| Thomas Tai | cf4182f | 2018-02-08 13:38:30 -0500 | [diff] [blame] | 1 | NVMe Fault Injection |
| 2 | ==================== |
| 3 | Linux's fault injection framework provides a systematic way to support |
| 4 | error injection via debugfs in the /sys/kernel/debug directory. When |
| 5 | enabled, the default NVME_SC_INVALID_OPCODE with no retry will be |
| 6 | injected into the nvme_end_request. Users can change the default status |
| 7 | code and no retry flag via the debugfs. The list of Generic Command |
| 8 | Status can be found in include/linux/nvme.h |
| 9 | |
| 10 | Following examples show how to inject an error into the nvme. |
| 11 | |
| 12 | First, enable CONFIG_FAULT_INJECTION_DEBUG_FS kernel config, |
| 13 | recompile the kernel. After booting up the kernel, do the |
| 14 | following. |
| 15 | |
| 16 | Example 1: Inject default status code with no retry |
| 17 | --------------------------------------------------- |
| 18 | |
| 19 | mount /dev/nvme0n1 /mnt |
| 20 | echo 1 > /sys/kernel/debug/nvme0n1/fault_inject/times |
| 21 | echo 100 > /sys/kernel/debug/nvme0n1/fault_inject/probability |
| 22 | cp a.file /mnt |
| 23 | |
| 24 | Expected Result: |
| 25 | |
| 26 | cp: cannot stat ‘/mnt/a.file’: Input/output error |
| 27 | |
| 28 | Message from dmesg: |
| 29 | |
| 30 | FAULT_INJECTION: forcing a failure. |
| 31 | name fault_inject, interval 1, probability 100, space 0, times 1 |
| 32 | CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.15.0-rc8+ #2 |
| 33 | Hardware name: innotek GmbH VirtualBox/VirtualBox, |
| 34 | BIOS VirtualBox 12/01/2006 |
| 35 | Call Trace: |
| 36 | <IRQ> |
| 37 | dump_stack+0x5c/0x7d |
| 38 | should_fail+0x148/0x170 |
| 39 | nvme_should_fail+0x2f/0x50 [nvme_core] |
| 40 | nvme_process_cq+0xe7/0x1d0 [nvme] |
| 41 | nvme_irq+0x1e/0x40 [nvme] |
| 42 | __handle_irq_event_percpu+0x3a/0x190 |
| 43 | handle_irq_event_percpu+0x30/0x70 |
| 44 | handle_irq_event+0x36/0x60 |
| 45 | handle_fasteoi_irq+0x78/0x120 |
| 46 | handle_irq+0xa7/0x130 |
| 47 | ? tick_irq_enter+0xa8/0xc0 |
| 48 | do_IRQ+0x43/0xc0 |
| 49 | common_interrupt+0xa2/0xa2 |
| 50 | </IRQ> |
| 51 | RIP: 0010:native_safe_halt+0x2/0x10 |
| 52 | RSP: 0018:ffffffff82003e90 EFLAGS: 00000246 ORIG_RAX: ffffffffffffffdd |
| 53 | RAX: ffffffff817a10c0 RBX: ffffffff82012480 RCX: 0000000000000000 |
| 54 | RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000 |
| 55 | RBP: 0000000000000000 R08: 000000008e38ce64 R09: 0000000000000000 |
| 56 | R10: 0000000000000000 R11: 0000000000000000 R12: ffffffff82012480 |
| 57 | R13: ffffffff82012480 R14: 0000000000000000 R15: 0000000000000000 |
| 58 | ? __sched_text_end+0x4/0x4 |
| 59 | default_idle+0x18/0xf0 |
| 60 | do_idle+0x150/0x1d0 |
| 61 | cpu_startup_entry+0x6f/0x80 |
| 62 | start_kernel+0x4c4/0x4e4 |
| 63 | ? set_init_arg+0x55/0x55 |
| 64 | secondary_startup_64+0xa5/0xb0 |
| 65 | print_req_error: I/O error, dev nvme0n1, sector 9240 |
| 66 | EXT4-fs error (device nvme0n1): ext4_find_entry:1436: |
| 67 | inode #2: comm cp: reading directory lblock 0 |
| 68 | |
| 69 | Example 2: Inject default status code with retry |
| 70 | ------------------------------------------------ |
| 71 | |
| 72 | mount /dev/nvme0n1 /mnt |
| 73 | echo 1 > /sys/kernel/debug/nvme0n1/fault_inject/times |
| 74 | echo 100 > /sys/kernel/debug/nvme0n1/fault_inject/probability |
| 75 | echo 1 > /sys/kernel/debug/nvme0n1/fault_inject/status |
| 76 | echo 0 > /sys/kernel/debug/nvme0n1/fault_inject/dont_retry |
| 77 | |
| 78 | cp a.file /mnt |
| 79 | |
| 80 | Expected Result: |
| 81 | |
| 82 | command success without error |
| 83 | |
| 84 | Message from dmesg: |
| 85 | |
| 86 | FAULT_INJECTION: forcing a failure. |
| 87 | name fault_inject, interval 1, probability 100, space 0, times 1 |
| 88 | CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.15.0-rc8+ #4 |
| 89 | Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 |
| 90 | Call Trace: |
| 91 | <IRQ> |
| 92 | dump_stack+0x5c/0x7d |
| 93 | should_fail+0x148/0x170 |
| 94 | nvme_should_fail+0x30/0x60 [nvme_core] |
| 95 | nvme_loop_queue_response+0x84/0x110 [nvme_loop] |
| 96 | nvmet_req_complete+0x11/0x40 [nvmet] |
| 97 | nvmet_bio_done+0x28/0x40 [nvmet] |
| 98 | blk_update_request+0xb0/0x310 |
| 99 | blk_mq_end_request+0x18/0x60 |
| 100 | flush_smp_call_function_queue+0x3d/0xf0 |
| 101 | smp_call_function_single_interrupt+0x2c/0xc0 |
| 102 | call_function_single_interrupt+0xa2/0xb0 |
| 103 | </IRQ> |
| 104 | RIP: 0010:native_safe_halt+0x2/0x10 |
| 105 | RSP: 0018:ffffc9000068bec0 EFLAGS: 00000246 ORIG_RAX: ffffffffffffff04 |
| 106 | RAX: ffffffff817a10c0 RBX: ffff88011a3c9680 RCX: 0000000000000000 |
| 107 | RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000 |
| 108 | RBP: 0000000000000001 R08: 000000008e38c131 R09: 0000000000000000 |
| 109 | R10: 0000000000000000 R11: 0000000000000000 R12: ffff88011a3c9680 |
| 110 | R13: ffff88011a3c9680 R14: 0000000000000000 R15: 0000000000000000 |
| 111 | ? __sched_text_end+0x4/0x4 |
| 112 | default_idle+0x18/0xf0 |
| 113 | do_idle+0x150/0x1d0 |
| 114 | cpu_startup_entry+0x6f/0x80 |
| 115 | start_secondary+0x187/0x1e0 |
| 116 | secondary_startup_64+0xa5/0xb0 |