LCOV - code coverage report
Current view: top level - discof/restore/utils - fd_ssparse.c (source / functions) Hit Total Coverage
Test: cov.lcov Lines: 0 506 0.0 %
Date: 2026-03-19 18:19:27 Functions: 0 19 0.0 %

          Line data    Source code
       1             : #include "fd_ssparse.h"
       2             : 
       3             : #include "../../../util/log/fd_log.h"
       4             : #include "../../../util/archive/fd_tar.h"
       5             : #include "../../../flamenco/runtime/fd_runtime_const.h"
       6             : #include "../../../flamenco/runtime/fd_system_ids.h"
       7             : 
       8             : #include <stdlib.h>
       9             : #include <errno.h>
      10             : 
      11           0 : #define FD_SSPARSE_STATE_TAR_HEADER             (0)
      12           0 : #define FD_SSPARSE_STATE_SCROLL_TAR_HEADER      (1)
      13           0 : #define FD_SSPARSE_STATE_VERSION                (2)
      14           0 : #define FD_SSPARSE_STATE_MANIFEST               (3)
      15           0 : #define FD_SSPARSE_STATE_ACCOUNT_HEADER         (4)
      16           0 : #define FD_SSPARSE_STATE_ACCOUNT_DATA           (5)
      17           0 : #define FD_SSPARSE_STATE_ACCOUNT_PADDING        (6)
      18           0 : #define FD_SSPARSE_STATE_STATUS_CACHE           (7)
      19           0 : #define FD_SSPARSE_STATE_SCROLL_ACCOUNT_GARBAGE (8)
      20             : #define FD_SSPARSE_STATE_ACCOUNT_BATCH          (9)
      21             : 
      22             : struct fd_ssparse_private {
      23             :   int state;
      24             :   uint batch_enabled : 1;
      25             : 
      26             :   struct {
      27             :     int seen_zero_tar_frame;
      28             :     int seen_manifest;
      29             :     int seen_status_cache;
      30             :     int seen_version;
      31             :   } flags;
      32             : 
      33             :   uchar version[ 5UL ];
      34             : 
      35             :   struct {
      36             :     acc_vec_map_t * acc_vec_map;
      37             :     acc_vec_t *     acc_vec_pool;
      38             :   } manifest;
      39             : 
      40             :   struct {
      41             :     uchar header[ 512UL ];
      42             :     ulong file_bytes;
      43             :     ulong file_bytes_consumed;
      44             :     ulong header_bytes_consumed;
      45             :   } tar;
      46             : 
      47             :   struct {
      48             :     uchar const * owner;
      49             :     uchar header[ 136UL ];
      50             :     ulong header_bytes_consumed;
      51             :     ulong data_bytes_consumed;
      52             :     ulong data_len;
      53             :   } account;
      54             : 
      55             :   ulong acc_vec_bytes;
      56             :   ulong slot;
      57             :   ulong bytes_consumed;
      58             : 
      59             :   ulong seed;
      60             :   ulong max_acc_vecs;
      61             :   ulong magic;
      62             : };
      63             : 
      64             : FD_FN_CONST ulong
      65           0 : fd_ssparse_align( void ) {
      66           0 :   return fd_ulong_max( alignof(fd_ssparse_t), fd_ulong_max( acc_vec_pool_align(), acc_vec_map_align() ) );
      67           0 : }
      68             : 
      69             : FD_FN_CONST ulong
      70           0 : fd_ssparse_footprint( ulong max_acc_vecs ) {
      71           0 :   ulong l = FD_LAYOUT_INIT;
      72           0 :   l = FD_LAYOUT_APPEND( l, fd_ssparse_align(),   sizeof(fd_ssparse_t)                   );
      73           0 :   l = FD_LAYOUT_APPEND( l, acc_vec_pool_align(), acc_vec_pool_footprint( max_acc_vecs ) );
      74           0 :   l = FD_LAYOUT_APPEND( l, acc_vec_map_align(),  acc_vec_map_footprint( max_acc_vecs )  );
      75           0 :   return FD_LAYOUT_FINI( l, fd_ssparse_align() );
      76           0 : }
      77             : 
      78             : void *
      79             : fd_ssparse_new( void *  shmem,
      80             :                 ulong   max_acc_vecs,
      81           0 :                 ulong   seed ) {
      82           0 :   if( FD_UNLIKELY( !shmem ) ) {
      83           0 :     FD_LOG_WARNING(( "NULL shmem" ));
      84           0 :     return NULL;
      85           0 :   }
      86             : 
      87           0 :   if( FD_UNLIKELY( !fd_ulong_is_aligned( (ulong)shmem, fd_ssparse_align() ) ) ) {
      88           0 :     FD_LOG_WARNING(( "unaligned shmem" ));
      89           0 :     return NULL;
      90           0 :   }
      91             : 
      92           0 :   FD_SCRATCH_ALLOC_INIT( l, shmem );
      93           0 :   fd_ssparse_t * ssparse       = FD_SCRATCH_ALLOC_APPEND( l, fd_ssparse_align(),   sizeof(fd_ssparse_t)                   );
      94           0 :   void *         _acc_vec_pool = FD_SCRATCH_ALLOC_APPEND( l, acc_vec_pool_align(), acc_vec_pool_footprint( max_acc_vecs ) );
      95           0 :   void *         _acc_vec_map  = FD_SCRATCH_ALLOC_APPEND( l, acc_vec_map_align(),  acc_vec_map_footprint( max_acc_vecs )  );
      96             : 
      97           0 :   ssparse->manifest.acc_vec_pool = acc_vec_pool_join( acc_vec_pool_new( _acc_vec_pool, max_acc_vecs ) );
      98           0 :   FD_TEST( ssparse->manifest.acc_vec_pool );
      99             : 
     100           0 :   ssparse->manifest.acc_vec_map = acc_vec_map_join( acc_vec_map_new( _acc_vec_map, max_acc_vecs, seed ) );
     101           0 :   FD_TEST( ssparse->manifest.acc_vec_map );
     102             : 
     103           0 :   ssparse->state = FD_SSPARSE_STATE_TAR_HEADER;
     104           0 :   fd_memset( &ssparse->flags, 0, sizeof(ssparse->flags) );
     105             : 
     106           0 :   ssparse->bytes_consumed = 0UL;
