netcdf-c/libdap4/d4file.c
Dennis Heimbigner 8b9253fef2 Fix various problem around VLEN's
re: https://github.com/Unidata/netcdf-c/issues/541
re: https://github.com/Unidata/netcdf-c/issues/1208
re: https://github.com/Unidata/netcdf-c/issues/2078
re: https://github.com/Unidata/netcdf-c/issues/2041
re: https://github.com/Unidata/netcdf-c/issues/2143

For a long time, there have been known problems with the
management of complex types containing VLENs.  This also
involves the string type because it is stored as a VLEN of
chars.

This PR (mostly) fixes this problem. But note that it adds new
functions to netcdf.h (see below) and this may require bumping
the .so number.  These new functions can be removed, if desired,
in favor of functions in netcdf_aux.h, but netcdf.h seems the
better place for them because they are intended as alternatives
to the nc_free_vlen and nc_free_string functions already in
netcdf.h.

The term complex type refers to any type that directly or
transitively references a VLEN type. So an array of VLENS, a
compound with a VLEN field, and so on.

In order to properly handle instances of these complex types, it
is necessary to have function that can recursively walk
instances of such types to perform various actions on them.  The
term "deep" is also used to mean recursive.

At the moment, the two operations needed by the netcdf library are:
* free'ing an instance of the complex type
* copying an instance of the complex type.

The current library does only shallow free and shallow copy of
complex types. This means that only the top level is properly
free'd or copied, but deep internal blocks in the instance are
not touched.

Note that the term "vector" will be used to mean a contiguous (in
memory) sequence of instances of some type. Given an array with,
say, dimensions 2 X 3 X 4, this will be stored in memory as a
vector of length 2*3*4=24 instances.

The use cases are primarily these.

## nc_get_vars
Suppose one is reading a vector of instances using nc_get_vars
(or nc_get_vara or nc_get_var, etc.).  These functions will
return the vector in the top-level memory provided.  All
interior blocks (form nested VLEN or strings) will have been
dynamically allocated.

After using this vector of instances, it is necessary to free
(aka reclaim) the dynamically allocated memory, otherwise a
memory leak occurs.  So, the recursive reclaim function is used
to walk the returned instance vector and do a deep reclaim of
the data.

Currently functions are defined in netcdf.h that are supposed to
handle this: nc_free_vlen(), nc_free_vlens(), and
nc_free_string().  Unfortunately, these functions only do a
shallow free, so deeply nested instances are not properly
handled by them.

Note that internally, the provided data is immediately written so
there is no need to copy it. But the caller may need to reclaim the
data it passed into the function.

## nc_put_att
Suppose one is writing a vector of instances as the data of an attribute
using, say, nc_put_att.

Internally, the incoming attribute data must be copied and stored
so that changes/reclamation of the input data will not affect
the attribute.

Again, the code inside the netcdf library does only shallow copying
rather than deep copy. As a result, one sees effects such as described
in Github Issue https://github.com/Unidata/netcdf-c/issues/2143.

Also, after defining the attribute, it may be necessary for the user
to free the data that was provided as input to nc_put_att().

## nc_get_att
Suppose one is reading a vector of instances as the data of an attribute
using, say, nc_get_att.

Internally, the existing attribute data must be copied and returned
to the caller, and the caller is responsible for reclaiming
the returned data.

Again, the code inside the netcdf library does only shallow copying
rather than deep copy. So this can lead to memory leaks and errors
because the deep data is shared between the library and the user.

# Solution

The solution is to build properly recursive reclaim and copy
functions and use those as needed.
These recursive functions are defined in libdispatch/dinstance.c
and their signatures are defined in include/netcdf.h.
For back compatibility, corresponding "ncaux_XXX" functions
are defined in include/netcdf_aux.h.
````
int nc_reclaim_data(int ncid, nc_type xtypeid, void* memory, size_t count);
int nc_reclaim_data_all(int ncid, nc_type xtypeid, void* memory, size_t count);
int nc_copy_data(int ncid, nc_type xtypeid, const void* memory, size_t count, void* copy);
int nc_copy_data_all(int ncid, nc_type xtypeid, const void* memory, size_t count, void** copyp);
````
There are two variants. The first two, nc_reclaim_data() and
nc_copy_data(), assume the top-level vector is managed by the
caller. For reclaim, this is so the user can use, for example, a
statically allocated vector. For copy, it assumes the user
provides the space into which the copy is stored.

