mirror of
https://github.com/Unidata/netcdf-c.git
synced 2024-11-21 03:13:42 +08:00
4db4393e69
strlcat provides better protection against buffer overflows. Code is taken from the FreeBSD project source code. Specifically: https://github.com/freebsd/freebsd/blob/master/lib/libc/string/strlcat.c License appears to be acceptable, but needs to be checked by e.g. Debian. Step 1: 1. Add to netcdf-c/include/ncconfigure.h to use our version if not already available as determined by HAVE_STRLCAT in config.h. 2. Add the strlcat code to libdispatch/dstring.c 3. Turns out that strlcat was already defined in several places. So remove it from: ncgen3/genlib.c ncdump/dumplib.c 3. Define strlcat extern definition in ncconfigure.h. 4. Modify following directories to use strlcat: libdap2 libdap4 ncdap_test dap4_test Will do others in subsequent steps.
171 lines
4.1 KiB
C
171 lines
4.1 KiB
C
/*********************************************************************
|
|
* Copyright 2016, UCAR/Unidata
|
|
* See netcdf/COPYRIGHT file for copying and redistribution conditions.
|
|
*********************************************************************/
|
|
|
|
#include "config.h"
|
|
#include <stdlib.h>
|
|
#include <assert.h>
|
|
#include "netcdf.h"
|
|
#include "d4util.h"
|
|
#include "d4odom.h"
|
|
|
|
/**********************************************/
|
|
/* Define methods for a dimension dapodometer*/
|
|
|
|
/* Build an odometer covering slices startslice upto,
|
|
but not including, stopslice
|
|
*/
|
|
|
|
/*
|
|
Future optimization. It is sometimes the case that
|
|
some suffix of the index set cover a contiguous chunk
|
|
of the index space. This happens when the index positions
|
|
starting at p upto rank have a start of zero, a count
|
|
the same as the max index size, and a stride of one.
|
|
Then we can effectively tread indices 0 thru p-1 as
|
|
a short odometer that steps through the chunks.
|
|
In practice, it is usually only ncdump that does this.
|
|
A simpler optimization that is to see if the (start,count)
|
|
covers the whole index space and has stride one.
|
|
We can detect and use this optimization elsewhere to
|
|
circumvent use of the odometer altogether.
|
|
See d4odometer#NCD4_odomWhole().
|
|
*/
|
|
|
|
D4odometer*
|
|
d4odom_new(size_t rank,
|
|
const size_t* start, const size_t* count,
|
|
const ptrdiff_t* stride, const size_t* size)
|
|
{
|
|
int i;
|
|
D4odometer* odom = (D4odometer*)calloc(1,sizeof(D4odometer));
|
|
if(odom == NULL)
|
|
return NULL;
|
|
odom->rank = rank;
|
|
assert(odom->rank <= NC_MAX_VAR_DIMS);
|
|
for(i=0;i<odom->rank;i++) {
|
|
size_t istart,icount,istop,ideclsize;
|
|
ptrdiff_t istride;
|
|
istart = (start != NULL ? start[i] : 0);
|
|
icount = (count != NULL ? count[i] : (size != NULL ? size[i] : 1));
|
|
istride = (size_t)(stride != NULL ? stride[i] : 1);
|
|
istop = istart + icount*istride;
|
|
ideclsize = (size != NULL ? size[i]: (istop - istart));
|
|
odom->start[i] = istart;
|
|
odom->stop[i] = istop;
|
|
odom->stride[i] = istride;
|
|
odom->declsize[i] = ideclsize;
|
|
odom->index[i] = odom->start[i];
|
|
}
|
|
return odom;
|
|
}
|
|
|
|
void
|
|
d4odom_free(D4odometer* odom)
|
|
{
|
|
if(odom) free(odom);
|
|
}
|
|
|
|
#if 0
|
|
char*
|
|
d4odom_print(D4odometer* odom)
|
|
{
|
|
int i;
|
|
static char line[1024];
|
|
char tmp[64];
|
|
line[0] = '\0';
|
|
if(odom->rank == 0) {
|
|
strlcat(line,"[]",sizeof(line));
|
|
} else for(i=0;i<odom->rank;i++) {
|
|
sprintf(tmp,"[%lu/%lu:%lu:%lu]",
|
|
(size_t)odom->index[i],
|
|
(size_t)odom->start[i],
|
|
(size_t)odom->stride[i],
|
|
(size_t)odom->length[i]);
|
|
strlcat(line,tmp,sizeof(line));
|
|
}
|
|
return line;
|
|
}
|
|
#endif
|
|
|
|
int
|
|
d4odom_more(D4odometer* odom)
|
|
{
|
|
return (odom->index[0] < odom->stop[0]);
|
|
}
|
|
|
|
d4size_t
|
|
d4odom_next(D4odometer* odom)
|
|
{
|
|
int i; /* do not make unsigned */
|
|
d4size_t count;
|
|
if(odom->rank == 0) { /*scalar*/
|
|
odom->index[0]++;
|
|
return 0;
|
|
}
|
|
count = d4odom_offset(odom); /* convenience */
|
|
for(i=odom->rank-1;i>=0;i--) {
|
|
odom->index[i] += odom->stride[i];
|
|
if(odom->index[i] < odom->stop[i]) break;
|
|
if(i == 0) break; /* leave the 0th entry if it overflows*/
|
|
odom->index[i] = odom->start[i]; /* reset this position*/
|
|
}
|
|
return count;
|
|
}
|
|
|
|
/* Convert current d4odometer settings to a single integer offset*/
|
|
d4size_t
|
|
d4odom_offset(D4odometer* odom)
|
|
{
|
|
int i;
|
|
d4size_t offset = 0;
|
|
for(i=0;i<odom->rank;i++) {
|
|
offset *= odom->declsize[i];
|
|
offset += odom->index[i];
|
|
}
|
|
return offset;
|
|
}
|
|
|
|
/**************************************************/
|
|
/*
|
|
Given a d4odometer, compute the total
|
|
number of elements in its space.
|
|
*/
|
|
|
|
d4size_t
|
|
d4odom_nelements(D4odometer* odom)
|
|
{
|
|
size_t i;
|
|
d4size_t count = 1;
|
|
for(i=0;i<odom->rank;i++) {
|
|
count *= odom->declsize[i];
|
|
}
|
|
return count;
|
|
}
|
|
|
|
int
|
|
d4odom_isWhole(D4odometer* odom)
|
|
{
|
|
int i;
|
|
for(i=0;i<odom->rank;i++) {
|
|
if(odom->start[i] != 0
|
|
|| odom->stride[i] != 1
|
|
|| odom->stop[i] != odom->declsize[i])
|
|
return 0;
|
|
}
|
|
return 1;
|
|
}
|
|
|
|
/* Scalar Odometer support */
|
|
D4odometer*
|
|
d4scalarodom_new(void)
|
|
{
|
|
D4odometer* odom = (D4odometer*)calloc(1,sizeof(D4odometer));
|
|
odom->rank = 0;
|
|
/* Fake things to execute exactly once */
|
|
odom->index[0] = 0;
|
|
odom->stop[0] = 1;
|
|
return odom;
|
|
}
|