     107           0 :   ssparse->seed           = seed;
     108           0 :   ssparse->max_acc_vecs   = max_acc_vecs;
     109             : 
     110           0 :   ssparse->tar.header_bytes_consumed = 0UL;
     111           0 :   ssparse->tar.file_bytes_consumed   = 0UL;
     112           0 :   ssparse->tar.file_bytes            = 0UL;
     113             : 
     114           0 :   ssparse->account.owner                 = NULL;
     115           0 :   ssparse->account.header_bytes_consumed = 0UL;
     116           0 :   ssparse->account.data_bytes_consumed   = 0UL;
     117           0 :   ssparse->account.data_len              = 0UL;
     118           0 :   ssparse->acc_vec_bytes                 = 0UL;
     119             : 
     120           0 :   FD_COMPILER_MFENCE();
     121           0 :   ssparse->magic = FD_SSPARSE_MAGIC;
     122           0 :   FD_COMPILER_MFENCE();
     123             : 
     124           0 :   return (void *)ssparse;
     125           0 : }
     126             : 
     127             : fd_ssparse_t *
     128           0 : fd_ssparse_join( void * shssparse ) {
     129           0 :   if( FD_UNLIKELY( !shssparse ) ) {
     130           0 :     FD_LOG_WARNING(( "NULL shssparse" ));
     131           0 :     return NULL;
     132           0 :   }
     133             : 
     134           0 :   if( FD_UNLIKELY( !fd_ulong_is_aligned( (ulong)shssparse, fd_ssparse_align() ) ) ) {
     135           0 :     FD_LOG_WARNING(( "misaligned shssparse" ));
     136           0 :     return NULL;
     137           0 :   }
     138             : 
     139           0 :   fd_ssparse_t * ssparse = (fd_ssparse_t *)shssparse;
     140             : 
     141           0 :   if( FD_UNLIKELY( ssparse->magic!=FD_SSPARSE_MAGIC ) ) {
     142           0 :     FD_LOG_WARNING(( "bad magic" ));
     143           0 :     return NULL;
     144           0 :   }
     145             : 
     146           0 :   return ssparse;
     147           0 : }
     148             : 
     149             : void
     150           0 : fd_ssparse_reset( fd_ssparse_t * ssparse ) {
     151           0 :   ssparse->state = FD_SSPARSE_STATE_TAR_HEADER;
     152           0 :   fd_memset( &ssparse->flags, 0, sizeof(ssparse->flags) );
     153           0 :   ssparse->bytes_consumed                = 0UL;
     154             : 
     155           0 :   ssparse->tar.header_bytes_consumed     = 0UL;
     156           0 :   ssparse->tar.file_bytes_consumed       = 0UL;
     157           0 :   ssparse->tar.file_bytes                = 0UL;
     158             : 
     159           0 :   ssparse->account.owner                 = NULL;
     160           0 :   ssparse->account.header_bytes_consumed = 0UL;
     161           0 :   ssparse->account.data_bytes_consumed   = 0UL;
     162           0 :   ssparse->account.data_len              = 0UL;
     163           0 :   ssparse->acc_vec_bytes                 = 0UL;
     164             : 
     165           0 :   acc_vec_map_reset( ssparse->manifest.acc_vec_map );
     166           0 :   acc_vec_pool_reset( ssparse->manifest.acc_vec_pool );
     167           0 : }
     168             : 
     169             : static int
     170             : parse_tar_header_name( char const * name,
     171             :                        ulong *      id,
     172           0 :                        ulong *      slot ) {
     173           0 :   char name_buf[ FD_TAR_NAME_SZ ];
     174           0 :   fd_memcpy( name_buf, name, FD_TAR_NAME_SZ );
     175           0 :   name_buf[ FD_TAR_NAME_SZ-1 ] = '\0';
     176             : 
     177           0 :   char const * ptr = name_buf;
     178             : 
     179           0 :   if( FD_UNLIKELY( strncmp( ptr, "accounts/", 9UL ) ) ) {
     180           0 :     *id   = ULONG_MAX;
     181           0 :     *slot = ULONG_MAX;
     182           0 :     return -1;
     183           0 :   }
     184             : 
     185           0 :   ptr += 9UL;
     186           0 :   char const * next = strchr( ptr, '.' );
     187           0 :   if( FD_UNLIKELY( !next ) ) {
     188           0 :     *id   = ULONG_MAX;
     189           0 :     *slot = ULONG_MAX;
     190           0 :     return -1;
     191           0 :   }
     192           0 :   errno = 0;
     193           0 :   char * endptr;
     194           0 :   *slot = strtoul( ptr, &endptr, 10 );
     195           0 :   if( FD_UNLIKELY( errno==ERANGE || *endptr!='.' || endptr==ptr ) ) {
     196           0 :     *id   = ULONG_MAX;
     197           0 :     *slot = ULONG_MAX;
     198           0 :     return -1;
     199           0 :   }
     200             : 
     201           0 :   errno = 0;
     202           0 :   ptr = next + 1;
     203           0 :   *id = strtoul( ptr, &endptr, 10 );
     204           0 :   if( FD_UNLIKELY( errno==ERANGE || *endptr!='\0' || endptr==ptr ) ) {
     205           0 :     *id   = ULONG_MAX;
     206           0 :     *slot = ULONG_MAX;
     207           0 :     return -1;
     208           0 :   }
     209             : 
     210           0 :   return 0;
     211           0 : }
     212             : 
     213             : static int
     214             : advance_tar( fd_ssparse_t *                ssparse,
     215             :              uchar const *                 data,
     216             :              ulong                         data_sz,
     217           0 :              fd_ssparse_advance_result_t * result ) {
     218           0 :   ulong consume = fd_ulong_min( data_sz, 512UL - ssparse->tar.header_bytes_consumed );
     219           0 :   if( FD_UNLIKELY( !consume ) ) {
     220           0 :     FD_LOG_WARNING(( "unexpected end of data in tar header, data_sz=%lu, header_bytes_consumed=%lu", data_sz, ssparse->tar.header_bytes_consumed ));
