Wind River Support Network

Description

In the Linux kernel, the following vulnerability has been resolved:[EOL][EOL]mm/memory-failure: fix VM_BUG_ON_PAGE(PagePoisoned(page)) when unpoison memory[EOL][EOL]When I did memory failure tests, below panic occurs:[EOL][EOL]page dumped because: VM_BUG_ON_PAGE(PagePoisoned(page))[EOL]kernel BUG at include/linux/page-flags.h:616![EOL]Oops: invalid opcode: 0000 [#1] PREEMPT SMP NOPTI[EOL]CPU: 3 PID: 720 Comm: bash Not tainted 6.10.0-rc1-00195-g148743902568 #40[EOL]RIP: 0010:unpoison_memory+0x2f3/0x590[EOL]RSP: 0018:ffffa57fc8787d60 EFLAGS: 00000246[EOL]RAX: 0000000000000037 RBX: 0000000000000009 RCX: ffff9be25fcdc9c8[EOL]RDX: 0000000000000000 RSI: 0000000000000027 RDI: ffff9be25fcdc9c0[EOL]RBP: 0000000000300000 R08: ffffffffb4956f88 R09: 0000000000009ffb[EOL]R10: 0000000000000284 R11: ffffffffb4926fa0 R12: ffffe6b00c000000[EOL]R13: ffff9bdb453dfd00 R14: 0000000000000000 R15: fffffffffffffffe[EOL]FS:  00007f08f04e4740(0000) GS:ffff9be25fcc0000(0000) knlGS:0000000000000000[EOL]CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033[EOL]CR2: 0000564787a30410 CR3: 000000010d4e2000 CR4: 00000000000006f0[EOL]Call Trace:[EOL] <TASK>[EOL] unpoison_memory+0x2f3/0x590[EOL] simple_attr_write_xsigned.constprop.0.isra.0+0xb3/0x110[EOL] debugfs_attr_write+0x42/0x60[EOL] full_proxy_write+0x5b/0x80[EOL] vfs_write+0xd5/0x540[EOL] ksys_write+0x64/0xe0[EOL] do_syscall_64+0xb9/0x1d0[EOL] entry_SYSCALL_64_after_hwframe+0x77/0x7f[EOL]RIP: 0033:0x7f08f0314887[EOL]RSP: 002b:00007ffece710078 EFLAGS: 00000246 ORIG_RAX: 0000000000000001[EOL]RAX: ffffffffffffffda RBX: 0000000000000009 RCX: 00007f08f0314887[EOL]RDX: 0000000000000009 RSI: 0000564787a30410 RDI: 0000000000000001[EOL]RBP: 0000564787a30410 R08: 000000000000fefe R09: 000000007fffffff[EOL]R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000009[EOL]R13: 00007f08f041b780 R14: 00007f08f0417600 R15: 00007f08f0416a00[EOL] </TASK>[EOL]Modules linked in: hwpoison_inject[EOL]---[ end trace 0000000000000000 ]---[EOL]RIP: 0010:unpoison_memory+0x2f3/0x590[EOL]RSP: 0018:ffffa57fc8787d60 EFLAGS: 00000246[EOL]RAX: 0000000000000037 RBX: 0000000000000009 RCX: ffff9be25fcdc9c8[EOL]RDX: 0000000000000000 RSI: 0000000000000027 RDI: ffff9be25fcdc9c0[EOL]RBP: 0000000000300000 R08: ffffffffb4956f88 R09: 0000000000009ffb[EOL]R10: 0000000000000284 R11: ffffffffb4926fa0 R12: ffffe6b00c000000[EOL]R13: ffff9bdb453dfd00 R14: 0000000000000000 R15: fffffffffffffffe[EOL]FS:  00007f08f04e4740(0000) GS:ffff9be25fcc0000(0000) knlGS:0000000000000000[EOL]CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033[EOL]CR2: 0000564787a30410 CR3: 000000010d4e2000 CR4: 00000000000006f0[EOL]Kernel panic - not syncing: Fatal exception[EOL]Kernel Offset: 0x31c00000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)[EOL]---[ end Kernel panic - not syncing: Fatal exception ]---[EOL][EOL]The root cause is that unpoison_memory() tries to check the PG_HWPoison[EOL]flags of an uninitialized page.  So VM_BUG_ON_PAGE(PagePoisoned(page)) is[EOL]triggered.  This can be reproduced by below steps:[EOL][EOL]1.Offline memory block:[EOL][EOL] echo offline > /sys/devices/system/memory/memory12/state[EOL][EOL]2.Get offlined memory pfn:[EOL][EOL] page-types -b n -rlN[EOL][EOL]3.Write pfn to unpoison-pfn[EOL][EOL] echo <pfn> > /sys/kernel/debug/hwpoison/unpoison-pfn[EOL][EOL]This scenario can be identified by pfn_to_online_page() returning NULL. [EOL]And ZONE_DEVICE pages are never expected, so we can simply fail if[EOL]pfn_to_online_page() == NULL to fix the bug.