diff options
Diffstat (limited to 'src/plugins/http/http.h')
| -rw-r--r-- | src/plugins/http/http.h | 530 |
1 files changed, 522 insertions, 8 deletions
diff --git a/src/plugins/http/http.h b/src/plugins/http/http.h index dbae5ac4611..e3ee93b6291 100644 --- a/src/plugins/http/http.h +++ b/src/plugins/http/http.h @@ -16,6 +16,8 @@ #ifndef SRC_PLUGINS_HTTP_HTTP_H_ #define SRC_PLUGINS_HTTP_HTTP_H_ +#include <ctype.h> + #include <vnet/plugin/plugin.h> #include <vpp/app/version.h> @@ -83,6 +85,14 @@ typedef enum http_msg_type_ HTTP_MSG_REPLY } http_msg_type_t; +typedef enum http_target_form_ +{ + HTTP_TARGET_ORIGIN_FORM, + HTTP_TARGET_ABSOLUTE_FORM, + HTTP_TARGET_AUTHORITY_FORM, + HTTP_TARGET_ASTERISK_FORM +} http_target_form_t; + #define foreach_http_content_type \ _ (APP_7Z, ".7z", "application / x - 7z - compressed") \ _ (APP_DOC, ".doc", "application / msword") \ @@ -172,12 +182,50 @@ typedef enum http_content_type_ } http_content_type_t; #define foreach_http_status_code \ + _ (100, CONTINUE, "100 Continue") \ + _ (101, SWITCHING_PROTOCOLS, "101 Switching Protocols") \ _ (200, OK, "200 OK") \ + _ (201, CREATED, "201 Created") \ + _ (202, ACCEPTED, "202 Accepted") \ + _ (203, NON_UTHORITATIVE_INFORMATION, "203 Non-Authoritative Information") \ + _ (204, NO_CONTENT, "204 No Content") \ + _ (205, RESET_CONTENT, "205 Reset Content") \ + _ (206, PARTIAL_CONTENT, "206 Partial Content") \ + _ (300, MULTIPLE_CHOICES, "300 Multiple Choices") \ _ (301, MOVED, "301 Moved Permanently") \ + _ (302, FOUND, "302 Found") \ + _ (303, SEE_OTHER, "303 See Other") \ + _ (304, NOT_MODIFIED, "304 Not Modified") \ + _ (305, USE_PROXY, "305 Use Proxy") \ + _ (307, TEMPORARY_REDIRECT, "307 Temporary Redirect") \ + _ (308, PERMANENT_REDIRECT, "308 Permanent Redirect") \ _ (400, BAD_REQUEST, "400 Bad Request") \ + _ (401, UNAUTHORIZED, "401 Unauthorized") \ + _ (402, PAYMENT_REQUIRED, "402 Payment Required") \ + _ (403, FORBIDDEN, "403 Forbidden") \ _ (404, NOT_FOUND, "404 Not Found") \ _ (405, METHOD_NOT_ALLOWED, "405 Method Not Allowed") \ - _ (500, INTERNAL_ERROR, "500 Internal Server Error") + _ (406, NOT_ACCEPTABLE, "406 Not Acceptable") \ + _ (407, PROXY_AUTHENTICATION_REQUIRED, "407 Proxy Authentication Required") \ + _ (408, REQUEST_TIMEOUT, "408 Request Timeout") \ + _ (409, CONFLICT, "409 Conflict") \ + _ (410, GONE, "410 Gone") \ + _ (411, LENGTH_REQUIRED, "411 Length Required") \ + _ (412, PRECONDITION_FAILED, "412 Precondition Failed") \ + _ (413, CONTENT_TOO_LARGE, "413 Content Too Large") \ + _ (414, URI_TOO_LONG, "414 URI Too Long") \ + _ (415, UNSUPPORTED_MEDIA_TYPE, "415 Unsupported Media Type") \ + _ (416, RANGE_NOT_SATISFIABLE, "416 Range Not Satisfiable") \ + _ (417, EXPECTATION_FAILED, "417 Expectation Failed") \ + _ (421, MISDIRECTED_REQUEST, "421 Misdirected Request") \ + _ (422, UNPROCESSABLE_CONTENT, "422 Unprocessable_Content") \ + _ (426, UPGRADE_REQUIRED, "426 Upgrade Required") \ + _ (500, INTERNAL_ERROR, "500 Internal Server Error") \ + _ (501, NOT_IMPLEMENTED, "501 Not Implemented") \ + _ (502, BAD_GATEWAY, "502 Bad Gateway") \ + _ (503, SERVICE_UNAVAILABLE, "503 Service Unavailable") \ + _ (504, GATEWAY_TIMEOUT, "504 Gateway Timeout") \ + _ (505, HTTP_VERSION_NOT_SUPPORTED, "505 HTTP Version Not Supported") typedef enum http_status_code_ { @@ -187,6 +235,51 @@ typedef enum http_status_code_ HTTP_N_STATUS } http_status_code_t; +#define HTTP_HEADER_ACCEPT "Accept" +#define HTTP_HEADER_ACCEPT_CHARSET "Accept-Charset" +#define HTTP_HEADER_ACCEPT_ENCODING "Accept-Encoding" +#define HTTP_HEADER_ACCEPT_LANGUAGE "Accept-Language" +#define HTTP_HEADER_ACCEPT_RANGES "Accept-Ranges" +#define HTTP_HEADER_ALLOW "Allow" +#define HTTP_HEADER_AUTHENTICATION_INFO "Authentication-Info" +#define HTTP_HEADER_AUTHORIZATION "Authorization" +#define HTTP_HEADER_CLOSE "Close" +#define