     221           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     222           0 :   }
     223             : 
     224           0 :   fd_memcpy( ssparse->tar.header+ssparse->tar.header_bytes_consumed, data, consume );
     225           0 :   ssparse->bytes_consumed            += consume;
     226           0 :   result->bytes_consumed              = consume;
     227           0 :   ssparse->tar.header_bytes_consumed += consume;
     228             : 
     229           0 :   if( FD_UNLIKELY( ssparse->tar.header_bytes_consumed<512UL ) ) return FD_SSPARSE_ADVANCE_AGAIN;
     230             : 
     231           0 :   fd_tar_meta_t const * hdr = (fd_tar_meta_t const *)ssparse->tar.header;
     232           0 :   ssparse->tar.header_bytes_consumed = 0UL;
     233             : 
     234             :   /* "ustar\x00" and "ustar  \x00" (overlaps with version) are both
     235             :      valid values for magic.  These are POSIX ustar and OLDGNU versions
     236             :      respectively. */
     237           0 :   if( FD_UNLIKELY( memcmp( hdr->magic, FD_TAR_MAGIC, FD_TAR_MAGIC_SZ ) ) ) {
     238           0 :     int not_zero = 0;
     239           0 :     for( ulong i=0UL; i<512UL; i++ ) not_zero |= ssparse->tar.header[ i ];
     240           0 :     if( FD_UNLIKELY( not_zero ) ) {
     241           0 :       FD_LOG_WARNING(( "invalid tar header magic `%." FD_EXPAND_THEN_STRINGIFY(FD_TAR_MAGIC_SZ) "s`", hdr->magic ));
     242           0 :       return FD_SSPARSE_ADVANCE_ERROR;
     243           0 :     }
     244             : 
     245           0 :     if( FD_LIKELY( ssparse->flags.seen_zero_tar_frame ) ) {
     246           0 :       if( FD_UNLIKELY( !ssparse->flags.seen_version || !ssparse->flags.seen_manifest || !ssparse->flags.seen_status_cache ) ) {
     247           0 :         FD_LOG_WARNING(( "unexpected end of file before version or manifest or status cache" ));
     248           0 :         return FD_SSPARSE_ADVANCE_ERROR;
     249           0 :       }
     250             : 
     251           0 :       return FD_SSPARSE_ADVANCE_DONE;
     252           0 :     }
     253             : 
     254           0 :     ssparse->flags.seen_zero_tar_frame = 1;
     255           0 :     return FD_SSPARSE_ADVANCE_AGAIN;
     256           0 :   }
     257             : 
     258           0 :   if( FD_UNLIKELY( ssparse->flags.seen_zero_tar_frame ) ) {
     259           0 :     FD_LOG_WARNING(( "unexpected valid tar header after zero frame" ));
     260           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     261           0 :   }
     262             : 
     263           0 :   ssparse->tar.file_bytes = fd_tar_meta_get_size( hdr );
     264           0 :   if( FD_UNLIKELY( ssparse->tar.file_bytes==ULONG_MAX ) ) {
     265           0 :     FD_LOG_WARNING(( "invalid tar header size %." FD_EXPAND_THEN_STRINGIFY(FD_TAR_SIZE_SZ) "s "
     266           0 :                      "for tar header name %." FD_EXPAND_THEN_STRINGIFY(FD_TAR_NAME_SZ) "s", hdr->size, hdr->name ));
     267           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     268           0 :   }
     269             : 
     270           0 :   if( FD_UNLIKELY( hdr->typeflag==FD_TAR_TYPE_DIR ) ) return FD_SSPARSE_ADVANCE_AGAIN;
     271             : 
     272           0 :   if( FD_UNLIKELY( !fd_tar_meta_is_reg( hdr ) ) ) {
     273           0 :     FD_LOG_WARNING(( "invalid tar header type %d", hdr->typeflag ));
     274           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     275           0 :   }
     276           0 :   if( FD_UNLIKELY( !ssparse->tar.file_bytes ) ) {
     277           0 :     FD_LOG_WARNING(( "invalid tar header size %lu", ssparse->tar.file_bytes ));
     278           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     279           0 :   }
     280             : 
     281             :   /* TODO: Check every header field here for validity? */
     282             : 
     283           0 :   int desired_state;
     284           0 :   if( FD_LIKELY( !strncmp( hdr->name, "version", 7UL ) ) ) {
     285           0 :     desired_state = FD_SSPARSE_STATE_VERSION;
     286           0 :     if( FD_UNLIKELY( ssparse->tar.file_bytes!=5UL ) ) {
     287           0 :       FD_LOG_WARNING(( "invalid version file size %lu", ssparse->tar.file_bytes ));
     288           0 :       return FD_SSPARSE_ADVANCE_ERROR;
     289           0 :     }
     290           0 :   } else if( FD_LIKELY( !strncmp( hdr->name, "accounts/", 9UL ) ) ) {
     291           0 :     ssparse->account.header_bytes_consumed = 0UL;
     292           0 :     desired_state = FD_SSPARSE_STATE_ACCOUNT_HEADER;
     293           0 :     ulong id, slot;
     294           0 :     if( FD_UNLIKELY( -1==parse_tar_header_name( hdr->name, &id, &slot ) ) ) {
     295           0 :       FD_LOG_WARNING(( "invalid account append vec name %." FD_EXPAND_THEN_STRINGIFY(FD_TAR_NAME_SZ) "s", hdr->name ));
     296           0 :       return FD_SSPARSE_ADVANCE_ERROR;
     297           0 :     }
     298             : 
     299           0 :     acc_vec_key_t key = { .slot = slot, .id = id };
     300           0 :     acc_vec_t const * acc_vec = acc_vec_map_ele_query_const( ssparse->manifest.acc_vec_map, &key, NULL, ssparse->manifest.acc_vec_pool );
     301           0 :     if( FD_UNLIKELY( !acc_vec ) ) {
     302           0 :       FD_LOG_WARNING(( "append vec %lu.%lu not found in manifest", slot, id ));
     303           0 :       return FD_SSPARSE_ADVANCE_ERROR;
     304           0 :     }
     305             : 
     306           0 :     ssparse->acc_vec_bytes = acc_vec->file_sz;
     307           0 :     if( FD_UNLIKELY( ssparse->acc_vec_bytes>ssparse->tar.file_bytes ) ) {
     308           0 :       FD_LOG_WARNING(( "invalid append vec file size %lu > %lu", ssparse->acc_vec_bytes, ssparse->tar.file_bytes ));
     309           0 :       return FD_SSPARSE_ADVANCE_ERROR;
