From 3bf92d26e3256d76766e77106502a68c9e30c14b Mon Sep 17 00:00:00 2001 From: Michael Vrable Date: Mon, 19 Jul 2010 15:31:40 -0700 Subject: [PATCH] Add checksumming to filesystem journal. This will be used to check for consistency during log recovery. --- bluesky/CMakeLists.txt | 2 +- bluesky/bluesky-private.h | 4 ++ bluesky/crc32c.c | 105 ++++++++++++++++++++++++++++++++++++++ bluesky/log.c | 43 +++++++++++++++- 4 files changed, 152 insertions(+), 2 deletions(-) create mode 100644 bluesky/crc32c.c diff --git a/bluesky/CMakeLists.txt b/bluesky/CMakeLists.txt index 3b88d16..daaeccb 100644 --- a/bluesky/CMakeLists.txt +++ b/bluesky/CMakeLists.txt @@ -3,7 +3,7 @@ include_directories("${LIBS3_BUILD_DIR}/include" ${KVSTORE_DIR}) link_directories("${LIBS3_BUILD_DIR}/lib" ${KVSTORE_DIR}) add_library(bluesky SHARED - cache.c crypto.c debug.c dir.c file.c init.c inode.c log.c + cache.c crc32c.c crypto.c debug.c dir.c file.c init.c inode.c log.c serialize.c store.c store-bdb.c store-kv.cc store-multi.c store-s3.c util.c) add_executable(bluesky-test main.c) diff --git a/bluesky/bluesky-private.h b/bluesky/bluesky-private.h index 039e253..83a383e 100644 --- a/bluesky/bluesky-private.h +++ b/bluesky/bluesky-private.h @@ -152,6 +152,10 @@ void bluesky_file_drop_cached(BlueSkyInode *inode); /* Logging infrastructure for ensuring operations are persistently recorded to * disk. */ +#define BLUESKY_CRC32C_SEED (~(uint32_t)0) +uint32_t crc32c(uint32_t crc, const char *buf, unsigned int length); +uint32_t crc32c_finalize(uint32_t crc); + struct _BlueSkyLog { char *log_directory; GAsyncQueue *queue; diff --git a/bluesky/crc32c.c b/bluesky/crc32c.c new file mode 100644 index 0000000..b29f18d --- /dev/null +++ b/bluesky/crc32c.c @@ -0,0 +1,105 @@ +/* Blue Sky: File Systems in the Cloud + * + * This file implements the Castagnoli CRC-32 computation, and is taken from + * crypto/crc32c.c in the Linux kernel: + * + * Copyright (c) 2004 Cisco Systems, Inc. + * Copyright (c) 2008 Herbert Xu + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * Adapted by Michael Vrable to fit with BlueSky. + */ + +#include +#include "bluesky-private.h" + +static const uint32_t crc32c_table[256] = { + 0x00000000L, 0xF26B8303L, 0xE13B70F7L, 0x1350F3F4L, + 0xC79A971FL, 0x35F1141CL, 0x26A1E7E8L, 0xD4CA64EBL, + 0x8AD958CFL, 0x78B2DBCCL, 0x6BE22838L, 0x9989AB3BL, + 0x4D43CFD0L, 0xBF284CD3L, 0xAC78BF27L, 0x5E133C24L, + 0x105EC76FL, 0xE235446CL, 0xF165B798L, 0x030E349BL, + 0xD7C45070L, 0x25AFD373L, 0x36FF2087L, 0xC494A384L, + 0x9A879FA0L, 0x68EC1CA3L, 0x7BBCEF57L, 0x89D76C54L, + 0x5D1D08BFL, 0xAF768BBCL, 0xBC267848L, 0x4E4DFB4BL, + 0x20BD8EDEL, 0xD2D60DDDL, 0xC186FE29L, 0x33ED7D2AL, + 0xE72719C1L, 0x154C9AC2L, 0x061C6936L, 0xF477EA35L, + 0xAA64D611L, 0x580F5512L, 