1What is hwpoison? 2 3Upcoming Intel CPUs have support for recovering from some memory errors 4(``MCA recovery''). This requires the OS to declare a page "poisoned", 5kill the processes associated with it and avoid using it in the future. 6 7This patchkit implements the necessary infrastructure in the VM. 8 9To quote the overview comment: 10 11 * High level machine check handler. Handles pages reported by the 12 * hardware as being corrupted usually due to a 2bit ECC memory or cache 13 * failure. 14 * 15 * This focusses on pages detected as corrupted in the background. 16 * When the current CPU tries to consume corruption the currently 17 * running process can just be killed directly instead. This implies 18 * that if the error cannot be handled for some reason it's safe to 19 * just ignore it because no corruption has been consumed yet. Instead 20 * when that happens another machine check will happen. 21 * 22 * Handles page cache pages in various states. The tricky part 23 * here is that we can access any page asynchronous to other VM 24 * users, because memory failures could happen anytime and anywhere, 25 * possibly violating some of their assumptions. This is why this code 26 * has to be extremely careful. Generally it tries to use normal locking 27 * rules, as in get the standard locks, even if that means the 28 * error handling takes potentially a long time. 29 * 30 * Some of the operations here are somewhat inefficient and have non 31 * linear algorithmic complexity, because the data structures have not 32 * been optimized for this case. This is in particular the case 33 * for the mapping from a vma to a process. Since this case is expected 34 * to be rare we hope we can get away with this. 35 36The code consists of a the high level handler in mm/memory-failure.c, 37a new page poison bit and various checks in the VM to handle poisoned 38pages. 39 40The main target right now is KVM guests, but it works for all kinds 41of applications. KVM support requires a recent qemu-kvm release. 42 43For the KVM use there was need for a new signal type so that 44KVM can inject the machine check into the guest with the proper 45address. This in theory allows other applications to handle 46memory failures too. The expection is that near all applications 47won't do that, but some very specialized ones might. 48 49--- 50 51There are two (actually three) modi memory failure recovery can be in: 52 53vm.memory_failure_recovery sysctl set to zero: 54 All memory failures cause a panic. Do not attempt recovery. 55 (on x86 this can be also affected by the tolerant level of the 56 MCE subsystem) 57 58early kill 59 (can be controlled globally and per process) 60 Send SIGBUS to the application as soon as the error is detected 61 This allows applications who can process memory errors in a gentle 62 way (e.g. drop affected object) 63 This is the mode used by KVM qemu. 64 65late kill 66 Send SIGBUS when the application runs into the corrupted page. 67 This is best for memory error unaware applications and default 68 Note some pages are always handled as late kill. 69 70--- 71 72User control: 73 74vm.memory_failure_recovery 75 See sysctl.txt 76 77vm.memory_failure_early_kill 78 Enable early kill mode globally 79 80PR_MCE_KILL 81 Set early/late kill mode/revert to system default 82 arg1: PR_MCE_KILL_CLEAR: Revert to system default 83 arg1: PR_MCE_KILL_SET: arg2 defines thread specific mode 84 PR_MCE_KILL_EARLY: Early kill 85 PR_MCE_KILL_LATE: Late kill 86 PR_MCE_KILL_DEFAULT: Use system global default 87 Note that if you want to have a dedicated thread which handles 88 the SIGBUS(BUS_MCEERR_AO) on behalf of the process, you should 89 call prctl(PR_MCE_KILL_EARLY) on the designated thread. Otherwise, 90 the SIGBUS is sent to the main thread. 91 92PR_MCE_KILL_GET 93 return current mode 94 95 96--- 97 98Testing: 99 100madvise(MADV_HWPOISON, ....) 101 (as root) 102 Poison a page in the process for testing 103 104 105hwpoison-inject module through debugfs 106 107/sys/debug/hwpoison/ 108 109corrupt-pfn 110 111Inject hwpoison fault at PFN echoed into this file. This does 112some early filtering to avoid corrupted unintended pages in test suites. 113 114unpoison-pfn 115 116Software-unpoison page at PFN echoed into this file. This 117way a page can be reused again. 118This only works for Linux injected failures, not for real 119memory failures. 120 121Note these injection interfaces are not stable and might change between 122kernel versions 123 124corrupt-filter-dev-major 125corrupt-filter-dev-minor 126 127Only handle memory failures to pages associated with the file system defined 128by block device major/minor. -1U is the wildcard value. 129This should be only used for testing with artificial injection. 130 131corrupt-filter-memcg 132 133Limit injection to pages owned by memgroup. Specified by inode number 134of the memcg. 135 136Example: 137 mkdir /sys/fs/cgroup/mem/hwpoison 138 139 usemem -m 100 -s 1000 & 140 echo `jobs -p` > /sys/fs/cgroup/mem/hwpoison/tasks 141 142 memcg_ino=$(ls -id /sys/fs/cgroup/mem/hwpoison | cut -f1 -d' ') 143 echo $memcg_ino > /debug/hwpoison/corrupt-filter-memcg 144 145 page-types -p `pidof init` --hwpoison # shall do nothing 146 page-types -p `pidof usemem` --hwpoison # poison its pages 147 148corrupt-filter-flags-mask 149corrupt-filter-flags-value 150 151When specified, only poison pages if ((page_flags & mask) == value). 152This allows stress testing of many kinds of pages. The page_flags 153are the same as in /proc/kpageflags. The flag bits are defined in 154include/linux/kernel-page-flags.h and documented in 155Documentation/vm/pagemap.txt 156 157Architecture specific MCE injector 158 159x86 has mce-inject, mce-test 160 161Some portable hwpoison test programs in mce-test, see blow. 162 163--- 164 165References: 166 167http://halobates.de/mce-lc09-2.pdf 168 Overview presentation from LinuxCon 09 169 170git://git.kernel.org/pub/scm/utils/cpu/mce/mce-test.git 171 Test suite (hwpoison specific portable tests in tsrc) 172 173git://git.kernel.org/pub/scm/utils/cpu/mce/mce-inject.git 174 x86 specific injector 175 176 177--- 178 179Limitations: 180 181- Not all page types are supported and never will. Most kernel internal 182objects cannot be recovered, only LRU pages for now. 183- Right now hugepage support is missing. 184 185--- 186Andi Kleen, Oct 2009 187 188