Line data    Source code

       1             : #define _GNU_SOURCE
       2             : #include "fd_sshttp.h"
       3             : #include "fd_ssarchive.h"
       4             : 
       5             : #include "../../../waltz/http/picohttpparser.h"
       6             : #include "../../../util/log/fd_log.h"
       7             : #include "../../../flamenco/types/fd_types_custom.h"
       8             : 
       9             : #include <unistd.h>
      10             : #include <errno.h>
      11             : #include <stdlib.h>
      12             : 
      13             : #include <sys/socket.h>
      14             : #include <netinet/tcp.h>
      15             : #include <netinet/in.h>
      16             : 
      17           0 : #define FD_SSHTTP_STATE_INIT  (0) /* start */
      18           0 : #define FD_SSHTTP_STATE_REQ   (1) /* sending request */
      19           0 : #define FD_SSHTTP_STATE_RESP  (2) /* receiving response headers */
      20           0 : #define FD_SSHTTP_STATE_DL    (3) /* downloading response body */
      21             : 
      22             : struct fd_sshttp_private {
      23             :   int  state;
      24             :   long deadline;
      25             :   int  full;
      26             : 
      27             :   int   hops;
      28             : 
      29             :   fd_ip4_port_t addr;
      30             :   int           sockfd;
      31             : 
      32             :   char  request[ 4096UL ];
      33             :   ulong request_len;
      34             :   ulong request_sent;
      35             : 
      36             :   ulong response_len;
      37             :   char  response[ USHORT_MAX ];
      38             : 
      39             :   char full_snapshot_name[ PATH_MAX ];
      40             :   char incremental_snapshot_name[ PATH_MAX ];
      41             : 
      42             :   ulong content_len;
      43             :   ulong content_read;
      44             : 
      45             :   ulong magic;
      46             : };
      47             : 
      48             : FD_FN_CONST ulong
      49           0 : fd_sshttp_align( void ) {
      50           0 :   return FD_SSHTTP_ALIGN;
      51           0 : }
      52             : 
      53             : FD_FN_CONST ulong
      54           0 : fd_sshttp_footprint( void ) {
      55           0 :   ulong l;
      56           0 :   l = FD_LAYOUT_INIT;
      57           0 :   l = FD_LAYOUT_APPEND( l, FD_SSHTTP_ALIGN,       sizeof(fd_sshttp_t) );
      58           0 :   return FD_LAYOUT_FINI( l, FD_SSHTTP_ALIGN );
      59           0 : }
      60             : 
      61             : void *
      62           0 : fd_sshttp_new( void * shmem ) {
      63           0 :   if( FD_UNLIKELY( !shmem ) ) {
      64           0 :     FD_LOG_WARNING(( "NULL shmem" ));
      65           0 :     return NULL;
      66           0 :   }
      67             : 
      68           0 :   if( FD_UNLIKELY( !fd_ulong_is_aligned( (ulong)shmem, fd_sshttp_align() ) ) ) {
      69           0 :     FD_LOG_WARNING(( "unaligned shmem" ));
      70           0 :     return NULL;
      71           0 :   }
      72             : 
      73           0 :   FD_SCRATCH_ALLOC_INIT( l, shmem );
      74           0 :   fd_sshttp_t * sshttp = FD_SCRATCH_ALLOC_APPEND( l, FD_SSHTTP_ALIGN, sizeof(fd_sshttp_t) );
      75             : 
      76           0 :   sshttp->state = FD_SSHTTP_STATE_INIT;
      77             : 
      78           0 :   FD_COMPILER_MFENCE();
      79           0 :   FD_VOLATILE( sshttp->magic ) = FD_SSHTTP_MAGIC;
      80           0 :   FD_COMPILER_MFENCE();
      81             : 
      82           0 :   return (void *)sshttp;
      83           0 : }
      84             : 
      85             : fd_sshttp_t *
      86           0 : fd_sshttp_join( void * shhttp ) {
      87           0 :   if( FD_UNLIKELY( !shhttp ) ) {
      88           0 :     FD_LOG_WARNING(( "NULL shhttp" ));
      89           0 :     return NULL;
      90           0 :   }
      91             : 
      92           0 :   if( FD_UNLIKELY( !fd_ulong_is_aligned( (ulong)shhttp, fd_sshttp_align() ) ) ) {
      93           0 :     FD_LOG_WARNING(( "misaligned shhttp" ));
      94           0 :     return NULL;
      95           0 :   }
      96             : 
      97           0 :   fd_sshttp_t * sshttp = (fd_sshttp_t *)shhttp;
      98             : 
      99           0 :   if( FD_UNLIKELY( sshttp->magic!=FD_SSHTTP_MAGIC ) ) {
