"value":"In the Linux kernel, the following vulnerability has been resolved:\n\nnilfs2: fix state management in error path of log writing function\n\nAfter commit a694291a6211 (\"nilfs2: separate wait function from\nnilfs_segctor_write\") was applied, the log writing function\nnilfs_segctor_do_construct() was able to issue I/O requests continuously\neven if user data blocks were split into multiple logs across segments,\nbut two potential flaws were introduced in its error handling.\n\nFirst, if nilfs_segctor_begin_construction() fails while creating the\nsecond or subsequent logs, the log writing function returns without\ncalling nilfs_segctor_abort_construction(), so the writeback flag set on\npages/folios will remain uncleared. This causes page cache operations to\nhang waiting for the writeback flag. For example,\ntruncate_inode_pages_final(), which is called via nilfs_evict_inode() when\nan inode is evicted from memory, will hang.\n\nSecond, the NILFS_I_COLLECTED flag set on normal inodes remain uncleared. \nAs a result, if the next log write involves checkpoint creation, that's\nfine, but if a partial log write is performed that does not, inodes with\nNILFS_I_COLLECTED set are erroneously removed from the \"sc_dirty_files\"\nlist, and their data and b-tree blocks may not be written to the device,\ncorrupting the block mapping.\n\nFix these issues by uniformly calling nilfs_segctor_abort_construction()\non failure of each step in the loop in nilfs_segctor_do_construct(),\nhaving it clean up logs and segment usages according to progress, and\ncorrecting the conditions for calling nilfs_redirty_inodes() to ensure\nthat the NILFS_I_COLLECTED flag is cleared."
