From 41c4d25f78c01ede13efee1f2e979f3f35dd26f6 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 12 Jun 2012 16:20:28 +0200 Subject: mm: Update file times from fault path only if .page_mkwrite is not set Filesystems wanting to properly support freezing need to have control when file_update_time() is called. After pushing file_update_time() to all relevant .page_mkwrite implementations we can just stop calling file_update_time() when filesystem implements .page_mkwrite. Tested-by: Kamal Mostafa Tested-by: Peter M. Petrakis Tested-by: Dann Frazier Tested-by: Massimo Morana Signed-off-by: Jan Kara Signed-off-by: Al Viro --- mm/memory.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'mm') diff --git a/mm/memory.c b/mm/memory.c index 2466d1250231..7c7fa7b4b6b6 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -2638,6 +2638,9 @@ reuse: if (!page_mkwrite) { wait_on_page_locked(dirty_page); set_page_dirty_balance(dirty_page, page_mkwrite); + /* file_update_time outside page_lock */ + if (vma->vm_file) + file_update_time(vma->vm_file); } put_page(dirty_page); if (page_mkwrite) { @@ -2655,10 +2658,6 @@ reuse: } } - /* file_update_time outside page_lock */ - if (vma->vm_file) - file_update_time(vma->vm_file); - return ret; } @@ -3327,12 +3326,13 @@ static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma, if (dirty_page) { struct address_space *mapping = page->mapping; + int dirtied = 0; if (set_page_dirty(dirty_page)) - page_mkwrite = 1; + dirtied = 1; unlock_page(dirty_page); put_page(dirty_page); - if (page_mkwrite && mapping) { + if ((dirtied || page_mkwrite) && mapping) { /* * Some device drivers do not set page.mapping but still * dirty their pages @@ -3341,7 +3341,7 @@ static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma, } /* file_update_time outside page_lock */ - if (vma->vm_file) + if (vma->vm_file && !page_mkwrite) file_update_time(vma->vm_file); } else { unlock_page(vmf.page); -- cgit v1.2.3 From 4fcf1c6205fcfc7a226a144ae4d83b7f5415cab8 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 12 Jun 2012 16:20:29 +0200 Subject: mm: Make default vm_ops provide ->page_mkwrite handler Make default vm_ops provide ->page_mkwrite handler. Currently it only updates file's modification times and gets locked page but later it will also handle filesystem freezing. BugLink: https://bugs.launchpad.net/bugs/897421 Tested-by: Kamal Mostafa Tested-by: Peter M. Petrakis Tested-by: Dann Frazier Tested-by: Massimo Morana Signed-off-by: Jan Kara Signed-off-by: Al Viro --- mm/filemap.c | 19 +++++++++++++++++++ mm/filemap_xip.c | 1 + 2 files changed, 20 insertions(+) (limited to 'mm') diff --git a/mm/filemap.c b/mm/filemap.c index a4a5260b0279..51efee65c2cc 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -1712,8 +1712,27 @@ page_not_uptodate: } EXPORT_SYMBOL(filemap_fault); +int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + struct page *page = vmf->page; + struct inode *inode = vma->vm_file->f_path.dentry->d_inode; + int ret = VM_FAULT_LOCKED; + + file_update_time(vma->vm_file); + lock_page(page); + if (page->mapping != inode->i_mapping) { + unlock_page(page); + ret = VM_FAULT_NOPAGE; + goto out; + } +out: + return ret; +} +EXPORT_SYMBOL(filemap_page_mkwrite); + const struct vm_operations_struct generic_file_vm_ops = { .fault = filemap_fault, + .page_mkwrite = filemap_page_mkwrite, }; /* This is used for a general mmap of a disk file */ diff --git a/mm/filemap_xip.c b/mm/filemap_xip.c index 213ca1f53409..80b34ef82dfe 100644 --- a/mm/filemap_xip.c +++ b/mm/filemap_xip.c @@ -304,6 +304,7 @@ out: static const struct vm_operations_struct xip_file_vm_ops = { .fault = xip_file_fault, + .page_mkwrite = filemap_page_mkwrite, }; int xip_file_mmap(struct file * file, struct