     310           0 :     }
     311             : 
     312           0 :     ssparse->slot = slot;
     313           0 :   } else if( FD_LIKELY( !strncmp( hdr->name, "snapshots/status_cache", 22UL ) ) ) desired_state = FD_SSPARSE_STATE_STATUS_CACHE;
     314           0 :   else if( FD_LIKELY( !strncmp( hdr->name, "snapshots/", 10UL ) ) ) {
     315           0 :     desired_state = FD_SSPARSE_STATE_MANIFEST;
     316           0 :   } else {
     317           0 :     FD_LOG_WARNING(( "unexpected tar header name `%." FD_EXPAND_THEN_STRINGIFY(FD_TAR_NAME_SZ) "s`", hdr->name ));
     318           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     319           0 :   }
     320             : 
     321           0 :   ssparse->tar.file_bytes_consumed = 0UL;
     322             : 
     323           0 :   switch( desired_state ) {
     324           0 :     case FD_SSPARSE_STATE_VERSION:
     325           0 :       if( FD_UNLIKELY( ssparse->flags.seen_version ) ) {
     326           0 :         FD_LOG_WARNING(( "unexpected duplicate version file" ));
     327           0 :         return FD_SSPARSE_ADVANCE_ERROR;
     328           0 :       }
     329             : 
     330           0 :       ssparse->flags.seen_version = 1;
     331           0 :       ssparse->state = FD_SSPARSE_STATE_VERSION;
     332           0 :       break;
     333           0 :     case FD_SSPARSE_STATE_MANIFEST:
     334           0 :       if( FD_UNLIKELY( ssparse->flags.seen_manifest ) ) {
     335           0 :         FD_LOG_WARNING(( "unexpected duplicate manifest file" ));
     336           0 :         return FD_SSPARSE_ADVANCE_ERROR;
     337           0 :       }
     338             : 
     339           0 :       ssparse->flags.seen_manifest = 1;
     340           0 :       ssparse->state = FD_SSPARSE_STATE_MANIFEST;
     341           0 :       break;
     342           0 :     case FD_SSPARSE_STATE_ACCOUNT_HEADER:
     343           0 :       if( FD_UNLIKELY( !ssparse->flags.seen_manifest ) ) {
     344           0 :         FD_LOG_WARNING(( "unexpected account append vec file before manifest" ));
     345           0 :         return FD_SSPARSE_ADVANCE_ERROR;
     346           0 :       }
     347             : 
     348           0 :       ssparse->account.header_bytes_consumed = 0UL;
     349           0 :       ssparse->state = FD_SSPARSE_STATE_ACCOUNT_HEADER;
     350           0 :       break;
     351           0 :     case FD_SSPARSE_STATE_STATUS_CACHE:
     352           0 :       if( FD_UNLIKELY( ssparse->flags.seen_status_cache ) ) {
     353           0 :         FD_LOG_WARNING(( "unexpected status cache file" ));
     354           0 :         return FD_SSPARSE_ADVANCE_ERROR;
     355           0 :       }
     356             : 
     357           0 :       ssparse->flags.seen_status_cache = 1;
     358           0 :       ssparse->state = FD_SSPARSE_STATE_STATUS_CACHE;
     359           0 :       break;
     360           0 :     default:
     361           0 :       FD_LOG_ERR(( "unexpected tar header desired state %d", desired_state ));
     362           0 :       break;
     363           0 :   }
     364             : 
     365           0 :   return FD_SSPARSE_ADVANCE_AGAIN;
     366           0 : }
     367             : 
     368             : static int
     369             : advance_version( fd_ssparse_t *                ssparse,
     370             :                  uchar const *                 data,
     371             :                  ulong                         data_sz,
     372           0 :                  fd_ssparse_advance_result_t * result ) {
     373           0 :   ulong consume = fd_ulong_min( data_sz, ssparse->tar.file_bytes-ssparse->tar.file_bytes_consumed );
     374           0 :   if( FD_UNLIKELY( !consume ) ) {
     375           0 :     FD_LOG_WARNING(( "unexpected end of data while parsing version file, data_sz=%lu, file_bytes_consumed=%lu, file_bytes=%lu", data_sz, ssparse->tar.file_bytes_consumed, ssparse->tar.file_bytes ));
     376           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     377           0 :   }
     378             : 
     379           0 :   fd_memcpy( ssparse->version+ssparse->tar.file_bytes_consumed, data, consume );
     380             : 
     381           0 :   ssparse->tar.file_bytes_consumed += consume;
     382           0 :   ssparse->bytes_consumed          += consume;
     383           0 :   result->bytes_consumed            = consume;
     384             : 
     385           0 :   if( FD_LIKELY( ssparse->tar.file_bytes_consumed<ssparse->tar.file_bytes ) ) return FD_SSPARSE_ADVANCE_AGAIN;
     386             : 
     387           0 :   FD_TEST( ssparse->tar.file_bytes_consumed==ssparse->tar.file_bytes );
     388           0 :   FD_TEST( ssparse->tar.file_bytes_consumed==5UL );
     389             : 
     390           0 :   if( FD_UNLIKELY( memcmp( ssparse->version, "1.2.0", 5UL ) ) ) {
     391           0 :     FD_LOG_WARNING(( "invalid version file %.*s", 5, ssparse->version ));
     392           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     393           0 :   }
     394             : 
     395           0 :   ssparse->state = FD_SSPARSE_STATE_SCROLL_TAR_HEADER;
     396           0 :   return FD_SSPARSE_ADVANCE_AGAIN;
     397           0 : }
     398             : 
     399             : static int
     400             : advance_status_cache( fd_ssparse_t *                 ssparse,
     401             :                        uchar const *                 data,
     402             :                        ulong                         data_sz,
     403           0 :                        fd_ssparse_advance_result_t * result ) {
     404           0 :   ulong consume = fd_ulong_min( data_sz, ssparse->tar.file_bytes-ssparse->tar.file_bytes_consumed );
     405           0 :   if( FD_UNLIKELY( !consume ) ) {