The second two, nc_reclaim_data_all() and
nc_copy_data_all(), allows the functions to manage the
top-level.  So for nc_reclaim_data_all, the top level is
assumed to be dynamically allocated and will be free'd by
nc_reclaim_data_all().  The nc_copy_data_all() function
will allocate the top level and return a pointer to it to the
user. The user can later pass that pointer to
nc_reclaim_data_all() to reclaim the instance(s).

# Internal Changes
The netcdf-c library internals are changed to use the proper
reclaim and copy functions.  It turns out that the places where
these functions are needed is quite pervasive in the netcdf-c
library code.  Using these functions also allows some
simplification of the code since the stdata and vldata fields of
NC_ATT_INFO are no longer needed.  Currently this is commented
out using the SEPDATA \#define macro.  When any bugs are largely
fixed, all this code will be removed.

# Known Bugs

1. There is still one known failure that has not been solved.
   All the failures revolve around some variant of this .cdl file.
   The proximate cause of failure is the use of a VLEN FillValue.
````
        netcdf x {
        types:
          float(*) row_of_floats ;
        dimensions:
          m = 5 ;
        variables:
          row_of_floats ragged_array(m) ;
              row_of_floats ragged_array:_FillValue = {-999} ;
        data:
          ragged_array = {10, 11, 12, 13, 14}, {20, 21, 22, 23}, {30, 31, 32},
                         {40, 41}, _ ;
        }
````
When a solution is found, I will either add it to this PR or post a new PR.

# Related Changes

* Mark nc_free_vlen(s) as deprecated in favor of ncaux_reclaim_data.
* Remove the --enable-unfixed-memory-leaks option.
* Remove the NC_VLENS_NOTEST code that suppresses some vlen tests.
* Document this change in docs/internal.md
* Disable the tst_vlen_data test in ncdump/tst_nccopy4.sh.
* Mark types as fixed size or not (transitively) to optimize the reclaim
  and copy functions.

# Misc. Changes

* Make Doxygen process libdispatch/daux.c
* Make sure the NC_ATT_INFO_T.container field is set.
2022-01-08 18:30:00 -07:00