HTTP_HEADER_CONNECTION "Connection" +#define HTTP_HEADER_CONTENT_ENCODING "Content-Encoding" +#define HTTP_HEADER_CONTENT_LANGUAGE "Content-Language" +#define HTTP_HEADER_CONTENT_LENGTH "Content-Length" +#define HTTP_HEADER_CONTENT_LOCATION "Content-Location" +#define HTTP_HEADER_CONTENT_RANGE "Content-Range" +#define HTTP_HEADER_CONTENT_TYPE "Content-Type" +#define HTTP_HEADER_DATE "Date" +#define HTTP_HEADER_ETAG "ETag" +#define HTTP_HEADER_EXPECT "Expect" +#define HTTP_HEADER_FROM "From" +#define HTTP_HEADER_HOST "Host" +#define HTTP_HEADER_IF_MATCH "If-Match" +#define HTTP_HEADER_IF_MODIFIED_SINCE "If-Modified-Since" +#define HTTP_HEADER_IF_NONE_MATCH "If-None-Match" +#define HTTP_HEADER_IF_RANGE "If-Range" +#define HTTP_HEADER_IF_UNMODIFIED_SINCE "If-Unmodified-Since" +#define HTTP_HEADER_LAST_MODIFIED "Last-Modified" +#define HTTP_HEADER_LOCATION "Location" +#define HTTP_HEADER_MAX_FORWARDS "Max-Forwards" +#define HTTP_HEADER_PROXY_AUTHENTICATE "Proxy-Authenticate" +#define HTTP_HEADER_PROXY_AUTHENTICATION_INFO "Proxy-Authentication-Info" +#define HTTP_HEADER_PROXY_AUTHORIZATION "Proxy-Authorization" +#define HTTP_HEADER_RANGE "Range" +#define HTTP_HEADER_REFERER "Referer" +#define HTTP_HEADER_RETRY_AFTER "Retry-After" +#define HTTP_HEADER_SERVER "Server" +#define HTTP_HEADER_TE "TE" +#define HTTP_HEADER_TRAILER "Trailer" +#define HTTP_HEADER_TRANSFER_ENCODING "Transfer-Encoding" +#define HTTP_HEADER_UPGRADE "Upgrade" +#define HTTP_HEADER_USER_AGENT "User-Agent" +#define HTTP_HEADER_VARY "Vary" +#define HTTP_HEADER_VIA "Via" +#define HTTP_HEADER_WWW_AUTHENTICATE "WWW-Authenticate" + typedef enum http_msg_data_type_ { HTTP_MSG_DATA_INLINE, @@ -197,6 +290,15 @@ typedef struct http_msg_data_ { http_msg_data_type_t type; u64 len; + http_target_form_t target_form; + u32 target_path_offset; + u32 target_path_len; + u32 target_query_offset; + u32 target_query_len; + u32 headers_offset; + u32 headers_len; + u32 body_offset; + u32 body_len; u8 data[0]; } http_msg_data_t; @@ -227,6 +329,7 @@ typedef struct http_tc_ http_conn_state_t state; u32 timer_handle; + u8 *app_name; /* * Current request @@ -238,6 +341,15 @@ typedef struct http_tc_ http_buffer_t tx_buf; u32 to_recv; u32 bytes_dequeued; + http_target_form_t target_form; + u32 target_path_offset; + u32 target_path_len; + u32 target_query_offset; + u32 target_query_len; + u32 headers_offset; + u32 headers_len; + u32 body_offset; + u32 body_len; } http_conn_t; typedef struct http_worker_ @@ -266,14 +378,416 @@ typedef struct http_main_ u32 fifo_size; } http_main_t; -static inline int -http_state_is_tx_valid (http_conn_t *hc) +always_inline int +_validate_target_syntax (u8 *target, int is_query, int *is_encoded) { - http_state_t state = hc->http_state; - return (state == HTTP_STATE_APP_IO_MORE_DATA || - state == HTTP_STATE_CLIENT_IO_MORE_DATA || - state == HTTP_STATE_WAIT_APP_REPLY || - state == HTTP_STATE_WAIT_APP_METHOD); + int i, encoded = 0; + + static uword valid_chars[4] = { + /* !$&'()*+,-./0123456789:;= */ + 0x2fffffd200000000, + /* @ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~ */ + 0x47fffffe87ffffff, + 0x0000000000000000, + 0x0000000000000000, + }; + + for (i = 0; i < vec_len (target); i++) + { + if (clib_bitmap_get_no_check (valid_chars, target[i])) + continue; + /* target was already split after first question mark, + * for query it is valid character */ + if (is_query && target[i] == '?') + continue; + /* pct-encoded = "%" HEXDIG HEXDIG */ + if (target[i] == '%') + { + if ((i + 2) > vec_len (target)) + return -1; + if (!isxdigit (target[i + 1]) || !isxdigit (target[i + 2])) + return -1; + i += 2; + encoded = 1; + continue; + } + clib_warning ("invalid character %d", target[i]); + return -1; + } + if (is_encoded) + *is_encoded = encoded; + return 0; +} + +/** + * An "absolute-path" rule validation (RFC9110 section 4.1). + * + * @param path Target path to validate. + * @param is_encoded Return flag that indicates if percent-encoded (optional). + * + * @return @c 0 on success. + */ +always_inline int +http_validate_abs_path_syntax (u8 *path, int *is_encoded) +{ + return _validate_target_syntax (path, 0, is_encoded); +} + +/** + * A "query" rule validation (RFC3986 section 2.1). + * + * @param query Target query to validate. + * @param is_encoded Return flag that indicates if percent-encoded (optional). + * + * @return @c 0 on success. + */ +always_inline int +http_validate_query_syntax (u8 *query, int *is_encoded) +{ + return _validate_target_syntax (query, 1, is_encoded); +} + +#define htoi(x) (isdigit (x) ? (x - '0') : (tolower (x) - 'a' + 10)) + +/** + * Decode percent-encoded data. + * + * @param src Data to decode. + * + * @return New vector with decoded data. + * + * The caller is always responsible to free the returned vector. + */ +always_inline u8 * +http_percent_decode (u8 *src) +{ + int i; + u8 *decoded_uri = 0; + + for (i = 0; i < vec_len (src); i++) + { + if (src[i] == '%') + { + u8 c = (htoi (src[i + 1]) << 4) | htoi (src[i + 2]); + vec_add1 (decoded_uri, c); + i += 2; + } + else + vec_add1 (decoded_uri, src[i]); + } + return decoded_uri; +} + +/** + * Remove dot segments from path (RFC3986 section 5.2.4) + * + * @param path Path to sanitize. + * + * @return New vector with sanitized path. + * + * The caller is always responsible to free the returned vector. + */ +always_inline u8 * +http_path_remove_dot_segments (u8 *path) +{ + u32 *segments = 0, *segments_len = 0, segment_len; + u8 *new_path = 0; + int i, ii; + + if (!path) + return vec_new (u8, 0); + + segments = vec_new (u32, 1); + /* first segment */ + segments[0] = 0; + /* find all segments */ + for (i = 1; i < (vec_len (path) - 1); i++) + { + if (path[i] == '/') + vec_add1 (segments, i + 1); + } + /* dummy tail */ + vec_add1 (segments, vec_len (path)); + + /* scan all segments for "." and ".." */ + segments_len = vec_new (u32, vec_len (segments) - 1); + for (i = 0; i < vec_len (segments_len); i++) + { + segment_len = segments[i + 1] - segments[i]; + if (segment_len == 2 && path[segments[i]] == '.') + segment_len = 0; + else if (segment_len == 3 && path[segments[i]] == '.' && + path[segments[i] + 1] == '.') + { + segment_len = 0; + /* remove parent (if any) */ + for (ii = i - 1; ii >= 0; ii--) + { + if (segments_len[ii]) + { + segments_len[ii] = 0; + break; + } + } + } + segments_len[i] = segment_len; + } + + /* we might end with empty path, so return at least empty vector */ + new_path = vec_new (u8, 0); + /* append all valid segments */ + for (i = 0; i < vec_len (segments_len); i++) + { + if (segments_len[i]) + vec_add (new_path, path + segments[i], segments_len[i]); + } + vec_free (segments); + vec_free (segments_len); + return new_path; +} + +always_inline int +_parse_field_name (u8 **pos, u8 *end, u8 **field_name_start, + u32 *field_name_len) +{ + u32 name_len = 0; + u8 *p; + + static uword tchar[4] = { + /* !#$%'*+-.0123456789 */ + 0x03ff6cba00000000, + /* ABCDEFGHIJKLMNOPQRSTUVWXYZ^_`abcdefghijklmnopqrstuvwxyz|~ */ + 0x57ffffffc7fffffe, + 0x0000000000000000, + 0x0000000000000000, + }; + + p = *pos; + + *field_name_start = p; + while (p != end) + { + if (clib_bitmap_get_no_check (tchar, *p)) + { + name_len++; + p++; + } + else if (*p == ':') + { + if (name_len == 0) + { + clib_warning ("empty field name"); + return -1; + } + *field_name_len = name_len; + p++; + *pos = p; + return 0; + } + else + { + clib_warning ("invalid character %d", *p); + return -1; + } + } + clib_warning ("field name end not found"); + return -1; +} + +always_inline int +_parse_field_value (u8 **pos, u8 *end, u8 **field_value_start, + u32 *field_value_len) +{ + u32 value_len = 0; + u8 *p; + + p = *pos; + + /* skip