1 // SPDX-License-Identifier: GPL-2.0 2 #include <linux/fs.h> 3 #include <linux/random.h> 4 #include <linux/buffer_head.h> 5 #include <linux/utsname.h> 6 #include <linux/kthread.h> 7 8 #include "ext4.h" 9 10 /* Checksumming functions */ 11 static __le32 ext4_mmp_csum(struct super_block *sb, struct mmp_struct *mmp) 12 { 13 struct ext4_sb_info *sbi = EXT4_SB(sb); 14 int offset = offsetof(struct mmp_struct, mmp_checksum); 15 __u32 csum; 16 17 csum = ext4_chksum(sbi, sbi->s_csum_seed, (char *)mmp, offset); 18 19 return cpu_to_le32(csum); 20 } 21 22 static int ext4_mmp_csum_verify(struct super_block *sb, struct mmp_struct *mmp) 23 { 24 if (!ext4_has_metadata_csum(sb)) 25 return 1; 26 27 return mmp->mmp_checksum == ext4_mmp_csum(sb, mmp); 28 } 29 30 static void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp) 31 { 32 if (!ext4_has_metadata_csum(sb)) 33 return; 34 35 mmp->mmp_checksum = ext4_mmp_csum(sb, mmp); 36 } 37 38 /* 39 * Write the MMP block using REQ_SYNC to try to get the block on-disk 40 * faster. 41 */ 42 static int write_mmp_block(struct super_block *sb, struct buffer_head *bh) 43 { 44 struct mmp_struct *mmp = (struct mmp_struct *)(bh->b_data); 45 46 /* 47 * We protect against freezing so that we don't create dirty buffers 48 * on frozen filesystem. 49 */ 50 sb_start_write(sb); 51 ext4_mmp_csum_set(sb, mmp); 52 lock_buffer(bh); 53 bh->b_end_io = end_buffer_write_sync; 54 get_bh(bh); 55 submit_bh(REQ_OP_WRITE, REQ_SYNC | REQ_META | REQ_PRIO, bh); 56 wait_on_buffer(bh); 57 sb_end_write(sb); 58 if (unlikely(!buffer_uptodate(bh))) 59 return 1; 60 61 return 0; 62 } 63 64 /* 65 * Read the MMP block. It _must_ be read from disk and hence we clear the 66 * uptodate flag on the buffer. 67 */ 68 static int read_mmp_block(struct super_block *sb, struct buffer_head **bh, 69 ext4_fsblk_t mmp_block) 70 { 71 struct mmp_struct *mmp; 72 int ret; 73 74 if (*bh) 75 clear_buffer_uptodate(*bh); 76 77 /* This would be sb_bread(sb, mmp_block), except we need to be sure 78 * that the MD RAID device cache has been bypassed, and that the read 79 * is not blocked in the elevator. */ 80 if (!*bh) { 81 *bh = sb_getblk(sb, mmp_block); 82 if (!*bh) { 83 ret = -ENOMEM; 84 goto warn_exit; 85 } 86 } 87 88 get_bh(*bh); 89 lock_buffer(*bh); 90 (*bh)->b_end_io = end_buffer_read_sync; 91 submit_bh(REQ_OP_READ, REQ_META | REQ_PRIO, *bh); 92 wait_on_buffer(*bh); 93 if (!buffer_uptodate(*bh)) { 94 ret = -EIO; 95 goto warn_exit; 96 } 97 mmp = (struct mmp_struct *)((*bh)->b_data); 98 if (le32_to_cpu(mmp->mmp_magic) != EXT4_MMP_MAGIC) { 99 ret = -EFSCORRUPTED; 100 goto warn_exit; 101 } 102 if (!ext4_mmp_csum_verify(sb, mmp)) { 103 ret = -EFSBADCRC; 104 goto warn_exit; 105 } 106 return 0; 107 warn_exit: 108 brelse(*bh); 109 *bh = NULL; 110 ext4_warning(sb, "Error %d while reading MMP block %llu", 111 ret, mmp_block); 112 return ret; 113 } 114 115 /* 116 * Dump as much information as possible to help the admin. 117 */ 118 void __dump_mmp_msg(struct super_block *sb, struct mmp_struct *mmp, 119 const char *function, unsigned int line, const char *msg) 120 { 121 __ext4_warning(sb, function, line, "%s", msg); 122 __ext4_warning(sb, function, line, 123 "MMP failure info: last update time: %llu, last update node: %.*s, last update device: %.