     406           0 :     FD_LOG_WARNING(( "unexpected end of data while parsing status cache, data_sz=%lu, file_bytes_consumed=%lu, file_bytes=%lu", data_sz, ssparse->tar.file_bytes_consumed, ssparse->tar.file_bytes ));
     407           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     408           0 :   }
     409             : 
     410           0 :   ssparse->tar.file_bytes_consumed += consume;
     411           0 :   ssparse->bytes_consumed          += consume;
     412             : 
     413           0 :   result->bytes_consumed            = consume;
     414           0 :   result->status_cache.data         = data;
     415           0 :   result->status_cache.data_sz      = consume;
     416             : 
     417           0 :   if( FD_LIKELY( ssparse->tar.file_bytes_consumed<ssparse->tar.file_bytes ) ) {
     418           0 :     return FD_SSPARSE_ADVANCE_STATUS_CACHE;
     419           0 :   }
     420           0 :   else { /* ssparse->tar.file_bytes_consumed==ssparse->tar.file_bytes */
     421             :     /* finished parsing status cache */
     422           0 :     ssparse->state = FD_SSPARSE_STATE_SCROLL_TAR_HEADER;
     423           0 :     return FD_SSPARSE_ADVANCE_STATUS_CACHE;
     424           0 :   }
     425           0 : }
     426             : 
     427             : static int
     428             : advance_manifest( fd_ssparse_t *                ssparse,
     429             :                   uchar const *                 data,
     430             :                   ulong                         data_sz,
     431           0 :                   fd_ssparse_advance_result_t * result ) {
     432           0 :   ulong consume = fd_ulong_min( data_sz, ssparse->tar.file_bytes-ssparse->tar.file_bytes_consumed );
     433           0 :   if( FD_UNLIKELY( !consume ) ) {
     434           0 :     FD_LOG_WARNING(( "unexpected end of data while parsing manifest, data_sz=%lu, file_bytes_consumed=%lu, file_bytes=%lu", data_sz, ssparse->tar.file_bytes_consumed, ssparse->tar.file_bytes ));
     435           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     436           0 :   }
     437             : 
     438           0 :   ssparse->tar.file_bytes_consumed += consume;
     439           0 :   ssparse->bytes_consumed          += consume;
     440             : 
     441           0 :   result->bytes_consumed           = consume;
     442           0 :   result->manifest.data            = data;
     443           0 :   result->manifest.data_sz         = consume;
     444           0 :   result->manifest.acc_vec_map     = ssparse->manifest.acc_vec_map;
     445           0 :   result->manifest.acc_vec_pool    = ssparse->manifest.acc_vec_pool;
     446             : 
     447           0 :   if( FD_LIKELY( ssparse->tar.file_bytes_consumed<ssparse->tar.file_bytes ) ) {
     448           0 :     return FD_SSPARSE_ADVANCE_MANIFEST;
     449           0 :   }
     450           0 :   else { /* ssparse->tar.file_bytes_consumed==ssparse->tar.file_bytes */
     451             :     /* finished parsing manifest */
     452           0 :     ssparse->state = FD_SSPARSE_STATE_SCROLL_TAR_HEADER;
     453           0 :     return FD_SSPARSE_ADVANCE_MANIFEST;
     454           0 :   }
     455           0 : }
     456             : 
     457             : static int
     458             : advance_next_tar( fd_ssparse_t *               ssparse,
     459             :                  uchar const *                 data,
     460             :                  ulong                         data_sz,
     461           0 :                  fd_ssparse_advance_result_t * result ) {
     462           0 :   (void)data;
     463             :   /* skip padding */
     464           0 :   ulong bytes_remaining    = fd_ulong_align_up( ssparse->bytes_consumed, 512UL ) - ssparse->bytes_consumed;
     465           0 :   ulong pad_sz             = bytes_remaining;
     466           0 :         pad_sz             = fd_ulong_min( pad_sz, data_sz );
     467           0 :   if( FD_UNLIKELY( !pad_sz && bytes_remaining ) ) {
     468           0 :     FD_LOG_WARNING(( "unexpected end of data while parsing tar header padding, data_sz=%lu, bytes_consumed=%lu, bytes_remaining=%lu", data_sz, ssparse->bytes_consumed, bytes_remaining ));
     469           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     470           0 :   }
     471             : 
     472           0 :   ssparse->bytes_consumed += pad_sz;
     473           0 :   result->bytes_consumed   = pad_sz;
     474           0 :   bytes_remaining         -= pad_sz;
     475             : 
     476           0 :   if( FD_LIKELY( !bytes_remaining ) ) ssparse->state = FD_SSPARSE_STATE_TAR_HEADER;
     477           0 :   return FD_SSPARSE_ADVANCE_AGAIN;
     478           0 : }
     479             : 
     480             : static int
     481             : advance_account_batch( fd_ssparse_t *                ssparse,
     482             :                        uchar const *                 data,
     483             :                        ulong                         data_sz,
     484           0 :                        fd_ssparse_advance_result_t * result ) {
     485             :   /* Cannot create a batch unless the parser is aligned to an account. */
     486           0 :   if( FD_UNLIKELY( ssparse->account.header_bytes_consumed ) ) return FD_SSPARSE_ADVANCE_AGAIN;
     487             : 
     488             :   /* Each account is at least 136 bytes large.  Don't attempt to create
     489             :      a batch unless at least 4 accounts fit. */
     490           0 :   ulong avail = fd_ulong_min( data_sz, ssparse->acc_vec_bytes - ssparse->tar.file_bytes_consumed );
     491           0 :   if( FD_UNLIKELY( avail<(4*136UL) ) ) return FD_SSPARSE_ADVANCE_AGAIN;
     492             : 