leading whitespace */ + while (1) + { + if (p == end) + { + clib_warning ("field value not found"); + return -1; + } + else if (*p != ' ' && *p != '\t') + { + break; + } + p++; + } + + *field_value_start = p; + while (p != end) + { + if (*p == '\r') + { + if ((end - p) < 1) + { + clib_warning ("incorrect field line end"); + return -1; + } + p++; + if (*p == '\n') + { + if (value_len == 0) + { + clib_warning ("empty field value"); + return -1; + } + p++; + *pos = p; + /* skip trailing whitespace */ + p = *field_value_start + value_len - 1; + while (*p == ' ' || *p == '\t') + { + p--; + value_len--; + } + *field_value_len = value_len; + return 0; + } + clib_warning ("CR without LF"); + return -1; + } + if (*p < ' ' && *p != '\t') + { + clib_warning ("invalid character %d", *p); + return -1; + } + p++; + value_len++; + } + + clib_warning ("field value end not found"); + return -1; +} + +typedef struct +{ + u8 *name; + u8 *value; +} http_header_t; + +typedef struct +{ + http_header_t *headers; + uword *value_by_name; +} http_header_table_t; + +/** + * Free header table's memory. + * + * @param ht Header table to free. + */ +always_inline void +http_free_header_table (http_header_table_t *ht) +{ + http_header_t *header; + vec_foreach (header, ht->headers) + { + vec_free (header->name); + vec_free (header->value); + } + vec_free (ht->headers); + hash_free (ht->value_by_name); + clib_mem_free (ht); +} + +/** + * Parse headers in given vector. + * + * @param headers Vector to parse. + * @param [out] header_table Parsed headers in case of success. + * + * @return @c 0 on success. + * + * The caller is responsible to free the returned @c header_table + * using @c http_free_header_table . + */ +always_inline int +http_parse_headers (u8 *headers, http_header_table_t **header_table) +{ + u8 *pos, *end, *name_start, *value_start, *name; + u32 name_len, value_len; + int rv; + http_header_t *header; + http_header_table_t *ht; + uword *p; + + end = headers + vec_len (headers); + pos = headers; + + ht = clib_mem_alloc (sizeof (*ht)); + ht->value_by_name = hash_create_string (0, sizeof (uword)); + ht->headers = 0; + do + { + rv = _parse_field_name (&pos, end, &name_start, &name_len); + if (rv != 0) + { + http_free_header_table (ht); + return rv; + } + rv = _parse_field_value (&pos, end, &value_start, &value_len); + if (rv != 0) + { + http_free_header_table (ht); + return rv; + } + name = vec_new (u8, name_len); + clib_memcpy (name, name_start, name_len); + vec_terminate_c_string (name); + /* check if header is repeated */ + p = hash_get_mem (ht->value_by_name, name); + if (p) + { + /* if yes combine values */ + header = vec_elt_at_index (ht->headers, p[0]); + vec_pop (header->value); /* drop null byte */ + header->value = format (header->value, ", %U%c", format_ascii_bytes, + value_start, value_len, 0); + vec_free (name); + continue; + } + /* or create new record */ + vec_add2 (ht->headers, header, sizeof (*header)); + header->name = name; + header->value = vec_new (u8, value_len); + clib_memcpy (header->value, value_start, value_len); + vec_terminate_c_string (header->value); + hash_set_mem (ht->value_by_name, header->name, header - ht->headers); + } + while (pos != end); + + *header_table = ht; + + return 0; +} + +/** + * Try to find given header name in header table. + * + * @param header_table Header table to search. + * @param name Header name to match. + * + * @return Header's value in case of success, @c 0 otherwise. + */ +always_inline const char * +http_get_header (http_header_table_t *header_table, const char *name) +{ + uword *p; + http_header_t *header; + + p = hash_get_mem (header_table->value_by_name, name); + if (p) + { + header = vec_elt_at_index (header_table->headers, p[0]); + return (const char *) header->value; + } + + return 0; } #endif /* SRC_PLUGINS_HTTP_HTTP_H_ */ |