565 lines
16 KiB
C

/*********************************************************************
* Copyright 2018, UCAR/Unidata
* See netcdf/COPYRIGHT file for copying and redistribution conditions.
*********************************************************************/
#include "ncdispatch.h"
#include "ncd4dispatch.h"
#include "d4includes.h"
#include "d4read.h"
#include "d4curlfunctions.h"
#ifdef _MSC_VER
#include <process.h>
#include <direct.h>
#endif
#ifdef HAVE_SYS_STAT_H
#include <sys/stat.h>
#endif
/**************************************************/
/* Forward */
static void applyclientmetacontrols(NCD4meta* meta);
static int constrainable(NCURI*);
static void freeCurl(NCD4curl*);
static void freeInfo(NCD4INFO*);
static int paramcheck(NCD4INFO*, const char* key, const char* subkey);
static const char* getparam(NCD4INFO* info, const char* key);
static int set_curl_properties(NCD4INFO*);
static int makesubstrate(NCD4INFO* d4info);
static void resetInfoforRead(NCD4INFO* d4info);
/**************************************************/
/* Constants */
static const char* checkseps = "+,:;";
/**************************************************/
int
NCD4_open(const char * path, int mode,
int basepe, size_t *chunksizehintp,
void *mpidata, const NC_Dispatch *dispatch, int ncid)
{
int ret = NC_NOERR;
NCD4INFO* d4info = NULL;
const char* value;
NC* nc;
NCD4meta* meta = NULL;
size_t len = 0;
void* contents = NULL;
if(path == NULL)
return THROW(NC_EDAPURL);
assert(dispatch != NULL);
/* Find pointer to NC struct for this file. */
ret = NC_check_id(ncid,&nc);
if(ret != NC_NOERR) {goto done;}
/* Setup our NC and NCDAPCOMMON state*/
d4info = (NCD4INFO*)calloc(1,sizeof(NCD4INFO));
if(d4info == NULL) {ret = NC_ENOMEM; goto done;}
nc->dispatchdata = d4info;
nc->int_ncid = nc__pseudofd(); /* create a unique id */
d4info->controller = (NC*)nc;
/* Parse url and params */
if(ncuriparse(nc->path,&d4info->uri))
{ret = NC_EDAPURL; goto done;}
/* Load auth info from rc file */
if((ret = NC_authsetup(&d4info->auth, d4info->uri)))
goto done;
NCD4_curl_protocols(d4info);
if(!constrainable(d4info->uri))
SETFLAG(d4info->controls.flags,NCF_UNCONSTRAINABLE);
/* fail if we are unconstrainable but have constraints */
if(FLAGSET(d4info->controls.flags,NCF_UNCONSTRAINABLE)) {
if(d4info->uri != NULL && d4info->uri->query != NULL) {
nclog(NCLOGWARN,"Attempt to constrain an unconstrainable data source: %s",
d4info->uri->query);
ret = THROW(NC_EDAPCONSTRAINT);
goto done;
}
}
/* process control client parameters */
NCD4_applyclientparamcontrols(d4info);
/* Use libsrc4 code (netcdf-4) for storing metadata */
{
char tmpname[NC_MAX_NAME];
/* Create fake file name: exact name must be unique,
but is otherwise irrelevant because we are using NC_DISKLESS
*/
if(strlen(d4info->controls.substratename) > 0)
snprintf(tmpname,sizeof(tmpname),"%s",d4info->controls.substratename);
else
snprintf(tmpname,sizeof(tmpname),"tmp_%d",nc->int_ncid);
/* Compute the relevant names for the substrate file */
d4info->substrate.filename = strdup(tmpname);
if(d4info->substrate.filename == NULL)
{ret = NC_ENOMEM; goto done;}
}
/* Turn on logging; only do this after oc_open*/
if((value = ncurifragmentlookup(d4info->uri,"log")) != NULL) {
ncloginit();
ncsetlogging(1);
}
/* Check env values */
if(getenv("CURLOPT_VERBOSE") != NULL)
d4info->auth->curlflags.verbose = 1;
/* Setup a curl connection */
{
CURL* curl = NULL; /* curl handle*/
d4info->curl = (NCD4curl*)calloc(1,sizeof(NCD4curl));
if(d4info->curl == NULL)
{ret = NC_ENOMEM; goto done;}
/* create the connection */
if((ret=NCD4_curlopen(&curl))!= NC_NOERR) goto done;