     100           0 :     FD_LOG_WARNING(( "bad magic" ));
     101           0 :     return NULL;
     102           0 :   }
     103             : 
     104           0 :   return sshttp;
     105           0 : }
     106             : 
     107             : void
     108             : fd_sshttp_init( fd_sshttp_t * http,
     109             :                 fd_ip4_port_t addr,
     110             :                 char const *  path,
     111             :                 ulong         path_len,
     112           0 :                 long          now ) {
     113           0 :   FD_TEST( http->state==FD_SSHTTP_STATE_INIT );
     114             : 
     115           0 :   http->hops = 4UL;
     116             : 
     117           0 :   http->request_sent = 0UL;
     118           0 :   FD_TEST( fd_cstr_printf_check( http->request, sizeof(http->request), &http->request_len,
     119           0 :     "GET %.*s HTTP/1.1\r\n"
     120           0 :     "User-Agent: Firedancer\r\n"
     121           0 :     "Accept: */*\r\n"
     122           0 :     "Accept-Encoding: identity\r\n"
     123           0 :     "Host: " FD_IP4_ADDR_FMT "\r\n\r\n",
     124           0 :     (int)path_len, path, FD_IP4_ADDR_FMT_ARGS( addr.addr ) ) );
     125             : 
     126           0 :   http->addr = addr;
     127           0 :   http->sockfd = socket( AF_INET, SOCK_STREAM|SOCK_NONBLOCK, 0 );
     128           0 :   if( FD_UNLIKELY( -1==http->sockfd ) ) FD_LOG_ERR(( "socket() failed (%d-%s)", errno, fd_io_strerror( errno ) ));
     129             : 
     130           0 :   int optval = 1;
     131           0 :   if( FD_UNLIKELY( -1==setsockopt( http->sockfd, SOL_TCP, TCP_NODELAY, &optval, sizeof(int) ) ) ) {
     132           0 :     FD_LOG_ERR(( "setsockopt() failed (%d-%s)", errno, fd_io_strerror( errno ) ));
     133           0 :   }
     134             : 
     135           0 :   struct sockaddr_in addr_in = {
     136           0 :     .sin_family = AF_INET,
     137           0 :     .sin_port   = addr.port,
     138           0 :     .sin_addr   = { .s_addr = addr.addr }
     139           0 :   };
     140             : 
     141           0 :   if( FD_UNLIKELY( -1==connect( http->sockfd, fd_type_pun_const( &addr_in ), sizeof(addr_in) ) ) ) {
     142           0 :     if( FD_UNLIKELY( errno!=EINPROGRESS ) ) {
     143           0 :       if( FD_UNLIKELY( -1==close( http->sockfd ) ) ) FD_LOG_ERR(( "close() failed (%d-%s)", errno, fd_io_strerror( errno ) ));
     144           0 :       return;
     145           0 :     }
     146           0 :   }
     147             : 
     148           0 :   http->state    = FD_SSHTTP_STATE_REQ;
     149           0 :   http->deadline = now + 500L*1000L*1000L;
     150           0 : }
     151             : 
     152             : void
     153           0 : fd_sshttp_cancel( fd_sshttp_t * http ) {
     154           0 :   if( FD_LIKELY( http->state!=FD_SSHTTP_STATE_INIT && -1!=http->sockfd ) ) {
     155           0 :     if( FD_UNLIKELY( -1==close( http->sockfd ) ) ) FD_LOG_ERR(( "close() failed (%d-%s)", errno, fd_io_strerror( errno ) ));
     156           0 :     http->sockfd = -1;
     157           0 :   }
     158           0 :   http->state = FD_SSHTTP_STATE_INIT;
     159           0 : }
     160             : 
     161             : static int
     162             : send_request( fd_sshttp_t * http,
     163           0 :               long          now ) {
     164           0 :   if( FD_UNLIKELY( now>http->deadline ) ) {
     165           0 :     fd_sshttp_cancel( http );
     166           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     167           0 :   }
     168             : 
     169           0 :   long sent = send( http->sockfd, http->request+http->request_sent, http->request_len-http->request_sent, 0 );
     170           0 :   if( FD_UNLIKELY( -1==sent && errno==EAGAIN ) ) return FD_SSHTTP_ADVANCE_AGAIN;
     171           0 :   else if( FD_UNLIKELY( -1==sent ) ) {
     172           0 :     fd_sshttp_cancel( http );
     173           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     174           0 :   }
     175             : 
     176           0 :   http->request_sent += (ulong)sent;
     177           0 :   if( FD_UNLIKELY( http->request_sent==http->request_len ) ) {
     178           0 :     http->state = FD_SSHTTP_STATE_RESP;
     179           0 :     http->response_len = 0UL;
     180           0 :     http->deadline = now + 500L*1000L*1000L;