*s", 124 (unsigned long long)le64_to_cpu(mmp->mmp_time), 125 (int)sizeof(mmp->mmp_nodename), mmp->mmp_nodename, 126 (int)sizeof(mmp->mmp_bdevname), mmp->mmp_bdevname); 127 } 128 129 /* 130 * kmmpd will update the MMP sequence every s_mmp_update_interval seconds 131 */ 132 static int kmmpd(void *data) 133 { 134 struct super_block *sb = ((struct mmpd_data *) data)->sb; 135 struct buffer_head *bh = ((struct mmpd_data *) data)->bh; 136 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 137 struct mmp_struct *mmp; 138 ext4_fsblk_t mmp_block; 139 u32 seq = 0; 140 unsigned long failed_writes = 0; 141 int mmp_update_interval = le16_to_cpu(es->s_mmp_update_interval); 142 unsigned mmp_check_interval; 143 unsigned long last_update_time; 144 unsigned long diff; 145 int retval; 146 147 mmp_block = le64_to_cpu(es->s_mmp_block); 148 mmp = (struct mmp_struct *)(bh->b_data); 149 mmp->mmp_time = cpu_to_le64(ktime_get_real_seconds()); 150 /* 151 * Start with the higher mmp_check_interval and reduce it if 152 * the MMP block is being updated on time. 153 */ 154 mmp_check_interval = max(EXT4_MMP_CHECK_MULT * mmp_update_interval, 155 EXT4_MMP_MIN_CHECK_INTERVAL); 156 mmp->mmp_check_interval = cpu_to_le16(mmp_check_interval); 157 BUILD_BUG_ON(sizeof(mmp->mmp_bdevname) < BDEVNAME_SIZE); 158 bdevname(bh->b_bdev, mmp->mmp_bdevname); 159 160 memcpy(mmp->mmp_nodename, init_utsname()->nodename, 161 sizeof(mmp->mmp_nodename)); 162 163 while (!kthread_should_stop()) { 164 if (++seq > EXT4_MMP_SEQ_MAX) 165 seq = 1; 166 167 mmp->mmp_seq = cpu_to_le32(seq); 168 mmp->mmp_time = cpu_to_le64(ktime_get_real_seconds()); 169 last_update_time = jiffies; 170 171 retval = write_mmp_block(sb, bh); 172 /* 173 * Don't spew too many error messages. Print one every 174 * (s_mmp_update_interval * 60) seconds. 175 */ 176 if (retval) { 177 if ((failed_writes % 60) == 0) { 178 ext4_error_err(sb, -retval, 179 "Error writing to MMP block"); 180 } 181 failed_writes++; 182 } 183 184 if (!(le32_to_cpu(es->s_feature_incompat) & 185 EXT4_FEATURE_INCOMPAT_MMP)) { 186 ext4_warning(sb, "kmmpd being stopped since MMP feature" 187 " has been disabled."); 188 goto exit_thread; 189 } 190 191 if (sb_rdonly(sb)) 192 break; 193 194 diff = jiffies - last_update_time; 195 if (diff < mmp_update_interval * HZ) 196 schedule_timeout_interruptible(mmp_update_interval * 197 HZ - diff); 198 199 /* 200 * We need to make sure that more than mmp_check_interval 201 * seconds have not passed since writing. If that has happened 202 * we need to check if the MMP block is as we left it. 203 */ 204 diff = jiffies - last_update_time; 205 if (diff > mmp_check_interval * HZ) { 206 struct buffer_head *bh_check = NULL; 207 struct mmp_struct *mmp_check; 208 209 retval = read_mmp_block(sb, &bh_check, mmp_block); 210 if (retval) { 211 ext4_error_err(sb, -retval, 212 "error reading MMP data: %d", 213 retval); 214 goto exit_thread; 215 } 216 217 mmp_check = (struct mmp_struct *)(bh_check->b_data); 218 if (mmp->mmp_seq != mmp_check->mmp_seq || 219 memcmp(mmp->mmp_nodename, mmp_check->mmp_nodename, 220 sizeof(mmp->mmp_nodename))) { 221 dump_mmp_msg(sb, mmp_check, 222 "Error while updating MMP info. " 223 "The filesystem seems to have been" 224 " multiply mounted."); 225 ext4_error_err(sb, EBUSY, "abort"); 226 put_bh(bh_check); 227 retval = -EBUSY; 228 goto exit_thread; 229 } 230 put_bh(bh_check); 231 } 232 233 /* 234 * Adjust the mmp_check_interval depending on how much time 235 * it took for the MMP block to be written. 236 */ 237 mmp_check_interval = max(min(EXT4_MMP_CHECK_MULT * diff / HZ, 238 EXT4_MMP_MAX_CHECK_INTERVAL), 239 EXT4_MMP_MIN_CHECK_INTERVAL); 240 mmp->mmp_check_interval = cpu_to_le16(mmp_check_interval); 241 } 242 243 /* 244 * Unmount seems to be clean. 