated transaction variable tr_num_revoke_rm, but in order to do so, it changed the accounting to always use tr_num_revoke for its math. Since you can remove more revokes than you add, tr_num_revoke could now become a negative value. This negative value broke the assert in function gfs2_trans_end: if (gfs2_assert_withdraw(sdp, (nbuf <=3D tr->tr_blocks) && (tr->tr_num_revoke <=3D tr->tr_revokes))) One way to fix this is to simply remove the tr_num_revoke clause from the assert and allow the value to become negative. Andreas didn't like that idea, so instead, we decided to revert e955537e32. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/incore.h | 1 + fs/gfs2/log.c | 3 ++- fs/gfs2/trans.c | 6 +++--- 3 files changed, 6 insertions(+), 4 deletions(-) diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h index 68d67d38a09d..9fd88ed18807 100644 --- a/fs/gfs2/incore.h +++ b/fs/gfs2/incore.h @@ -503,6 +503,7 @@ struct gfs2_trans { unsigned int tr_num_buf_rm; unsigned int tr_num_databuf_rm; unsigned int tr_num_revoke; + unsigned int tr_num_revoke_rm; struct list_head tr_list; struct list_head tr_databuf; diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c index ca8fcdb35822..00a2e721a374 100644 --- a/fs/gfs2/log.c +++ b/fs/gfs2/log.c @@ -895,6 +895,7 @@ static void gfs2_merge_trans(struct gfs2_trans *old, struct gfs2_trans *new) old->tr_num_buf_rm += new->tr_num_buf_rm; old->tr_num_databuf_rm += new->tr_num_databuf_rm; old->tr_num_revoke += new->tr_num_revoke; + old->tr_num_revoke_rm += new->tr_num_revoke_rm; list_splice_tail_init(&new->tr_databuf, &old->tr_databuf); list_splice_tail_init(&new->tr_buf, &old->tr_buf); @@ -916,7 +917,7 @@ static void log_refund(struct gfs2_sbd *sdp, struct gfs2_trans *tr) set_bit(TR_ATTACHED, &tr->tr_flags); } - sdp->sd_log_committed_revoke += tr->tr_num_revoke; + sdp->sd_log_committed_revoke += tr->tr_num_revoke - tr->tr_num_revoke_rm; reserved = calc_reserved(sdp); maxres = sdp->sd_log_blks_reserved + tr->tr_reserved; gfs2_assert_withdraw(sdp, maxres >= reserved); diff --git a/fs/gfs2/trans.c b/fs/gfs2/trans.c index 4d01fe19c125..a685637a5b55 100644 --- a/fs/gfs2/trans.c +++ b/fs/gfs2/trans.c @@ -76,10 +76,10 @@ static void gfs2_print_trans(struct gfs2_sbd *sdp, const struct gfs2_trans *tr) fs_warn(sdp, "blocks=%u revokes=%u reserved=%u touched=%u\n", tr->tr_blocks, tr->tr_revokes, tr->tr_reserved, test_bit(TR_TOUCHED, &tr->tr_flags)); - fs_warn(sdp, "Buf %u/%u Databuf %u/%u Revoke %u\n", + fs_warn(sdp, "Buf %u/%u Databuf %u/%u Revoke %u/%u\n", tr->tr_num_buf_new, tr->tr_num_buf_rm, tr->tr_num_databuf_new, tr->tr_num_databuf_rm, - tr->tr_num_revoke); + tr->tr_num_revoke, tr->tr_num_revoke_rm); } void gfs2_trans_end(struct gfs2_sbd *sdp) @@ -264,7 +264,7 @@ void gfs2_trans_remove_revoke(struct gfs2_sbd *sdp, u64 blkno, unsigned int len) if (bd->bd_gl) gfs2_glock_remove_revoke(bd->bd_gl); kmem_cache_free(gfs2_bufdata_cachep, bd); - tr->tr_num_revoke--; + tr->tr_num_revoke_rm++; if (--n == 0) break; }