     181           0 :   }
     182             : 
     183           0 :   return FD_SSHTTP_ADVANCE_AGAIN;
     184           0 : }
     185             : 
     186             : static int
     187             : follow_redirect( fd_sshttp_t *        http,
     188             :                   struct phr_header * headers,
     189             :                   ulong               header_cnt,
     190           0 :                   long                now ) {
     191           0 :   if( FD_UNLIKELY( !http->hops ) ) {
     192           0 :     FD_LOG_WARNING(( "too many redirects" ));
     193           0 :     fd_sshttp_cancel( http );
     194           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     195           0 :   }
     196             : 
     197           0 :   http->hops--;
     198             : 
     199           0 :   ulong        location_len;
     200           0 :   char const * location = NULL;
     201             : 
     202           0 :   for( ulong i=0UL; i<header_cnt; i++ ) {
     203           0 :     if( FD_UNLIKELY( !strncasecmp( headers[ i ].name, "location", headers[ i ].name_len ) ) ) {
     204           0 :       if( FD_UNLIKELY( !headers [ i ].value_len || headers[ i ].value[ 0 ]!='/' ) ) {
     205           0 :         FD_LOG_WARNING(( "invalid location header `%.*s`", (int)headers[ i ].value_len, headers[ i ].value ));
     206           0 :         fd_sshttp_cancel( http );
     207           0 :         return FD_SSHTTP_ADVANCE_ERROR;
     208           0 :       }
     209             : 
     210           0 :       location_len = headers[ i ].value_len;
     211           0 :       location     = headers[ i ].value;
     212             : 
     213           0 :       if( FD_UNLIKELY( location_len>=PATH_MAX-1UL ) ) {
     214           0 :         fd_sshttp_cancel( http );
     215           0 :         return FD_SSHTTP_ADVANCE_ERROR;
     216           0 :       }
     217             : 
     218           0 :       char snapshot_name[ PATH_MAX ];
     219           0 :       fd_memcpy( snapshot_name, location+1UL, location_len-1UL );
     220           0 :       snapshot_name[ location_len-1UL ] = '\0';
     221             : 
     222           0 :       ulong full_entry_slot, incremental_entry_slot;
     223           0 :       uchar decoded_hash[ FD_HASH_FOOTPRINT ];
     224           0 :       int err = fd_ssarchive_parse_filename( snapshot_name, &full_entry_slot, &incremental_entry_slot, decoded_hash );
     225             : 
     226           0 :       if( FD_UNLIKELY( err ) ) {
     227           0 :         FD_LOG_WARNING(( "unrecognized snapshot file `%s` in redirect location header", snapshot_name ));
     228           0 :         fd_sshttp_cancel( http );
     229           0 :         return FD_SSHTTP_ADVANCE_ERROR;
     230           0 :       }
     231             : 
     232           0 :       char encoded_hash[ FD_BASE58_ENCODED_32_SZ ];
     233           0 :       fd_base58_encode_32( decoded_hash, NULL, encoded_hash );
     234             : 
     235           0 :       if( FD_LIKELY( incremental_entry_slot!=ULONG_MAX ) ) {
     236           0 :         FD_TEST( fd_cstr_printf_check( http->incremental_snapshot_name, PATH_MAX, NULL, "incremental-snapshot-%lu-%lu-%s.tar.zst", full_entry_slot, incremental_entry_slot, encoded_hash ) );
     237           0 :       } else {
     238           0 :         FD_TEST( fd_cstr_printf_check( http->full_snapshot_name, PATH_MAX, NULL, "snapshot-%lu-%s.tar.zst", full_entry_slot, encoded_hash ) );
     239           0 :       }
     240           0 :       break;
     241           0 :     }
     242           0 :   }
     243             : 
     244           0 :   if( FD_UNLIKELY( !location ) ) {
     245           0 :     FD_LOG_WARNING(( "no location header in redirect response" ));
     246           0 :     fd_sshttp_cancel( http );
     247           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     248           0 :   }
     249             : 
     250           0 :   if( FD_UNLIKELY( !fd_cstr_printf_check( http->request, sizeof(http->request), &http->request_len,
     251           0 :     "GET %.*s HTTP/1.1\r\n"
     252           0 :     "User-Agent: Firedancer\r\n"
     253           0 :     "Accept: */*\r\n"
     254           0 :     "Accept-Encoding: identity\r\n"
     255           0 :     "Host: " FD_IP4_ADDR_FMT "\r\n\r\n",
     256           0 :     (int)location_len, location, FD_IP4_ADDR_FMT_ARGS( http->addr.addr ) ) ) ) {
