From: Xiubo Li <xiubli@xxxxxxxxxx> When fscrypt is enabled we will align the truncate size up to the CEPH_FSCRYPT_BLOCK_SIZE always, so if we truncate the size in the same block more than once, the latter ones will be skipped being invalidated from the page caches. This will force invalidating the page caches by using the smaller size than the real file size. At the same time add more debug log and fix the debug log for truncate code. URL: https://tracker.ceph.com/issues/58834 Signed-off-by: Xiubo Li <xiubli@xxxxxxxxxx> --- This is for fscrypt feature and based on the testing branch. fs/ceph/caps.c | 4 ++-- fs/ceph/inode.c | 35 ++++++++++++++++++++++++----------- 2 files changed, 26 insertions(+), 13 deletions(-) diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 4a17ee6942ac..07b29b72ac39 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -3928,8 +3928,8 @@ static bool handle_cap_trunc(struct inode *inode, if (IS_ENCRYPTED(inode) && size) size = extra_info->fscrypt_file_size; - dout("handle_cap_trunc inode %p mds%d seq %d to %lld seq %d\n", - inode, mds, seq, truncate_size, truncate_seq); + dout("%s inode %p mds%d seq %d to %lld truncate seq %d\n", + __func__, inode, mds, seq, truncate_size, truncate_seq); queue_trunc = ceph_fill_file_size(inode, issued, truncate_seq, truncate_size, size, 0); diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index b9a3ff7ecbcb..dfa69bc346bf 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -765,7 +765,7 @@ int ceph_fill_file_size(struct inode *inode, int issued, ceph_fscache_update(inode); ci->i_reported_size = size; if (truncate_seq != ci->i_truncate_seq) { - dout("truncate_seq %u -> %u\n", + dout("%s truncate_seq %u -> %u\n", __func__, ci->i_truncate_seq, truncate_seq); ci->i_truncate_seq = truncate_seq; @@ -799,15 +799,26 @@ int ceph_fill_file_size(struct inode *inode, int issued, } } } - if (ceph_seq_cmp(truncate_seq, ci->i_truncate_seq) >= 0 && - ci->i_truncate_size != truncate_size) { - dout("truncate_size %lld -> %llu\n", ci->i_truncate_size, - truncate_size); + + /* + * It's possible that the new sizes of the two consecutive + * size truncations will be in the same fscrypt last block, + * and we need to truncate the corresponding page caches + * anyway. + */ + if (ceph_seq_cmp(truncate_seq, ci->i_truncate_seq) >= 0) { + dout("%s truncate_size %lld -> %llu, encrypted %d\n", __func__, + ci->i_truncate_size, truncate_size, !!IS_ENCRYPTED(inode)); + ci->i_truncate_size = truncate_size; - if (IS_ENCRYPTED(inode)) + + if (IS_ENCRYPTED(inode)) { + dout("%s truncate_pagecache_size %lld -> %llu\n", + __func__, ci->i_truncate_pagecache_size, size); ci->i_truncate_pagecache_size = size; - else + } else { ci->i_truncate_pagecache_size = truncate_size; + } } return queue_trunc; } @@ -2162,7 +2173,7 @@ void __ceph_do_pending_vmtruncate(struct inode *inode) retry: spin_lock(&ci->i_ceph_lock); if (ci->i_truncate_pending == 0) { - dout("__do_pending_vmtruncate %p none pending\n", inode); + dout("%s %p none pending\n", __func__, inode); spin_unlock(&ci->i_ceph_lock); mutex_unlock(&ci->i_truncate_mutex); return; @@ -2174,8 +2185,7 @@ void __ceph_do_pending_vmtruncate(struct inode *inode) */ if (ci->i_wrbuffer_ref_head < ci->i_wrbuffer_ref) { spin_unlock(&ci->i_ceph_lock); - dout("__do_pending_vmtruncate %p flushing snaps first\n", - inode); + dout("%s %p flushing snaps first\n", __func__, inode); filemap_write_and_wait_range(&inode->i_data, 0, inode->i_sb->s_maxbytes); goto retry; @@ -2186,7 +2196,7 @@ void __ceph_do_pending_vmtruncate(struct inode *inode) to = ci->i_truncate_pagecache_size; wrbuffer_refs = ci->i_wrbuffer_ref; - dout("__do_pending_vmtruncate %p (%d) to %lld\n", inode, + dout("%s %p (%d) to %lld\n", __func__, inode, ci->i_truncate_pending, to); spin_unlock(&ci->i_ceph_lock); @@ -2373,6 +2383,9 @@ static int fill_fscrypt_truncate(struct inode *inode, header.data_len = cpu_to_le32(8 + 8 + 4 + CEPH_FSCRYPT_BLOCK_SIZE); header.file_offset = cpu_to_le64(orig_pos); + dout("%s encrypt block boff/bsize %d/%lu\n", __func__, + boff, CEPH_FSCRYPT_BLOCK_SIZE); + /* truncate and zero out the extra contents for the last block */ memset(iov.iov_base + boff, 0, PAGE_SIZE - boff); -- 2.31.1
