"value":"In the Linux kernel, the following vulnerability has been resolved:\n\nx86/mm/pat: fix VM_PAT handling in COW mappings\n\nPAT handling won't do the right thing in COW mappings: the first PTE (or,\nin fact, all PTEs) can be replaced during write faults to point at anon\nfolios. Reliably recovering the correct PFN and cachemode using\nfollow_phys() from PTEs will not work in COW mappings.\n\nUsing follow_phys(), we might just get the address+protection of the anon\nfolio (which is very wrong), or fail on swap/nonswap entries, failing\nfollow_phys() and triggering a WARN_ON_ONCE() in untrack_pfn() and\ntrack_pfn_copy(), not properly calling free_pfn_range().\n\nIn free_pfn_range(), we either wouldn't call memtype_free() or would call\nit with the wrong range, possibly leaking memory.\n\nTo fix that, let's update follow_phys() to refuse returning anon folios,\nand fallback to using the stored PFN inside vma->vm_pgoff for COW mappings\nif we run into that.\n\nWe will now properly handle untrack_pfn() with COW mappings, where we\ndon't need the cachemode. We'll have to fail fork()->track_pfn_copy() if\nthe first page was replaced by an anon folio, though: we'd have to store\nthe cachemode in the VMA to make this work, likely growing the VMA size.\n\nFor now, lets keep it simple and let track_pfn_copy() just fail in that\ncase: it would have failed in the past with swap/nonswap entries already,\nand it would have done the wrong thing with anon folios.\n\nSimple reproducer to trigger the WARN_ON_ONCE() in untrack_pfn():\n\n<--- C reproducer --->\n #include <stdio.h>\n #include <sys/mman.h>\n #include <unistd.h>\n #include <liburing.h>\n\n int main(void)\n {\n struct io_uring_params p = {};\n int ring_fd;\n size_t size;\n char *map;\n\n ring_fd = io_uring_setup(1, &p);\n if (ring_fd < 0) {\n perror(\"io_uring_setup\");\n return 1;\n }\n size = p.sq_off.array + p.sq_entries * sizeof(unsigned);\n\n /* Map the submission queue ring MAP_PRIVATE */\n map = mmap(0, size, PROT_READ | PROT_WRITE, MAP_PRIVATE,\n ring_fd, IORING_OFF_SQ_RING);\n if (map == MAP_FAILED) {\n perror(\"mmap\");\nreturn1;\n}\n\n/*Wehaveatleastonepage.Let'sCOWit.*/\n*map=0;\npause();\nreturn0;\n}\n<---Creproducer--->\n\nOnasystemwith16GiBRAMandswapconfigured:\n#./iouring&\n#memhog16G\n#killalliouring\n[301.552930]------------[cuthere]------------\n[301.553285]WARNING:CPU:7PID:1402atarch/x86/mm/pat/memtype.c:1060untrack_pfn+0xf4/0x100\n[301.553989]Moduleslinkedin:binfmt_miscnft_fib_inetnft_fib_ipv4nft_fib_ipv6nft_fibnft_reject_g\n[301.558232]CPU:7PID:1402Comm:iouringNottainted6.7.5-100.fc38.x86_64#1\n[301.558772]Hardwarename:QEMUStandardPC(Q35+ICH9,2009),BIOSrel-1.16.3-0-ga6ed6b701f0a-prebu4\n[301.559569]RIP:0010:untrack_pfn+0xf4/0x100\n[301.559893]Code:75c4ebcf488b43108ba8e80000003b6b2874b8488b7b30e8ea1af7000\n[301.561189]RSP:0018:ffffba2c0377fab8EFLAGS:00010282\n[301.561590]RAX:00000000ffffffeaRBX:ffff9208c8ce9cc0RCX:000000010455e047\n[301.562105]RDX:07fffffff0eb1e0aRSI:0000000000000000RDI:ffff9208c391d200\n[301.562628]RBP:0000000000000000R08:ffffba2c0377fab8R09:0000000000000000\n[301.563145]R10:ffff9208d2292d50R11:0000000000000002R12:00007fea890e0000\n[301.563669]R13:0000000000000000R14:ffffba2c0377fc08R15:0000000000000000\n[301.564186]FS:0000000000000000(0000)GS:ffff920c2fbc0000(0000)knlGS:0000000000000000\n[301.564773]CS:0010DS:0000ES:0000CR0:0000000080050033\n[301.565197]CR2:00007fea88ee8a20CR3:00000001033a8000CR4:0000000000750ef0\n[301.565725]PKRU:55555554\n[301.565944]CallTrace:\n[301.566148]<TASK>\n[301.566325]?untrack_pfn+0xf4/0x100\n[301.566618]?__warn+0x81/0x130\n[301.566876]?untrack_pfn+0xf4/0