Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751956Ab3FNPjn (ORCPT ); Fri, 14 Jun 2013 11:39:43 -0400 Received: from m59-178.qiye.163.com ([123.58.178.59]:37188 "EHLO m59-178.qiye.163.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751545Ab3FNPjm (ORCPT ); Fri, 14 Jun 2013 11:39:42 -0400 From: Li Wang To: ceph-devel@vger.kernel.org Cc: Sage Weil , linux-kernel@vger.kernel.org, Li Wang , Yunchuan Wen Subject: [PATCH 2/2] Punch hole support against 3.10-rc5 Date: Fri, 14 Jun 2013 23:39:00 +0800 Message-Id: <1371224340-4926-1-git-send-email-liwang@ubuntukylin.com> X-Mailer: git-send-email 1.7.9.5 In-Reply-To: <1371224186-4809-1-git-send-email-liwang@ubuntukylin.com> References: <1371224186-4809-1-git-send-email-liwang@ubuntukylin.com> X-HM-Spam-Status: e1koWUFPN1dZCBgUCR5ZQU5VQ0hLS0tLT0tLTk9ISUpIV1kJDhceCFlBWSgrPSQoND0vPToyNyQy NSQzPjo*PilBS1VLQDYjJCI#KCQyNSQzPjo*PilBS1VLQCsvKSQ1NCQyNSQzPjo*PilBSVVLQDg0 LjUvKSIkODVBS1VLQCk#PDI0NSQ6KDI6QUtVS0ArKTQtMjU4PiQzLjU6NUFLVUtAPyI1OjYyOCQy KyQ1NCQyNSQzPjo*PilBS1VLQDYuNy8yJCk4Ky8kPzI9PT4pPjUvJDI1JDM#Oj8#KUFJVUtAMisk LzQ*OiIkODUvJEskSktLQUtVS0AyKyRKJDM0LikkODUvJEskSktLQUtVS0AyKyRKJDYyNS4vPiQ4 NS8kSyRKS0FLVUtAMiskSEskNjI1Li8#JDg1LyRLJE5LQUtVS0AyKyROJDYyNS4vPiQ4NS8kSyRK S0FLVUtAKC45JD5BSlVOTkA9NSQoLjkkPjUsNCk*KCQzNzEkSktLSUtKQUtVSUNZBg++ X-HM-Sender-Digest: e1kSHx4VD1lBWUc6MQg6Cjo4LDo4EDorKjhIOj4qOkMwCjFVSlVKSExKSUlPSENLSkhLVTMWGhIX VRcSDBoVHDsOGQ4VDw4QAhcSFVUYFBZFWVdZDB4ZWUEdGhcIHgY+ Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 9768 Lines: 321 This patch implements punch hole (fallocate) support against Linux kernel 3.10-rc5. Signed-off-by: Li Wang Signed-off-by: Yunchuan Wen --- fs/ceph/file.c | 245 +++++++++++++++++++++++++++++++++++++++++++++++++ net/ceph/osd_client.c | 8 +- 2 files changed, 251 insertions(+), 2 deletions(-) diff --git a/fs/ceph/file.c b/fs/ceph/file.c index 656e169..e092b69 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -8,6 +8,7 @@ #include #include #include +#include #include "super.h" #include "mds_client.h" @@ -882,6 +883,249 @@ out: return offset; } +static inline void ceph_zero_partial_page(struct inode *inode, pgoff_t index, unsigned start, unsigned size) +{ + struct page *page; + + page = find_lock_page(inode->i_mapping, index); + if (page) { + zero_user(page, start, size); + unlock_page(page); + page_cache_release(page); + } +} + +static void ceph_truncate_and_zero_page_cache(struct inode *inode, loff_t offset, loff_t length) +{ + loff_t first_page; + loff_t last_page; + loff_t zero_len; + + first_page =((offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) << PAGE_CACHE_SHIFT; + last_page = ((offset + length) >> PAGE_CACHE_SHIFT) << PAGE_CACHE_SHIFT; + if (last_page > first_page) { + truncate_pagecache_range(inode, first_page, last_page - 1); + } + if (first_page > last_page) { + ceph_zero_partial_page(inode, offset >> PAGE_CACHE_SHIFT, offset & (PAGE_CACHE_SIZE - 1), length); + return; + } + /* + * zero out the partial page that contains + * the start of the hole + */ + zero_len = first_page - offset; + if (zero_len > 0) { + ceph_zero_partial_page(inode, offset >> PAGE_CACHE_SHIFT, offset & (PAGE_CACHE_SIZE -1), zero_len); + } + /* + * zero out the partial page that contains + * the end of the hole + */ + zero_len = offset + length - last_page; + if (zero_len > 0) { + ceph_zero_partial_page(inode, (offset + length) >> PAGE_CACHE_SHIFT, 0, zero_len); + } + /* + * If i_size is contained in the last page, we