0x4B5FA6E6L, 0xB93425E5L, + 0x6DFE410EL, 0x9F95C20DL, 0x8CC531F9L, 0x7EAEB2FAL, + 0x30E349B1L, 0xC288CAB2L, 0xD1D83946L, 0x23B3BA45L, + 0xF779DEAEL, 0x05125DADL, 0x1642AE59L, 0xE4292D5AL, + 0xBA3A117EL, 0x4851927DL, 0x5B016189L, 0xA96AE28AL, + 0x7DA08661L, 0x8FCB0562L, 0x9C9BF696L, 0x6EF07595L, + 0x417B1DBCL, 0xB3109EBFL, 0xA0406D4BL, 0x522BEE48L, + 0x86E18AA3L, 0x748A09A0L, 0x67DAFA54L, 0x95B17957L, + 0xCBA24573L, 0x39C9C670L, 0x2A993584L, 0xD8F2B687L, + 0x0C38D26CL, 0xFE53516FL, 0xED03A29BL, 0x1F682198L, + 0x5125DAD3L, 0xA34E59D0L, 0xB01EAA24L, 0x42752927L, + 0x96BF4DCCL, 0x64D4CECFL, 0x77843D3BL, 0x85EFBE38L, + 0xDBFC821CL, 0x2997011FL, 0x3AC7F2EBL, 0xC8AC71E8L, + 0x1C661503L, 0xEE0D9600L, 0xFD5D65F4L, 0x0F36E6F7L, + 0x61C69362L, 0x93AD1061L, 0x80FDE395L, 0x72966096L, + 0xA65C047DL, 0x5437877EL, 0x4767748AL, 0xB50CF789L, + 0xEB1FCBADL, 0x197448AEL, 0x0A24BB5AL, 0xF84F3859L, + 0x2C855CB2L, 0xDEEEDFB1L, 0xCDBE2C45L, 0x3FD5AF46L, + 0x7198540DL, 0x83F3D70EL, 0x90A324FAL, 0x62C8A7F9L, + 0xB602C312L, 0x44694011L, 0x5739B3E5L, 0xA55230E6L, + 0xFB410CC2L, 0x092A8FC1L, 0x1A7A7C35L, 0xE811FF36L, + 0x3CDB9BDDL, 0xCEB018DEL, 0xDDE0EB2AL, 0x2F8B6829L, + 0x82F63B78L, 0x709DB87BL, 0x63CD4B8FL, 0x91A6C88CL, + 0x456CAC67L, 0xB7072F64L, 0xA457DC90L, 0x563C5F93L, + 0x082F63B7L, 0xFA44E0B4L, 0xE9141340L, 0x1B7F9043L, + 0xCFB5F4A8L, 0x3DDE77ABL, 0x2E8E845FL, 0xDCE5075CL, + 0x92A8FC17L, 0x60C37F14L, 0x73938CE0L, 0x81F80FE3L, + 0x55326B08L, 0xA759E80BL, 0xB4091BFFL, 0x466298FCL, + 0x1871A4D8L, 0xEA1A27DBL, 0xF94AD42FL, 0x0B21572CL, + 0xDFEB33C7L, 0x2D80B0C4L, 0x3ED04330L, 0xCCBBC033L, + 0xA24BB5A6L, 0x502036A5L, 0x4370C551L, 0xB11B4652L, + 0x65D122B9L, 0x97BAA1BAL, 0x84EA524EL, 0x7681D14DL, + 0x2892ED69L, 0xDAF96E6AL, 0xC9A99D9EL, 0x3BC21E9DL, + 0xEF087A76L, 0x1D63F975L, 0x0E330A81L, 0xFC588982L, + 0xB21572C9L, 0x407EF1CAL, 0x532E023EL, 0xA145813DL, + 0x758FE5D6L, 0x87E466D5L, 0x94B49521L, 0x66DF1622L, + 0x38CC2A06L, 0xCAA7A905L, 0xD9F75AF1L, 0x2B9CD9F2L, + 0xFF56BD19L, 0x0D3D3E1AL, 0x1E6DCDEEL, 0xEC064EEDL, + 0xC38D26C4L, 0x31E6A5C7L, 0x22B65633L, 0xD0DDD530L, + 0x0417B1DBL, 0xF67C32D8L, 0xE52CC12CL, 0x1747422FL, + 0x49547E0BL, 0xBB3FFD08L, 0xA86F0EFCL, 0x5A048DFFL, + 0x8ECEE914L, 0x7CA56A17L, 0x6FF599E3L, 0x9D9E1AE0L, + 0xD3D3E1ABL, 0x21B862A8L, 0x32E8915CL, 0xC083125FL, + 0x144976B4L, 0xE622F5B7L, 0xF5720643L, 0x07198540L, + 0x590AB964L, 0xAB613A67L, 0xB831C993L, 0x4A5A4A90L, + 0x9E902E7BL, 0x6CFBAD78L, 0x7FAB5E8CL, 0x8DC0DD8FL, + 0xE330A81AL, 0x115B2B19L, 0x020BD8EDL, 0xF0605BEEL, + 0x24AA3F05L, 0xD6C1BC06L, 0xC5914FF2L, 0x37FACCF1L, + 0x69E9F0D5L, 0x9B8273D6L, 0x88D28022L, 0x7AB90321L, + 0xAE7367CAL, 0x5C18E4C9L, 0x4F48173DL, 0xBD23943EL, + 0xF36E6F75L, 0x0105EC76L, 0x12551F82L, 0xE03E9C81L, + 0x34F4F86AL, 0xC69F7B69L, 0xD5CF889DL, 