     493             :   /* Skip over accounts until we reached EOF or batch is full */
     494           0 :   result->account_batch.batch_cnt = 0;
     495           0 :   ulong off = 0UL;
     496           0 :   for( ulong idx=0UL; idx<FD_SSPARSE_ACC_BATCH_MAX && off+136UL<=avail; idx++ ) {
     497           0 :     uchar const * acc_hdr     = (uchar *)data+off;
     498             : 
     499             :     /* We want ConfigProgram accounts to go through the slow path,
     500             :        since they are published from there to consumers for monitoring. */
     501           0 :     if( FD_UNLIKELY( !memcmp( acc_hdr+64UL, fd_solana_config_program_id.key, sizeof(fd_hash_t) ) ) ) {
     502           0 :       if( FD_UNLIKELY( idx==0UL  ) ) return FD_SSPARSE_ADVANCE_AGAIN; /* At the front of the batch, abort */
     503           0 :       else                           break; /* otherwise, break early.  */
     504           0 :     }
     505             : 
     506           0 :     ulong         acc_data_sz = fd_ulong_load_8_fast( acc_hdr+8 );
     507           0 :     ulong         next_off    = off+136UL+acc_data_sz;
     508           0 :     ulong         pad_sz      = fd_ulong_align_up( ssparse->tar.file_bytes_consumed+next_off, 8UL ) -
     509           0 :                                                  ( ssparse->tar.file_bytes_consumed+next_off );
     510           0 :     next_off += pad_sz;
     511           0 :     if( FD_UNLIKELY( next_off>avail ) ) break; /* account is fragmented */
     512           0 :     if( FD_UNLIKELY( acc_data_sz > (24UL<<20) ) ) {
     513           0 :       FD_LOG_ERR(( "invalid account data size %lu", acc_data_sz ));
     514           0 :     }
     515           0 :     result->account_batch.batch_cnt        = idx+1UL;
     516           0 :     result->account_batch.batch[ idx ]     = acc_hdr;
     517           0 :     ssparse->account.header_bytes_consumed = 136UL;
     518           0 :     ssparse->account.data_bytes_consumed   = acc_data_sz;
     519           0 :     ssparse->account.data_len              = acc_data_sz;
     520           0 :     off = next_off;
     521           0 :   }
     522             : 
     523             :   /* Not worth batching if current chunk contains too few accounts. */
     524           0 :   if( FD_UNLIKELY( result->account_batch.batch_cnt!=FD_SSPARSE_ACC_BATCH_MAX ) ) {
     525           0 :     return FD_SSPARSE_ADVANCE_AGAIN;
     526           0 :   }
     527             : 
     528           0 :   ssparse->tar.file_bytes_consumed += off;
     529           0 :   ssparse->bytes_consumed          += off;
     530           0 :   result->bytes_consumed            = off;
     531             : 
     532             :   /* reset state */
     533             : 
     534           0 :   ssparse->state = FD_SSPARSE_STATE_ACCOUNT_PADDING;
     535             : 
     536           0 :   result->account_batch.slot = ssparse->slot;
     537             : 
     538           0 :   return FD_SSPARSE_ADVANCE_ACCOUNT_BATCH;
     539           0 : }
     540             : 
     541             : static int
     542             : advance_account_header( fd_ssparse_t *                ssparse,
     543             :                         uchar const *                 data,
     544             :                         ulong                         data_sz,
     545           0 :                         fd_ssparse_advance_result_t * result ) {
     546           0 :   ulong consume = fd_ulong_min( 136UL-ssparse->account.header_bytes_consumed, fd_ulong_min( data_sz, ssparse->acc_vec_bytes-ssparse->tar.file_bytes_consumed ) );
     547             : 
     548           0 :   if( FD_UNLIKELY( !consume ) ) {
     549           0 :     if( FD_LIKELY( ssparse->tar.file_bytes_consumed==ssparse->acc_vec_bytes ) ) {
     550           0 :       ssparse->state = FD_SSPARSE_STATE_SCROLL_ACCOUNT_GARBAGE;
     551           0 :       return FD_SSPARSE_ADVANCE_AGAIN;
     552           0 :     } else {
     553           0 :       FD_LOG_WARNING(( "unexpected end of data while advancing account header, data_sz=%lu, file_bytes_consumed=%lu, acc_vec_bytes=%lu", data_sz, ssparse->tar.file_bytes_consumed, ssparse->acc_vec_bytes ));
     554           0 :       return FD_SSPARSE_ADVANCE_ERROR;
     555           0 :     }
     556           0 :   }
     557             : 
     558           0 :   if( FD_UNLIKELY( consume<136UL ) ) {
     559           0 :     fd_memcpy( ssparse->account.header+ssparse->account.header_bytes_consumed, data, consume );
     560           0 :   } else if( ssparse->batch_enabled ) {
     561             :     /* fast path */
     562           0 :     int res = advance_account_batch( ssparse, data, data_sz, result );
     563           0 :     if( res==FD_SSPARSE_ADVANCE_ACCOUNT_BATCH ) return res;
     564             :     /* fall through and continue processing account header */
     565           0 :   }
     566             : 
     567           0 :   ssparse->account.header_bytes_consumed += consume;
     568           0 :   ssparse->tar.file_bytes_consumed       += consume;
     569           0 :   ssparse->bytes_consumed                += consume;
     570           0 :   result->bytes_consumed                  = consume;
     571             : 
     572           0 :   if( FD_UNLIKELY( ssparse->account.header_bytes_consumed<136UL ) ) return FD_SSPARSE_ADVANCE_AGAIN;
     573             : 
     574           0 :   uchar const * hdr = ssparse->account.header;
     575           0 :   if( FD_LIKELY( consume==136UL ) ) hdr = data;
     576             : 
     577           0 :   result->account_header.data_len = fd_ulong_load_8_fast( hdr+8UL );
     578           0 :   if( FD_UNLIKELY( result->account_header.data_len>FD_RUNTIME_ACC_SZ_MAX ) ) {
     579           0 :     FD_LOG_WARNING(( "invalid account header data length %lu", result->account_header.data_len ));