d4info->curl->curl = curl;
/* Load misc rc properties */
NCD4_get_rcproperties(d4info);
if((ret=set_curl_properties(d4info))!= NC_NOERR) goto done;
/* Set the one-time curl flags */
if((ret=NCD4_set_flags_perlink(d4info))!= NC_NOERR) goto done;
#if 1 /* temporarily make per-link */
if((ret=NCD4_set_flags_perfetch(d4info))!= NC_NOERR) goto done;
#endif
}
d4info->curl->packet = ncbytesnew();
ncbytessetalloc(d4info->curl->packet,DFALTPACKETSIZE); /*initial reasonable size*/
/* Reset the substrate */
if((ret=makesubstrate(d4info))) goto done;
/* Always start by reading the DMR only */
/* reclaim substrate.metadata */
resetInfoforRead(d4info);
/* Rebuild metadata */
if((d4info->substrate.metadata=NCD4_newmeta(d4info))==NULL)
{ret = NC_ENOMEM; goto done;}
if((ret=NCD4_readDMR(d4info, d4info->controls.flags.flags))) goto done;
/* set serial.rawdata */
len = ncbyteslength(d4info->curl->packet);
contents = ncbytesextract(d4info->curl->packet);
NCD4_attachraw(d4info->substrate.metadata, len, contents);
meta = d4info->substrate.metadata;
/* process meta control parameters */
applyclientmetacontrols(meta);
/* Infer the mode */
if((ret=NCD4_infermode(meta))) goto done;
#ifdef D4DUMPDMR
{
fprintf(stderr,"=============\n");
fputs(d4info->substrate.metadata->serial.dmr,stderr);
fprintf(stderr,"\n=============\n");
fflush(stderr);
}
#endif
/* Process the dmr part */
if((ret=NCD4_dechunk(meta))) goto done;
if((ret = NCD4_parse(d4info->substrate.metadata))) goto done;
#ifdef D4DEBUGMETA
{
fprintf(stderr,"\n/////////////\n");
NCbytes* buf = ncbytesnew();
NCD4_print(d4info->substrate.metadata,buf);
ncbytesnull(buf);
fputs(ncbytescontents(buf),stderr);
ncbytesfree(buf);
fprintf(stderr,"\n/////////////\n");
fflush(stderr);
}
#endif
/* Build the substrate metadata */
ret = NCD4_metabuild(d4info->substrate.metadata,d4info->substrate.metadata->ncid);
if(ret != NC_NOERR && ret != NC_EVARSIZE) goto done;
done:
if(ret) {
freeInfo(d4info);
nc->dispatchdata = NULL;
}
return THROW(ret);
}
int
NCD4_close(int ncid, void* ignore)
{
int ret = NC_NOERR;
NC* nc;
NCD4INFO* d4info;
int substrateid;
ret = NC_check_id(ncid, (NC**)&nc);
if(ret != NC_NOERR) goto done;
d4info = (NCD4INFO*)nc->dispatchdata;
substrateid = makenc4id(nc,ncid);
/* We call abort rather than close to avoid trying to write anything,
except if we are debugging
*/
if(FLAGSET(d4info->controls.debugflags,NCF_DEBUG_COPY)) {
/* Dump the data into the substrate */
if((ret = NCD4_debugcopy(d4info)))
goto done;
ret = nc_close(substrateid);
} else {
ret = nc_abort(substrateid);
}
freeInfo(d4info);
done:
return THROW(ret);
}
int
NCD4_abort(int ncid)
{
return NCD4_close(ncid,NULL);
}
/**************************************************/
/* Reclaim an NCD4INFO instance */
static void
freeInfo(NCD4INFO* d4info)
{
if(d4info == NULL) return;
d4info->controller = NULL; /* break link */
nullfree(d4info->rawurltext);
nullfree(d4info->urltext);
ncurifree(d4info->uri);
freeCurl(d4info->curl);
nullfree(d4info->data.memory);
nullfree(d4info->data.ondiskfilename);
if(d4info->data.ondiskfile != NULL)
fclose(d4info->data.ondiskfile);
nullfree(d4info->fileproto.filename);
if(d4info->substrate.realfile
&& !FLAGSET(d4info->controls.debugflags,NCF_DEBUG_COPY)) {
/* We used real file, so we need to delete the temp file
unless we are debugging.
Assume caller has done nc_close|nc_abort on the ncid.
Note that in theory, this should not be necessary since
AFAIK the substrate file is still in def mode, and
when aborted, it should be deleted. But that is not working
for some reason, so we delete it ourselves.
*/
#if 0