     257           0 :     FD_LOG_WARNING(( "location header too long `%.*s`", (int)location_len, location ));
     258           0 :     fd_sshttp_cancel( http );
     259           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     260           0 :   }
     261             : 
     262           0 :   FD_LOG_NOTICE(( "following redirect to http://" FD_IP4_ADDR_FMT ":%hu%.*s",
     263           0 :                   FD_IP4_ADDR_FMT_ARGS( http->addr.addr ), fd_ushort_bswap( http->addr.port ),
     264           0 :                   (int)headers[ 0 ].value_len, headers[ 0 ].value ));
     265             : 
     266           0 :   fd_sshttp_cancel( http );
     267           0 :   fd_sshttp_init( http, http->addr, location, location_len, now );
     268             : 
     269           0 :   return FD_SSHTTP_ADVANCE_AGAIN;
     270           0 : }
     271             : 
     272             : static int
     273             : read_response( fd_sshttp_t * http,
     274             :                ulong *       data_len,
     275             :                uchar *       data,
     276           0 :                long          now ) {
     277           0 :   if( FD_UNLIKELY( now>http->deadline ) ) {
     278           0 :     FD_LOG_WARNING(( "timeout reading response" ));
     279           0 :     fd_sshttp_cancel( http );
     280           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     281           0 :   }
     282             : 
     283           0 :   long read = recv( http->sockfd, http->response+http->response_len, sizeof(http->response)-http->response_len, 0 );
     284           0 :   if( FD_UNLIKELY( -1==read && errno==EAGAIN ) ) return 0;
     285           0 :   else if( FD_UNLIKELY( -1==read ) ) {
     286           0 :     FD_LOG_WARNING(( "recv() failed (%d-%s)", errno, fd_io_strerror( errno ) ));
     287           0 :     fd_sshttp_cancel( http );
     288           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     289           0 :   }
     290             : 
     291           0 :   http->response_len += (ulong)read;
     292             : 
     293           0 :   int               minor_version;
     294           0 :   int               status;
     295           0 :   const char *      message;
     296           0 :   ulong             message_len;
     297           0 :   struct phr_header headers[ 128UL ];
     298           0 :   ulong             header_cnt = 128UL;
     299           0 :   int parsed = phr_parse_response( http->response,
     300           0 :                                     http->response_len,
     301           0 :                                     &minor_version,
     302           0 :                                     &status,
     303           0 :                                     &message,
     304           0 :                                     &message_len,
     305           0 :                                     headers,
     306           0 :                                     &header_cnt,
     307           0 :                                     http->response_len - (ulong)read );
     308           0 :   if( FD_UNLIKELY( parsed==-1 ) ) {
     309           0 :     FD_LOG_WARNING(( "malformed response body" ));
     310           0 :     fd_sshttp_cancel( http );
     311           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     312           0 :   } else if( parsed==-2 ) {
     313           0 :     return FD_SSHTTP_ADVANCE_AGAIN;
     314           0 :   }
     315             : 
     316           0 :   int is_redirect = (status==301) | (status==303) | (status==304) | (status==307) | (status==308);
     317           0 :   if( FD_UNLIKELY( is_redirect ) ) {
     318           0 :     return follow_redirect( http, headers, header_cnt, now );
     319           0 :   }
     320             : 
     321           0 :   if( FD_UNLIKELY( status!=200 ) ) {
     322           0 :     FD_LOG_WARNING(( "unexpected response status %d", status ));
     323           0 :     fd_sshttp_cancel( http );
     324           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     325           0 :   }
     326             : 
     327           0 :   http->content_read = 0UL;
     328           0 :   http->content_len = ULONG_MAX;
     329           0 :   for( ulong i=0UL; i<header_cnt; i++ ) {
     330           0 :     if( FD_LIKELY( headers[i].name_len!=14UL ) ) continue;
