Skip to content

Commit

Permalink
Merge pull request #202 from tim-griesbach/feature-scda
Browse files Browse the repository at this point in the history 
scda: Fuzzy error synchronization and padding optimization
  • Loading branch information
@cburstedde
cburstedde authored Sep 19, 2024
2 parents 0f05ae2 + 4334546 commit 74397d3
Showing 1 changed file with 78 additions and 15 deletions.
93 changes: 78 additions & 15 deletions src/sc_scda.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -131,6 +131,7 @@
SC_CHECK_MPI(sc_MPI_Bcast(&errcode->mpiret,\
1, sc_MPI_INT, root, \
fc->mpicomm)); \
sc_scda_fuzzy_sync_state (fc); \
if (!sc_scda_ferror_is_success (*errcode)) {\
sc_scda_file_error_cleanup (&fc->file); \
SC_FREE (fc); \
Expand Down Expand Up @@ -343,29 +344,24 @@ sc_scda_pad_to_fix_len_inplace (const char *input_data, size_t input_len,
}

/** This function checks if \b padded_data is actually padded to \b pad_len.
* Moreover, the raw data is extracted.
*
* \param [in] padded_data The padded data.
* \param [in] pad_len The length of \b padded_data in number of bytes.
* \param [out] raw_data On output the raw data extracted from
* \b padded_data. The byte count of \b raw_data
* is \b raw_len. \b raw_data must be at least
* \b pad_len - 4. Undefined data if the function
* returns true.
* \param [out] raw_len The length of \b raw_data in number of bytes.
* Undefined if the function returns true.
* \return True if \b padded_data does not satisfy the
* scda padding convention for fixed-length paddding.
* False, otherwise.
*/
static int
sc_scda_get_pad_to_fix_len (const char *padded_data, size_t pad_len,
char *raw_data, size_t *raw_len)
sc_scda_check_pad_to_fix_len (const char *padded_data, size_t pad_len,
size_t *raw_len)
{
size_t si;

*raw_len = 0;

SC_ASSERT (padded_data != NULL);
SC_ASSERT (raw_data != NULL);
SC_ASSERT (raw_len != NULL);

if (pad_len < 4) {
Expand All @@ -387,8 +383,44 @@ sc_scda_get_pad_to_fix_len (const char *padded_data, size_t pad_len,
return -1;
}

/* the padding was valid and the remaining data is the actual data */
/* the padding is valid */
*raw_len = si;

return 0;
}

/** Checks the padding convention and extracts the data in case of success.
*
* In more concrete terms, the function checks if \b padded_data is actually
* padded to \b pad_len and extracts the raw data in case of success.
*
* \param [in] padded_data The padded data.
* \param [in] pad_len The length of \b padded_data in number of bytes.
* \param [out] raw_data On output the raw data extracted from
* \b padded_data. The byte count of \b raw_data
* is \b raw_len. \b raw_data must be at least
* \b pad_len - 4. Undefined data if the function
* returns true.
* \param [out] raw_len The length of \b raw_data in number of bytes.
* Undefined if the function returns true.
* \return True if \b padded_data does not satisfy the
* scda padding convention for fixed-length paddding.
* False, otherwise.
*/
static int
sc_scda_get_pad_to_fix_len (const char *padded_data, size_t pad_len,
char *raw_data, size_t *raw_len)
{
SC_ASSERT (padded_data != NULL);
SC_ASSERT (raw_data != NULL);
SC_ASSERT (raw_len != NULL);

if (sc_scda_check_pad_to_fix_len (padded_data, pad_len, raw_len)) {
/* invalid padding */
return -1;
}

/* the padding is valid and the remaining data is the actual data */
sc_scda_copy_bytes (raw_data, padded_data, *raw_len);

return 0;
Expand Down Expand Up @@ -991,6 +1023,39 @@ sc_scda_mpiret_to_errcode (int mpiret, sc_scda_ferror_t * scda_errorcode,
scda_errorcode->mpiret = mpiret_internal;
}

/** Synchronize the random state.
*
* This function is required since the fuzzy error state (cf. fuzzy_seed in
* \ref sc_scda_fcontext) may become unsynchronized due to a different number
* of samples on different MPI ranks, which is due to non-collective code
* paths.
*
* \param [in, out] fc The file context used in the last \ref
* sc_scda_scdaret_to_errcode or \ref
* sc_scda_mpiret_to_errcode call.
* On output \b fuzzy_seed is set to the global
* maximum of all local \b fuzzy_seed values.
*/
static void
sc_scda_fuzzy_sync_state (sc_scda_fcontext_t *fc)
{
int mpiret;
unsigned long fuzzy_state, global_state;

SC_ASSERT (fc != NULL);

/* get local fuzzy_state */
fuzzy_state = (unsigned long) fc->fuzzy_seed;

/* determine maximal state */
mpiret = sc_MPI_Allreduce (&fuzzy_state, &global_state, 1,
sc_MPI_UNSIGNED_LONG, sc_MPI_MAX, fc->mpicomm);
SC_CHECK_MPI (mpiret);

/* assign global state */
fc->fuzzy_seed = (sc_rand_state_t) global_state;
}

/** Check if an error code is valid. */
static int
sc_scda_errcode_is_valid (sc_scda_ferror_t errcode)
Expand Down Expand Up @@ -1718,7 +1783,6 @@ sc_scda_check_file_header (const char *file_header_data, char *user_string,
size_t *len)
{
int current_pos;
char vendor_string[SC_SCDA_VENDOR_STRING_BYTES];
size_t vendor_len;

SC_ASSERT (file_header_data != NULL);
Expand Down Expand Up @@ -1749,13 +1813,12 @@ sc_scda_check_file_header (const char *file_header_data, char *user_string,
current_pos = SC_SCDA_MAGIC_BYTES + 1;

/* check the padding of the vendor string */
if (sc_scda_get_pad_to_fix_len (&file_header_data[current_pos],
SC_SCDA_VENDOR_STRING_FIELD, vendor_string,
&vendor_len)) {
if (sc_scda_check_pad_to_fix_len (&file_header_data[current_pos],
SC_SCDA_VENDOR_STRING_FIELD, &vendor_len)) {
/* wrong padding format */
return -1;
}
/* vendor string content is not checked */
/* vendor string content is not checked and hence not read */

current_pos += SC_SCDA_VENDOR_STRING_FIELD + 2;
/* check the user string */
Expand Down

0 comments on commit 74397d3

Please sign in to comment.