if(d4info->substrate.filename != NULL) {
unlink(d4info->substrate.filename);
}
#endif
}
nullfree(d4info->substrate.filename); /* always reclaim */
NCD4_reclaimMeta(d4info->substrate.metadata);
NC_authfree(d4info->auth);
nclistfree(d4info->blobs);
free(d4info);
}
/* Reset NCD4INFO instance for new read request */
static void
resetInfoforRead(NCD4INFO* d4info)
{
if(d4info == NULL) return;
if(d4info->substrate.realfile
&& !FLAGSET(d4info->controls.debugflags,NCF_DEBUG_COPY)) {
/* We used real file, so we need to delete the temp file
unless we are debugging.
Assume caller has done nc_close|nc_abort on the ncid.
Note that in theory, this should not be necessary since
AFAIK the substrate file is still in def mode, and
when aborted, it should be deleted. But that is not working
for some reason, so we delete it ourselves.
*/
if(d4info->substrate.filename != NULL) {
unlink(d4info->substrate.filename);
}
}
NCD4_resetMeta(d4info->substrate.metadata);
nullfree(d4info->substrate.metadata);
d4info->substrate.metadata = NULL;
}
static void
freeCurl(NCD4curl* curl)
{
if(curl == NULL) return;
NCD4_curlclose(curl->curl);
ncbytesfree(curl->packet);
nullfree(curl->errdata.code);
nullfree(curl->errdata.message);
free(curl);
}
/* Define the set of protocols known to be constrainable */
static const char* constrainableprotocols[] = {"http", "https",NULL};
static int
constrainable(NCURI* durl)
{
const char** protocol = constrainableprotocols;
for(;*protocol;protocol++) {
if(strcmp(durl->protocol,*protocol)==0)
return 1;
}
return 0;
}
/*
Set curl properties for link based on rc files etc.
*/
static int
set_curl_properties(NCD4INFO* d4info)
{
int ret = NC_NOERR;
if(d4info->auth->curlflags.useragent == NULL) {
char* agent;
size_t len = strlen(DFALTUSERAGENT) + strlen(VERSION);
len++; /*strlcat nul*/
agent = (char*)malloc(len+1);
strncpy(agent,DFALTUSERAGENT,len);
strlcat(agent,VERSION,len);
d4info->auth->curlflags.useragent = agent;
}
/* Some servers (e.g. thredds and columbia) appear to require a place
to put cookies in order for some security functions to work
*/
if(d4info->auth->curlflags.cookiejar != NULL
&& strlen(d4info->auth->curlflags.cookiejar) == 0) {
free(d4info->auth->curlflags.cookiejar);
d4info->auth->curlflags.cookiejar = NULL;
}
if(d4info->auth->curlflags.cookiejar == NULL) {
/* If no cookie file was defined, define a default */
char* path = NULL;
char* newpath = NULL;
int len;
errno = 0;
NCRCglobalstate* globalstate = ncrc_getglobalstate();
/* Create the unique cookie file name */
len =
strlen(globalstate->tempdir)
+ 1 /* '/' */
+ strlen("ncd4cookies");
path = (char*)malloc(len+1);
if(path == NULL) return NC_ENOMEM;
snprintf(path,len,"%s/nc4cookies",globalstate->tempdir);
/* Create the unique cookie file name */
newpath = NC_mktmp(path);
free(path);
if(newpath == NULL) {
fprintf(stderr,"Cannot create cookie file\n");
goto fail;
}
d4info->auth->curlflags.cookiejar = newpath;
d4info->auth->curlflags.cookiejarcreated = 1;
errno = 0;
}
assert(d4info->auth->curlflags.cookiejar != NULL);
/* Make sure the cookie jar exists and can be read and written */
{
FILE* f = NULL;
char* fname = d4info->auth->curlflags.cookiejar;
/* See if the file exists already */
f = fopen(fname,"r");
if(f == NULL) {
/* Ok, create it */
f = fopen(fname,"w+");
if(f == NULL) {
fprintf(stderr,"Cookie file cannot be read and written: %s\n",fname);
{ret= NC_EPERM; goto fail;}
}
} else { /* test if file can be written */
fclose(f);
f = fopen(fname,"r+");
if(f == NULL) {
fprintf(stderr,"Cookie file is cannot be written: %s\n",fname);
{ret = NC_EPERM; goto fail;}
}
}
if(f != NULL) fclose(f);
}
