fs/ext4/inode.c

254 			     "couldn't mark inode dirty (err %d)", err);
311 	 * Check out some where else accidentally dirty the evicting inode,
979  * dirty so that writeback code knows about this page (and inode) contains
980  * dirty data. ext4_writepages() then commits appropriate transaction to
992 	int dirty = buffer_dirty(bh);
999 	 * the dirty bit as jbd2_journal_get_write_access() could complain
1000 	 * otherwise about fs integrity issues. Setting of the dirty bit
1005 	if (dirty)
1010 	if (!ret && dirty)
1305 	 * Don't mark the inode dirty under folio lock. First, it unnecessarily
1340  * set the buffer to be dirty, since in data=journalled mode we need
1510 	/* update fs dirty data blocks counter */
2187  *                     dirty pages to avoid infinite loops.
2279 				       "Failed to mark inode %lu dirty",
2343  * mpage_prepare_extent_to_map - find & lock contiguous range of dirty pages
2348  * Walk dirty pages in the mapping. If they are fully mapped, submit them for
2350  * buffers in the page for IO and keep page dirty. When we can map blocks and
2401 			 * Accumulated enough dirty pages? This doesn't apply
2405 			 * newly appeared dirty pages, but have not synced all
2406 			 * of the old dirty pages.
2426 			 * If the page is no longer dirty, or its mapping no
2486 				/* Add all dirty buffers to mpd */
2592 		 * the page and we may dirty the inode.
2837 	 * Start pushing delalloc when 1/2 of free blocks are dirty.
2845 		 * free block count is less than 150% of dirty blocks
2949 	 * block_write_end() will mark the inode as dirty with I_DIRTY_PAGES
3185 	/* Page has dirty journalled data -> cannot release */
3221 	 * so consider them to be dirty for the purpose of O_DSYNC, even if
3483  * For data=journal mode, folio should be marked dirty only when it was
3490  * transaction but we cannot due to lock ordering.  We cannot just dirty the
3491  * folio and leave attached buffers clean, because the buffers' dirty state is
3492  * "definitive".  We cannot just set the buffers dirty or jbddirty because all
3494  * "pending dirty" and next time ext4_writepages() is called, attach buffers
3882 	 * Write out all dirty pages to avoid race conditions
4125 	 * marks the inode dirty, to catch the new size.
5143  * superblock's dirty inode list.
5210 	 * strip all buffers from the folio but keep the folio dirty which can then
5211 	 * confuse e.g. concurrent ext4_writepages() seeing dirty folio without
5248  * dirty pages and freeing of disk blocks, we can guarantee that any
5256  * we start writeout of all the dirty pages which are being truncated.
5878  * dirtiness (it may still be data-dirty).
5921  * Also, dquot_alloc_block() will always dirty the inode when blocks
5967 	 * to flush all dirty data the inode has. There can be outstanding
5970 	 * dirty data which can be converted only after flushing the dirty
6012 	/* Finally we can mark the inode as dirty. */
6126 	 * thus might hit warning messages for dirty metadata buffers.