vm_area_struct * vma) -- cgit v1.2.3 From 14da9200140f8d722ad1767dfabadebd8b34f2ad Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 12 Jun 2012 16:20:37 +0200 Subject: fs: Protect write paths by sb_start_write - sb_end_write There are several entry points which dirty pages in a filesystem. mmap (handled by block_page_mkwrite()), buffered write (handled by __generic_file_aio_write()), splice write (generic_file_splice_write), truncate, and fallocate (these can dirty last partial page - handled inside each filesystem separately). Protect these places with sb_start_write() and sb_end_write(). ->page_mkwrite() calls are particularly complex since they are called with mmap_sem held and thus we cannot use standard sb_start_write() due to lock ordering constraints. We solve the problem by using a special freeze protection sb_start_pagefault() which ranks below mmap_sem. BugLink: https://bugs.launchpad.net/bugs/897421 Tested-by: Kamal Mostafa Tested-by: Peter M. Petrakis Tested-by: Dann Frazier Tested-by: Massimo Morana Signed-off-by: Jan Kara Signed-off-by: Al Viro --- mm/filemap.c | 12 ++++++++++-- mm/filemap_xip.c | 5 +++-- 2 files changed, 13 insertions(+), 4 deletions(-) (limited to 'mm') diff --git a/mm/filemap.c b/mm/filemap.c index 51efee65c2cc..fa5ca304148e 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -1718,6 +1718,7 @@ int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) struct inode *inode = vma->vm_file->f_path.dentry->d_inode; int ret = VM_FAULT_LOCKED; + sb_start_pagefault(inode->i_sb); file_update_time(vma->vm_file); lock_page(page); if (page->mapping != inode->i_mapping) { @@ -1725,7 +1726,14 @@ int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) ret = VM_FAULT_NOPAGE; goto out; } + /* + * We mark the page dirty already here so that when freeze is in + * progress, we are guaranteed that writeback during freezing will + * see the dirty page and writeprotect it again. + */ + set_page_dirty(page); out: + sb_end_pagefault(inode->i_sb); return ret; } EXPORT_SYMBOL(filemap_page_mkwrite); @@ -2426,8 +2434,6 @@ ssize_t __generic_file_aio_write(struct kiocb *iocb, const struct iovec *iov, count = ocount; pos = *ppos; - vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE); - /* We can write back this queue in page reclaim */ current->backing_dev_info = mapping->backing_dev_info; written = 0; @@ -2526,6 +2532,7 @@ ssize_t generic_file_aio_write(struct kiocb *iocb, const struct iovec *iov, BUG_ON(iocb->ki_pos != pos); + sb_start_write(inode->i_sb); mutex_lock(&inode->i_mutex); blk_start_plug(&plug); ret = __generic_file_aio_write(iocb, iov, nr_segs, &iocb->ki_pos); @@ -2539,6 +2546,7 @@ ssize_t generic_file_aio_write(struct kiocb *iocb, const struct iovec *iov, ret = err; } blk_finish_plug(&plug); + sb_end_write(inode->i_sb); return ret; } EXPORT_SYMBOL(generic_file_aio_write); diff --git a/mm/filemap_xip.c b/mm/filemap_xip.c index 80b34ef82dfe..13e013b1270c 100644 --- a/mm/filemap_xip.c +++ b/mm/filemap_xip.c @@ -402,6 +402,8 @@ xip_file_write(struct file *filp, const char __user *buf, size_t len, loff_t pos; ssize_t ret; + sb_start_write(inode->i_sb); + mutex_lock(&inode->i_mutex); if (!access_ok(VERIFY_READ, buf, len)) { @@ -412,8 +414,6 @@ xip_file_write(struct file *filp, const char __user *buf, size_t len, pos = *ppos; count = len; - vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE); - /* We can write back this queue in page reclaim */ current->backing_dev_info = mapping->backing_dev_info; @@ -437,6 +437,7 @@ xip_file_write(struct file *filp, const char __user *buf, size_t len, current->backing_dev_info = NULL; out_up: mutex_unlock(&inode->i_mutex); + sb_end_write(inode->i_sb); return ret; } EXPORT_SYMBOL_GPL(xip_file_write); -- cgit v1.2.3