     331           0 :     if( FD_LIKELY( strncasecmp( headers[i].name, "content-length", 14UL ) ) ) continue;
     332             : 
     333           0 :     http->content_len = strtoul( headers[i].value, NULL, 10 );
     334           0 :     break;
     335           0 :   }
     336             : 
     337           0 :   if( FD_UNLIKELY( http->content_len==ULONG_MAX ) ) {
     338           0 :     FD_LOG_WARNING(( "no content-length header in response" ));
     339           0 :     fd_sshttp_cancel( http );
     340           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     341           0 :   }
     342             : 
     343           0 :   http->state = FD_SSHTTP_STATE_DL;
     344           0 :   if( FD_UNLIKELY( (ulong)parsed<http->response_len ) ) {
     345           0 :     if( FD_UNLIKELY( *data_len<http->response_len-(ulong)parsed ) ) FD_LOG_ERR(( "data buffer too small %lu %lu %lu", *data_len, http->response_len, (ulong)parsed ));
     346           0 :     FD_TEST( *data_len>=http->response_len-(ulong)parsed );
     347           0 :     *data_len = http->response_len - (ulong)parsed;
     348           0 :     fd_memcpy( data, http->response+parsed, *data_len );
     349           0 :     http->content_read += *data_len;
     350           0 :     return FD_SSHTTP_ADVANCE_DATA;
     351           0 :   } else {
     352           0 :     FD_TEST( http->response_len==(ulong)parsed );
     353           0 :     return FD_SSHTTP_ADVANCE_AGAIN;
     354           0 :   }
     355           0 : }
     356             : 
     357             : static int
     358             : read_body( fd_sshttp_t * http,
     359             :            ulong *       data_len,
     360           0 :            uchar *       data ) {
     361           0 :   if( FD_UNLIKELY( http->content_read>=http->content_len ) ) {
     362           0 :     fd_sshttp_cancel( http );
     363           0 :     http->state = FD_SSHTTP_STATE_INIT;
     364           0 :     return FD_SSHTTP_ADVANCE_DONE;
     365           0 :   }
     366             : 
     367           0 :   FD_TEST( http->content_read<http->content_len );
     368           0 :   long read = recv( http->sockfd, data, fd_ulong_min( *data_len, http->content_len-http->content_read ), 0 );
     369           0 :   if( FD_UNLIKELY( -1==read && errno==EAGAIN ) ) return FD_SSHTTP_ADVANCE_AGAIN;
     370           0 :   else if( FD_UNLIKELY( -1==read ) ) {
     371           0 :     fd_sshttp_cancel( http );
     372           0 :     return FD_SSHTTP_ADVANCE_ERROR;
     373           0 :   }
     374             : 
     375           0 :   if( FD_UNLIKELY( !read ) ) return FD_SSHTTP_ADVANCE_AGAIN;
     376             : 
     377           0 :   *data_len = (ulong)read;
     378           0 :   http->content_read += (ulong)read;
     379             : 
     380           0 :   return FD_SSHTTP_ADVANCE_DATA;
     381           0 : }
     382             : 
     383             : void
     384             : fd_sshttp_snapshot_names( fd_sshttp_t const * http,
     385             :                           char const **       full_snapshot_name,
     386           0 :                           char const **       incremental_snapshot_name ) {
     387           0 :   *full_snapshot_name        = http->full_snapshot_name;
     388           0 :   *incremental_snapshot_name = http->incremental_snapshot_name;
     389           0 : }
     390             : 
     391             : ulong
     392           0 : fd_sshttp_content_len( fd_sshttp_t const * http ) {
     393           0 :   return http->content_len;
     394           0 : }
     395             : 
     396             : int
     397             : fd_sshttp_advance( fd_sshttp_t * http,
     398             :                    ulong *       data_len,
     399             :                    uchar *       data,
     400           0 :                    long          now ) {
     401             :   /* TODO: Add timeouts ... */
     402             : 
     403           0 :   switch( http->state ) {
     404           0 :     case FD_SSHTTP_STATE_INIT: return FD_SSHTTP_ADVANCE_AGAIN;
     405           0 :     case FD_SSHTTP_STATE_REQ: return send_request( http, now );
     406           0 :     case FD_SSHTTP_STATE_RESP: return read_response( http, data_len, data, now );
     407           0 :     case FD_SSHTTP_STATE_DL: return read_body( http, data_len, data );
     408           0 :     default: return 0;
     409           0 :   }
     410           0 : }