     580           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     581           0 :   }
     582             : 
     583           0 :   result->account_header.pubkey     = hdr+16UL;
     584           0 :   result->account_header.lamports   = fd_ulong_load_8_fast( hdr+48UL );
     585           0 :   result->account_header.rent_epoch = fd_ulong_load_8_fast( hdr+56UL );
     586           0 :   result->account_header.owner      = hdr+64UL;
     587           0 :   result->account_header.executable = hdr[ 96UL ];
     588           0 :   if( FD_UNLIKELY( result->account_header.executable>1 ) ) {
     589           0 :     char pubkey_str[ FD_BASE58_ENCODED_32_SZ ];
     590           0 :     fd_base58_encode_32( result->account_header.pubkey, NULL, pubkey_str );
     591           0 :     FD_LOG_WARNING(( "invalid account header executable %d for account %s", result->account_header.executable, pubkey_str ));
     592           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     593           0 :   }
     594           0 :   result->account_header.hash       = hdr+104UL;
     595           0 :   result->account_header.slot       = ssparse->slot;
     596             : 
     597           0 :   ssparse->account.owner               = hdr+64UL;
     598           0 :   ssparse->account.data_len            = result->account_header.data_len;
     599           0 :   ssparse->account.data_bytes_consumed = 0UL;
     600           0 :   ssparse->state = FD_SSPARSE_STATE_ACCOUNT_DATA;
     601             : 
     602           0 :   return FD_SSPARSE_ADVANCE_ACCOUNT_HEADER;
     603           0 : }
     604             : 
     605             : static int
     606             : advance_account_data( fd_ssparse_t *                ssparse,
     607             :                       uchar const *                 data,
     608             :                       ulong                         data_sz,
     609           0 :                       fd_ssparse_advance_result_t * result ) {
     610           0 :   if( FD_UNLIKELY( ssparse->account.data_bytes_consumed==ssparse->account.data_len ) ) {
     611           0 :     ssparse->state = FD_SSPARSE_STATE_ACCOUNT_PADDING;
     612           0 :     return FD_SSPARSE_ADVANCE_AGAIN;
     613           0 :   }
     614             : 
     615           0 :   ulong consume = fd_ulong_min( data_sz, ssparse->acc_vec_bytes-ssparse->tar.file_bytes_consumed );
     616           0 :   if( FD_UNLIKELY( !consume ) ) {
     617           0 :     FD_LOG_WARNING(( "account data extends beyond append vec size" ));
     618           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     619           0 :   }
     620             : 
     621           0 :   consume = fd_ulong_min( consume, ssparse->account.data_len-ssparse->account.data_bytes_consumed );
     622           0 :   if( FD_UNLIKELY( !consume ) ) {
     623           0 :     FD_LOG_WARNING(( "unexpected end of data while parsing account data, data_sz=%lu, data_bytes_consumed=%lu, data_len=%lu", data_sz, ssparse->account.data_bytes_consumed, ssparse->account.data_len ));
     624           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     625           0 :   }
     626             : 
     627           0 :   ssparse->tar.file_bytes_consumed     += consume;
     628           0 :   ssparse->bytes_consumed              += consume;
     629           0 :   ssparse->account.data_bytes_consumed += consume;
     630           0 :   result->bytes_consumed                = consume;
     631             : 
     632           0 :   result->account_data.owner    = ssparse->account.owner;
     633           0 :   result->account_data.data_sz  = consume;
     634           0 :   result->account_data.data     = data;
     635             : 
     636           0 :   FD_TEST( ssparse->account.data_bytes_consumed<=ssparse->account.data_len );
     637           0 :   if( FD_LIKELY( ssparse->account.data_bytes_consumed==ssparse->account.data_len ) ) {
     638           0 :     ssparse->state = FD_SSPARSE_STATE_ACCOUNT_PADDING;
     639           0 :   }
     640             : 
     641           0 :   return FD_SSPARSE_ADVANCE_ACCOUNT_DATA;
     642           0 : }
     643             : 
     644             : static int
     645             : advance_account_padding( fd_ssparse_t *                ssparse,
     646             :                          uchar const *                 data,
     647             :                          ulong                         data_sz,
     648           0 :                          fd_ssparse_advance_result_t * result ) {
     649           0 :   (void)data;
     650             : 
     651           0 :   ulong pad_sz = fd_ulong_align_up( ssparse->tar.file_bytes_consumed, 8UL ) - ssparse->tar.file_bytes_consumed;
     652           0 :         pad_sz = fd_ulong_min( pad_sz, ssparse->acc_vec_bytes - ssparse->tar.file_bytes_consumed );
     653           0 :   if( FD_UNLIKELY( !pad_sz ) ) {
     654           0 :     if( FD_LIKELY( ssparse->tar.file_bytes_consumed==ssparse->acc_vec_bytes ) ) ssparse->state = FD_SSPARSE_STATE_SCROLL_TAR_HEADER;
     655           0 :     else                                                                        ssparse->state = FD_SSPARSE_STATE_ACCOUNT_HEADER;
     656             : 
     657           0 :     ssparse->account.header_bytes_consumed = 0UL;
     658           0 :     return FD_SSPARSE_ADVANCE_AGAIN;
     659           0 :   }
     660             : 
     661           0 :   ulong consume = fd_ulong_min( data_sz, pad_sz );
     662           0 :   if( FD_UNLIKELY( !consume ) ) {
     663           0 :     FD_LOG_WARNING(( "unexpected end of data while parsing account padding, data_sz=%lu, file_bytes_consumed=%lu, acc_vec_bytes=%lu", data_sz, ssparse->tar.file_bytes_consumed, ssparse->acc_vec_bytes ));