need to + * zero the partial page after i_size + */ + if (inode->i_size >> PAGE_CACHE_SHIFT == (offset + length) >> PAGE_CACHE_SHIFT && inode->i_size % PAGE_CACHE_SIZE != 0) { + zero_len = PAGE_CACHE_SIZE - + (inode->i_size & (PAGE_CACHE_SIZE - 1)); + if (zero_len > 0) { + ceph_zero_partial_page(inode, inode->i_size >> PAGE_CACHE_SHIFT, inode->i_size & (PAGE_CACHE_SIZE -1), zero_len); + } + } +} + +static int ceph_delete_object_range(struct inode *inode, loff_t lstart, loff_t lend) +{ + struct ceph_inode_info *ci = ceph_inode(inode); + struct ceph_fs_client *fsc = ceph_inode_to_client(inode); + struct ceph_osd_request *req; + u64 length = ceph_file_layout_object_size(ci->i_layout); + loff_t offset; + int ret = 0; + + if (lstart > lend || length <= 0) + goto out; + for (offset = lstart; offset <= lend; offset += length) { + req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout, + ceph_vino(inode), offset, &length, + 1, CEPH_OSD_OP_DELETE, CEPH_OSD_FLAG_ONDISK, + NULL, + ci->i_truncate_seq, ci->i_truncate_size, + false); + if (IS_ERR(req)) { + ret = PTR_ERR(req); + goto out; + } + + ret = ceph_osdc_start_request(&fsc->client->osdc, req, false); + if (!ret) { + ret = ceph_osdc_wait_request(&fsc->client->osdc, req); + } + ceph_osdc_put_request(req); + /* object deleted */ + if (ret == -ENOENT) + ret = 0; + } + + out: + return ret; +} + +static int ceph_zero_partial_object(struct file *file, loff_t offset, loff_t length) +{ + struct ceph_file_info *fi = file->private_data; + struct inode *inode = file->f_dentry->d_inode; + struct ceph_inode_info *ci = ceph_inode(inode); + struct ceph_fs_client *fsc = ceph_inode_to_client(inode); + struct ceph_osd_request *req; + int want, got = 0, ret = 0; + + if (length <= 0) + goto out; + + + if (fi->fmode & CEPH_FILE_MODE_LAZY) + want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO; + else + want = CEPH_CAP_FILE_BUFFER; + + ret = ceph_get_caps(ci, CEPH_CAP_FILE_WR, want, &got, offset+length); + if (ret < 0) + goto out; + if (!(got & (CEPH_CAP_FILE_BUFFER|CEPH_CAP_FILE_LAZYIO))) { + ceph_put_cap_refs(ci, got); + ret = -EAGAIN; + goto out; + } + req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout, + ceph_vino(inode), offset, &length, 1, + CEPH_OSD_OP_ZERO, CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, + NULL, + ci->i_truncate_seq, ci->i_truncate_size, + false); + if (IS_ERR(req)) { + ret = PTR_ERR(req); + goto out; + } + + ret = ceph_osdc_start_request(&fsc->client->osdc, req, false); + if (!ret) { + ret = ceph_osdc_wait_request(&fsc->client->osdc, req); + } + ceph_osdc_put_request(req); + ceph_put_cap_refs(ci, got); + + out: + return ret; +} + +static int ceph_delete_and_zero_objects(struct file *file, loff_t offset, loff_t length) +{ + unsigned long first_object; + unsigned long last_object; + struct inode *inode = file->f_dentry->d_inode; + struct ceph_inode_info *ci = ceph_inode(inode); + __s32 object_size; + __u32 object_shift; + loff_t zero_len; + int ret = 0; + + if (!