CVE-2026-31702

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> f2fs: fix use-after-free of sbi in f2fs_compress_write_end_io()<br /> <br /> In f2fs_compress_write_end_io(), dec_page_count(sbi, type) can bring<br /> the F2FS_WB_CP_DATA counter to zero, unblocking<br /> f2fs_wait_on_all_pages() in f2fs_put_super() on a concurrent unmount<br /> CPU. The unmount path then proceeds to call<br /> f2fs_destroy_page_array_cache(sbi), which destroys<br /> sbi-&gt;page_array_slab via kmem_cache_destroy(), and eventually<br /> kfree(sbi). Meanwhile, the bio completion callback is still executing:<br /> when it reaches page_array_free(sbi, ...), it dereferences<br /> sbi-&gt;page_array_slab — a destroyed slab cache — to call<br /> kmem_cache_free(), causing a use-after-free.<br /> <br /> This is the same class of bug as CVE-2026-23234 (which fixed the<br /> equivalent race in f2fs_write_end_io() in data.c), but in the<br /> compressed writeback completion path that was not covered by that fix.<br /> <br /> Fix this by moving dec_page_count() to after page_array_free(), so<br /> that all sbi accesses complete before the counter decrement that can<br /> unblock unmount. For non-last folios (where atomic_dec_return on<br /> cic-&gt;pending_pages is nonzero), dec_page_count is called immediately<br /> before returning — page_array_free is not reached on this path, so<br /> there is no post-decrement sbi access. For the last folio,<br /> page_array_free runs while the F2FS_WB_CP_DATA counter is still<br /> nonzero (this folio has not yet decremented it), keeping sbi alive,<br /> and dec_page_count runs as the final operation.