return THROW(ret);
fail:
return THROW(ret);
}
void
NCD4_applyclientparamcontrols(NCD4INFO* info)
{
const char* value;
/* clear the flags */
CLRFLAG(info->controls.flags,NCF_CACHE);
CLRFLAG(info->controls.flags,NCF_SHOWFETCH);
CLRFLAG(info->controls.flags,NCF_NC4);
CLRFLAG(info->controls.flags,NCF_NCDAP);
CLRFLAG(info->controls.flags,NCF_FILLMISMATCH);
/* Turn on any default on flags */
SETFLAG(info->controls.flags,DFALT_ON_FLAGS);
SETFLAG(info->controls.flags,(NCF_NC4|NCF_NCDAP));
if(paramcheck(info,"show","fetch"))
SETFLAG(info->controls.flags,NCF_SHOWFETCH);
if(paramcheck(info,"translate","nc4"))
info->controls.translation = NCD4_TRANSNC4;
/* Look at the debug flags */
if(paramcheck(info,"debug","copy"))
SETFLAG(info->controls.debugflags,NCF_DEBUG_COPY); /* => close */
value = getparam(info,"substratename");
if(value != NULL)
strncpy(info->controls.substratename,value,(NC_MAX_NAME-1));
info->controls.opaquesize = DFALTOPAQUESIZE;
value = getparam(info,"opaquesize");
if(value != NULL) {
long long len = 0;
if(sscanf(value,"%lld",&len) != 1 || len == 0)
nclog(NCLOGWARN,"bad [opaquesize] tag: %s",value);
else
info->controls.opaquesize = (size_t)len;
}
value = getparam(info,"fillmismatch");
if(value != NULL)
SETFLAG(info->controls.flags,NCF_FILLMISMATCH);
value = getparam(info,"nofillmismatch");
if(value != NULL)
CLRFLAG(info->controls.debugflags,NCF_FILLMISMATCH);
}
static void
applyclientmetacontrols(NCD4meta* meta)
{
NCD4INFO* info = meta->controller;
const char* value = getparam(info,"checksummode");
if(value != NULL) {
if(strcmp(value,"ignore")==0)
meta->ignorechecksums = 1;
}
}
/* Search for substring in value of param. If substring == NULL; then just
check if param is defined.
*/
static int
paramcheck(NCD4INFO* info, const char* key, const char* subkey)
{
const char* value;
char* p;
value = getparam(info, key);
if(value == NULL)
return 0;
if(subkey == NULL) return 1;
p = strstr(value,subkey);
if(p == NULL) return 0;
p += strlen(subkey);
if(*p != '\0' && strchr(checkseps,*p) == NULL) return 0;
return 1;
}
/*
Given a parameter key, return its value or NULL if not defined.
*/
static const char*
getparam(NCD4INFO* info, const char* key)
{
const char* value;
if(info == NULL || key == NULL) return NULL;
if((value=ncurifragmentlookup(info->uri,key)) == NULL)
return NULL;
return value;
}
/**************************************************/
static int
makesubstrate(NCD4INFO* d4info)
{
int ret = NC_NOERR;
int new = NC_NETCDF4;
int old = 0;
int ncid = 0;
int ncflags = NC_NETCDF4|NC_CLOBBER;
if(d4info->substrate.nc4id != 0) {
/* reset the substrate */
nc_abort(d4info->substrate.nc4id);
d4info->substrate.nc4id = 0;
}
/* Create the hidden substrate netcdf file.
We want this hidden file to always be NC_NETCDF4, so we need to
force default format temporarily in case user changed it.
Since diskless is enabled, create file in-memory.
*/
ncflags |= NC_DISKLESS;
if(FLAGSET(d4info->controls.debugflags,NCF_DEBUG_COPY)) {
/* Cause data to be dumped to real file */
ncflags |= NC_WRITE;
ncflags &= ~(NC_DISKLESS); /* use real file */
}
nc_set_default_format(new,&old); /* save and change */
ret = nc_create(d4info->substrate.filename,ncflags,&ncid);
nc_set_default_format(old,&new); /* restore */
/* Avoid fill on the substrate */
nc_set_fill(ncid,NC_NOFILL,NULL);
d4info->substrate.nc4id = ncid;
return THROW(ret);
}
int
NCD4_get_substrate(int ncid)
{
NC* nc = NULL;
NCD4INFO* d4 = NULL;
int subncid = 0;
/* Find pointer to NC struct for this file. */
(void)NC_check_id(ncid,&nc);
d4 = (NCD4INFO*)nc->dispatchdata;
subncid = d4->substrate.nc4id;
return subncid;
}