protobuf/upb/decode.int.h

/*
** Internal implementation details of the decoder that are shared between
** decode.c and decode_fast.c.
*/

#ifndef UPB_DECODE_INT_H_
#define UPB_DECODE_INT_H_

#include <setjmp.h>

#include "upb/msg.h"
#include "upb/upb.int.h"
#include "upb/decode_fast.h"

/* Must be last. */
#include "upb/port_def.inc"

typedef struct upb_decstate {
  const char *end;         /* Can read up to 16 bytes slop beyond this. */
  const char *limit_ptr;   /* = end + UPB_MIN(limit, 0) */
  upb_msg *unknown_msg;    /* If non-NULL, add unknown data at buffer flip. */
  const char *unknown;     /* Start of unknown data. */
  int limit;               /* Submessage limit relative to end. */
  int depth;
  uint32_t end_group; /* Set to field number of END_GROUP tag, if any. */
  bool alias;
  char patch[32];
  upb_arena arena;
  jmp_buf err;
} upb_decstate;

/* Error function that will abort decoding with longjmp(). We can't declare this
 * UPB_NORETURN, even though it is appropriate, because if we do then compilers
 * will "helpfully" refuse to tailcall to it
 * (see: https://stackoverflow.com/a/55657013), which will defeat a major goal
 * of our optimizations. That is also why we must declare it in a separate file,
 * otherwise the compiler will see that it calls longjmp() and deduce that it is
 * noreturn. */
const char *fastdecode_err(upb_decstate *d);

extern const uint8_t upb_utf8_offsets[];

UPB_INLINE
bool decode_verifyutf8_inl(const char *buf, int len) {
  int i, j;
  uint8_t offset;

  i = 0;
  while (i < len) {
    offset = upb_utf8_offsets[(uint8_t)buf[i]];
    if (offset == 0 || i + offset > len) {
      return false;
    }
    for (j = i + 1; j < i + offset; j++) {
      if ((buf[j] & 0xc0) != 0x80) {
        return false;
      }
    }
    i += offset;
  }
  return i == len;
}

/* x86-64 pointers always have the high 16 bits matching. So we can shift
 * left 8 and right 8 without loss of information. */
UPB_INLINE intptr_t decode_totable(const upb_msglayout *tablep) {
  return ((intptr_t)tablep << 8) | tablep->table_mask;
}

UPB_INLINE const upb_msglayout *decode_totablep(intptr_t table) {
  return (void*)(table >> 8);
}

UPB_INLINE
const char *decode_isdonefallback_inl(upb_decstate *d, const char *ptr,
                                      int overrun) {
  if (overrun < d->limit) {
    /* Need to copy remaining data into patch buffer. */
    UPB_ASSERT(overrun < 16);
    if (d->unknown_msg) {
      if (!_upb_msg_addunknown(d->unknown_msg, d->unknown, ptr - d->unknown,
                               &d->arena)) {
        return NULL;
      }
      d->unknown = &d->patch[0] + overrun;
    }
    memset(d->patch + 16, 0, 16);
    memcpy(d->patch, d->end, 16);
    ptr = &d->patch[0] + overrun;
    d->end = &d->patch[16];
    d->limit -= 16;
    d->limit_ptr = d->end + d->limit;
    d->alias = false;
    UPB_ASSERT(ptr < d->limit_ptr);
    return ptr;
  } else {
    return NULL;
  }
}

const char *decode_isdonefallback(upb_decstate *d, const char *ptr,
                                  int overrun);

UPB_INLINE
bool decode_isdone(upb_decstate *d, const char **ptr) {
  int overrun = *ptr - d->end;
  if (UPB_LIKELY(*ptr < d->limit_ptr)) {
    return false;
  } else if (UPB_LIKELY(overrun == d->limit)) {
    return true;
  } else {
    *ptr = decode_isdonefallback(d, *ptr, overrun);
    return false;
  }
}

UPB_INLINE
const char *fastdecode_tagdispatch(upb_decstate *d, const char *ptr,
                                    upb_msg *msg, intptr_t table,
                                    uint64_t hasbits, uint32_t tag) {
  const upb_msglayout *table_p = decode_totablep(table);
  uint8_t mask = table;
  uint64_t data;
  size_t idx = tag & mask;
  UPB_ASSUME((idx & 7) == 0);
  idx >>= 3;
  data = table_p->fasttable[idx].field_data ^ tag;
  return table_p->fasttable[idx].field_parser(d, ptr, msg, table, hasbits, data);
}

UPB_INLINE uint32_t fastdecode_loadtag(const char* ptr) {
  uint16_t tag;
  memcpy(&tag, ptr, 2);
  return tag;
}

UPB_INLINE int decode_pushlimit(upb_decstate *d, const char *ptr, int size) {
  int limit = size + (int)(ptr - d->end);
  int delta = d->limit - limit;
  d->limit = limit;
  d->limit_ptr = d->end + UPB_MIN(0, limit);
  return delta;
}

UPB_INLINE void decode_poplimit(upb_decstate *d, const char *ptr,
                                int saved_delta) {
  UPB_ASSERT(ptr - d->end == d->limit);
  d->limit += saved_delta;
  d->limit_ptr = d->end + UPB_MIN(0, d->limit);
  UPB_ASSERT(d->limit_ptr == d->end + UPB_MIN(0, d->limit));
}

#include "upb/port_undef.inc"

#endif  /* UPB_DECODE_INT_H_ */