0x27A40B9EL, + 0x79B737BAL, 0x8BDCB4B9L, 0x988C474DL, 0x6AE7C44EL, + 0xBE2DA0A5L, 0x4C4623A6L, 0x5F16D052L, 0xAD7D5351L +}; + +/* + * Steps through buffer one byte at at time, calculates reflected + * crc using table. + */ + +uint32_t crc32c(uint32_t crc, const char *buf, unsigned int length) +{ + const uint8_t *data = (const uint8_t *)buf; + + while (length--) + crc = crc32c_table[(crc ^ *data++) & 0xFFL] ^ (crc >> 8); + + return crc; +} + +uint32_t crc32c_finalize(uint32_t crc) +{ + return ~GUINT32_TO_LE(crc); +} diff --git a/bluesky/log.c b/bluesky/log.c index c3d0c2b..ee96976 100644 --- a/bluesky/log.c +++ b/bluesky/log.c @@ -37,6 +37,21 @@ // no absolute guarantees on the size of a log segment. #define LOG_SEGMENT_SIZE (1 << 23) +#define HEADER_MAGIC 0x676f4c0a +#define FOOTER_MAGIC 0x2e435243 + +struct log_header { + uint32_t magic; // HEADER_MAGIC + uint64_t offset; // Starting byte offset of the log header + uint32_t keysize; // Size of the log key (bytes) + uint32_t size; // Size of the data item (bytes) +} __attribute__((packed)); + +struct log_footer { + uint32_t magic; // FOOTER_MAGIC + uint32_t crc; // Computed from log_header to log_footer.magic +} __attribute__((packed)); + static void writebuf(int fd, const char *buf, size_t len) { while (len > 0) { @@ -97,14 +112,40 @@ static gpointer log_thread(gpointer d) BlueSkyLogItem *item = (BlueSkyLogItem *)g_async_queue_pop(log->queue); g_mutex_lock(item->lock); + + off_t logsize = lseek(log->fd, 0, SEEK_CUR); + struct log_header header; + struct log_footer footer; + + header.magic = GUINT32_TO_LE(HEADER_MAGIC); + header.offset = GUINT64_TO_LE(logsize); + header.keysize = GUINT32_TO_LE(strlen(item->key)); + header.size = GUINT32_TO_LE(item->data->len); + footer.magic = GUINT32_TO_LE(FOOTER_MAGIC); + + uint32_t crc = BLUESKY_CRC32C_SEED; + + writebuf(log->fd, (const char *)&header, sizeof(header)); + crc = crc32c(crc, (const char *)&header, sizeof(header)); + writebuf(log->fd, item->key, strlen(item->key)); + crc = crc32c(crc, item->key, strlen(item->key)); + writebuf(log->fd, item->data->data, item->data->len); + crc = crc32c(crc, item->data->data, item->data->len); + + crc = crc32c(crc, (const char *)&footer, + sizeof(footer) - sizeof(uint32_t)); + footer.crc = crc32c_finalize(crc); + writebuf(log->fd, (const char *)&footer, sizeof(footer)); + committed = g_slist_prepend(committed, item); /* Force an fsync either if we will be closing this log segment and * opening a new file, or if there are no other log items currently * waiting to be written. */ - off_t logsize = lseek(log->fd, 0, SEEK_CUR); + logsize += strlen(item->key) + item->data->len + sizeof(header) + sizeof(footer); + if (logsize >= LOG_SEGMENT_SIZE || g_async_queue_length(log->queue) <= 0) { -- 2.20.1