245 */ 246 mmp->mmp_seq = cpu_to_le32(EXT4_MMP_SEQ_CLEAN); 247 mmp->mmp_time = cpu_to_le64(ktime_get_real_seconds()); 248 249 retval = write_mmp_block(sb, bh); 250 251 exit_thread: 252 EXT4_SB(sb)->s_mmp_tsk = NULL; 253 kfree(data); 254 brelse(bh); 255 return retval; 256 } 257 258 /* 259 * Get a random new sequence number but make sure it is not greater than 260 * EXT4_MMP_SEQ_MAX. 261 */ 262 static unsigned int mmp_new_seq(void) 263 { 264 u32 new_seq; 265 266 do { 267 new_seq = prandom_u32(); 268 } while (new_seq > EXT4_MMP_SEQ_MAX); 269 270 return new_seq; 271 } 272 273 /* 274 * Protect the filesystem from being mounted more than once. 275 */ 276 int ext4_multi_mount_protect(struct super_block *sb, 277 ext4_fsblk_t mmp_block) 278 { 279 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 280 struct buffer_head *bh = NULL; 281 struct mmp_struct *mmp = NULL; 282 struct mmpd_data *mmpd_data; 283 u32 seq; 284 unsigned int mmp_check_interval = le16_to_cpu(es->s_mmp_update_interval); 285 unsigned int wait_time = 0; 286 int retval; 287 288 if (mmp_block < le32_to_cpu(es->s_first_data_block) || 289 mmp_block >= ext4_blocks_count(es)) { 290 ext4_warning(sb, "Invalid MMP block in superblock"); 291 goto failed; 292 } 293 294 retval = read_mmp_block(sb, &bh, mmp_block); 295 if (retval) 296 goto failed; 297 298 mmp = (struct mmp_struct *)(bh->b_data); 299 300 if (mmp_check_interval < EXT4_MMP_MIN_CHECK_INTERVAL) 301 mmp_check_interval = EXT4_MMP_MIN_CHECK_INTERVAL; 302 303 /* 304 * If check_interval in MMP block is larger, use that instead of 305 * update_interval from the superblock. 306 */ 307 if (le16_to_cpu(mmp->mmp_check_interval) > mmp_check_interval) 308 mmp_check_interval = le16_to_cpu(mmp->mmp_check_interval); 309 310 seq = le32_to_cpu(mmp->mmp_seq); 311 if (seq == EXT4_MMP_SEQ_CLEAN) 312 goto skip; 313 314 if (seq == EXT4_MMP_SEQ_FSCK) { 315 dump_mmp_msg(sb, mmp, "fsck is running on the filesystem"); 316 goto failed; 317 } 318 319 wait_time = min(mmp_check_interval * 2 + 1, 320 mmp_check_interval + 60); 321 322 /* Print MMP interval if more than 20 secs. */ 323 if (wait_time > EXT4_MMP_MIN_CHECK_INTERVAL * 4) 324 ext4_warning(sb, "MMP interval %u higher than expected, please" 325 " wait.\n", wait_time * 2); 326 327 if (schedule_timeout_interruptible(HZ * wait_time) != 0) { 328 ext4_warning(sb, "MMP startup interrupted, failing mount\n"); 329 goto failed; 330 } 331 332 retval = read_mmp_block(sb, &bh, mmp_block); 333 if (retval) 334 goto failed; 335 mmp = (struct mmp_struct *)(bh->b_data); 336 if (seq != le32_to_cpu(mmp->mmp_seq)) { 337 dump_mmp_msg(sb, mmp, 338 "Device is already active on another node."); 339 goto failed; 340 } 341 342 skip: 343 /* 344 * write a new random sequence number. 345 */ 346 seq = mmp_new_seq(); 347 mmp->mmp_seq = cpu_to_le32(seq); 348 349 retval = write_mmp_block(sb, bh); 350 if (retval) 351 goto failed; 352 353 /* 354 * wait for MMP interval and check mmp_seq. 355 */ 356 if (schedule_timeout_interruptible(HZ * wait_time) != 0) { 357 ext4_warning(sb, "MMP startup interrupted, failing mount"); 358 goto failed; 359 } 360 361 retval = read_mmp_block(sb, &bh, mmp_block); 362 if (retval) 363 goto failed; 364 mmp = (struct mmp_struct *)(bh->b_data); 365 if (seq != le32_to_cpu(mmp->mmp_seq)) { 366 dump_mmp_msg(sb, mmp, 367 "Device is already active on another node."); 368 goto failed; 369 } 370 371 mmpd_data = kmalloc(sizeof(*mmpd_data), GFP_KERNEL); 372 if (!mmpd_data) { 373 ext4_warning(sb, "not enough memory for mmpd_data"); 374 goto failed; 375 } 376 mmpd_data->sb = sb; 377 mmpd_data->bh = bh; 378 379 /* 380 * Start a kernel thread to update the MMP block periodically. 381 */ 382 EXT4_SB(sb)->s_mmp_tsk = kthread_run(kmmpd, mmpd_data, "kmmpd-%.*s", 383 (int)sizeof(mmp->mmp_bdevname), 384 bdevname(bh->b_bdev, 385 mmp->mmp_bdevname)); 386 if (IS_ERR(EXT4_SB(sb)->s_mmp_tsk)) { 387 EXT4_SB(sb)->s_mmp_tsk = NULL; 388 kfree(mmpd_data); 389 ext4_warning(sb, "Unable to create kmmpd thread for %s.", 390 sb->s_id); 391 goto failed; 392 } 393 394 return 0; 395 396 failed: 397 brelse(bh); 398 return 1; 399 } 400 401 402