     664           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     665           0 :   }
     666             : 
     667           0 :   ssparse->tar.file_bytes_consumed += consume;
     668           0 :   ssparse->bytes_consumed          += consume;
     669           0 :   result->bytes_consumed            = consume;
     670             : 
     671           0 :   ulong remaining = fd_ulong_align_up( ssparse->tar.file_bytes_consumed, 8UL ) - ssparse->tar.file_bytes_consumed;
     672           0 :   if( FD_LIKELY( !remaining ) ) {
     673           0 :     ssparse->account.header_bytes_consumed = 0UL;
     674           0 :     ssparse->state = FD_SSPARSE_STATE_ACCOUNT_HEADER;
     675           0 :   }
     676           0 :   return FD_SSPARSE_ADVANCE_AGAIN;
     677           0 : }
     678             : 
     679             : static int
     680             : advance_account_garbage( fd_ssparse_t *                ssparse,
     681             :                          uchar const *                 data,
     682             :                          ulong                         data_sz,
     683           0 :                          fd_ssparse_advance_result_t * result ) {
     684           0 :   (void)data;
     685           0 :   ulong rem = ssparse->tar.file_bytes-ssparse->tar.file_bytes_consumed;
     686           0 :   if( FD_UNLIKELY( !rem ) ) {
     687           0 :     ssparse->state = FD_SSPARSE_STATE_SCROLL_TAR_HEADER;
     688           0 :     return FD_SSPARSE_ADVANCE_AGAIN;
     689           0 :   }
     690             : 
     691           0 :   if( FD_UNLIKELY( !data_sz ) ) {
     692           0 :     FD_LOG_WARNING(( "unexpected end of data while parsing append vec garbage, data_sz=%lu, remaining_bytes=%lu", data_sz, rem ));
     693           0 :     return FD_SSPARSE_ADVANCE_ERROR;
     694           0 :   }
     695             : 
     696           0 :   ulong consume = fd_ulong_min( data_sz, rem );
     697           0 :   ssparse->tar.file_bytes_consumed += consume;
     698           0 :   ssparse->bytes_consumed          += consume;
     699           0 :   result->bytes_consumed            = consume;
     700             : 
     701           0 :   if( FD_LIKELY( ssparse->tar.file_bytes_consumed<ssparse->tar.file_bytes ) ) return FD_SSPARSE_ADVANCE_AGAIN;
     702             : 
     703           0 :   ssparse->state = FD_SSPARSE_STATE_SCROLL_TAR_HEADER;
     704           0 :   return FD_SSPARSE_ADVANCE_AGAIN;
     705           0 : }
     706             : 
     707             : int
     708             : fd_ssparse_advance( fd_ssparse_t *                ssparse,
     709             :                     uchar const *                 data,
     710             :                     ulong                         data_sz,
     711           0 :                     fd_ssparse_advance_result_t * result ) {
     712           0 :   result->bytes_consumed = 0UL;
     713             : 
     714           0 :   switch( ssparse->state ) {
     715           0 :     case FD_SSPARSE_STATE_TAR_HEADER:             return advance_tar( ssparse, data, data_sz, result );
     716           0 :     case FD_SSPARSE_STATE_SCROLL_TAR_HEADER:      return advance_next_tar( ssparse, data, data_sz, result );
     717           0 :     case FD_SSPARSE_STATE_VERSION:                return advance_version( ssparse, data, data_sz, result );
     718           0 :     case FD_SSPARSE_STATE_MANIFEST:               return advance_manifest( ssparse, data, data_sz, result );
     719           0 :     case FD_SSPARSE_STATE_ACCOUNT_HEADER:         return advance_account_header( ssparse, data, data_sz, result );
     720           0 :     case FD_SSPARSE_STATE_ACCOUNT_DATA:           return advance_account_data( ssparse, data, data_sz, result );
     721           0 :     case FD_SSPARSE_STATE_ACCOUNT_PADDING:        return advance_account_padding( ssparse, data, data_sz, result );
     722           0 :     case FD_SSPARSE_STATE_STATUS_CACHE:           return advance_status_cache( ssparse, data, data_sz, result );
     723           0 :     case FD_SSPARSE_STATE_SCROLL_ACCOUNT_GARBAGE: return advance_account_garbage( ssparse, data, data_sz, result );
     724           0 :     default: FD_LOG_ERR(( "invalid state %d", ssparse->state ));
     725           0 :   }
     726           0 : }
     727             : 
     728             : void
     729             : fd_ssparse_batch_enable( fd_ssparse_t * ssparse,
     730           0 :                          int            enabled ) {
     731           0 :   ssparse->batch_enabled = !!enabled;
     732           0 : }
     733             : 
     734             : int
     735             : fd_ssparse_populate_acc_vec_map( fd_ssparse_t * ssparse,
     736             :                                  ulong *        slots,
     737             :                                  ulong *        ids,
     738             :                                  ulong *        file_szs,
     739           0 :                                  ulong          cnt ) {
     740           0 :   for( ulong i=0UL; i<cnt; i++ ) {
     741           0 :     acc_vec_key_t key = { .slot=slots[ i ], .id=ids[ i ] };
     742           0 :     if( FD_UNLIKELY( acc_vec_map_ele_query( ssparse->manifest.acc_vec_map, &key, NULL, ssparse->manifest.acc_vec_pool ) ) ) return -1;
     743           0 :     acc_vec_t * acc_vec = acc_vec_pool_ele_acquire( ssparse->manifest.acc_vec_pool );
     744           0 :     acc_vec->key.id   = ids[ i ];
     745           0 :     acc_vec->key.slot = slots[ i ];
     746           0 :     acc_vec->file_sz  = file_szs[ i ];
     747           0 :     acc_vec_map_ele_insert( ssparse->manifest.acc_vec_map, acc_vec, ssparse->manifest.acc_vec_pool );
     748           0 :   }
     749           0 :   return 0;
     750           0 : }

Generated by: LCOV version 1.14