(object_size = ceph_file_layout_object_size(ci->i_layout))) + goto out; + if (object_size == 1) { + object_shift = 0; + } else { + for (object_shift = 0; ;object_shift++) { + if (2 << object_shift == object_size) + break; + } + object_shift++; + } + + first_object =((offset + object_size - 1) >> object_shift) << object_shift; + last_object = ((offset + length) >> object_shift) << object_shift; + if (last_object > first_object) { + ret = ceph_delete_object_range(inode, first_object, last_object - 1); + if (ret) + goto out; + } + if (first_object > last_object) { + ret = ceph_zero_partial_object(file, offset, length); + goto out; + } + /* + * zero out the partial object that contains + * the start of the hole + */ + zero_len = first_object - offset; + if (zero_len > 0) { + ret = ceph_zero_partial_object(file, offset, zero_len); + if (ret) + goto out; + } + /* + * zero out the partial object that contains + * the end of the hole + */ + zero_len = offset + length - last_object; + if (zero_len > 0) { + ret = ceph_zero_partial_object(file, last_object, zero_len); + } + + out: + return ret; +} + +static int ceph_punch_hole(struct file *file, loff_t offset, loff_t length) +{ + struct inode *inode = file->f_dentry->d_inode; + int ret = 0; + + if (!S_ISREG(inode->i_mode)) { + return -EOPNOTSUPP; + } + if (IS_SWAPFILE(inode)) { + return -ETXTBSY; + } + mutex_lock(&inode->i_mutex); + + /* No need to punch hole beyond i_size */ + if (offset >= inode->i_size) + goto out_unlock; + + /* + * If the hole extends beyond i_size, set the hole + * to end after the page that contains i_size + */ + if (offset + length > inode->i_size) { + length = inode->i_size + + PAGE_CACHE_SIZE - (inode->i_size & (PAGE_CACHE_SIZE - 1)) - + offset; + } + + ceph_truncate_and_zero_page_cache(inode, offset, length); + ret = ceph_delete_and_zero_objects(file, offset, length); + + out_unlock: + mutex_unlock(&inode->i_mutex); + return ret; +} + +static long ceph_fallocate(struct file *file, int mode, loff_t offset, loff_t length) +{ + /* FALLOC_FL_PUNCH_HOLE must be used with FALLOC_FL_KEEP_SIZE */ + if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) + return -EOPNOTSUPP; + if (mode & FALLOC_FL_PUNCH_HOLE) + return ceph_punch_hole(file, offset, length); + return -EOPNOTSUPP; +} + const struct file_operations ceph_file_fops = { .open = ceph_open, .release = ceph_release, @@ -898,5 +1142,6 @@ const struct file_operations ceph_file_fops = { .splice_write = generic_file_splice_write, .unlocked_ioctl = ceph_ioctl, .compat_ioctl = ceph_ioctl, + .fallocate = ceph_fallocate, }; diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c index 3a246a6..a6d9671 100644 --- a/net/ceph/osd_client.c +++ b/net/ceph/osd_client.c @@ -503,7 +503,8 @@ void osd_req_op_extent_init(struct ceph_osd_request *osd_req, struct ceph_osd_req_op *op = _osd_req_op_init(osd_req, which, opcode); size_t payload_len = 0; - BUG_ON(opcode != CEPH_OSD_OP_READ && opcode != CEPH_OSD_OP_WRITE); + BUG_ON(opcode != CEPH_OSD_OP_READ && opcode != CEPH_OSD_OP_WRITE && + opcode != CEPH_OSD_OP_DELETE && opcode != CEPH_OSD_OP_ZERO); op->extent.offset = offset; op->extent.length = length; @@ -631,6 +632,8 @@ static u64 osd_req_encode_op(struct ceph_osd_request *req, break; case CEPH_OSD_OP_READ: case CEPH_OSD_OP_WRITE: + case CEPH_OSD_OP_DELETE: + case CEPH_OSD_OP_ZERO: if (src->op == CEPH_OSD_OP_WRITE) request_data_len = src->extent.length; dst->extent.offset = cpu_to_le64(src->extent.offset); @@ -715,7 +718,8 @@ struct ceph_osd_request *ceph_osdc_new_request(struct ceph_osd_client *osdc, u64 object_base; int r; - BUG_ON(opcode != CEPH_OSD_OP_READ && opcode != CEPH_OSD_OP_WRITE); + BUG_ON(opcode != CEPH_OSD_OP_READ && opcode != CEPH_OSD_OP_WRITE && + opcode != CEPH_OSD_OP_DELETE && opcode != CEPH_OSD_OP_ZERO); req = ceph_osdc_alloc_request(osdc, snapc, num_ops, use_mempool, GFP_NOFS); -- 1.7.9.5 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/