As follow-up to the rework of thread-handling, likewise also the implementation base for locking was switched over from direct usage of POSIX primitives to the portable wrappers available in the C++ standard library. All usages have been reviewed and modernised to prefer λ-functions where possible. With this series of changes, the old threadpool implementation and a lot of further low-level support facilities are not used any more and can be dismantled. Due to the integration efforts spurred by the »Playback Vertical Slice«, several questions of architecture could be decided over the last months. The design of the Scheduler and Engine turned out different than previously anticipated; notably the Scheduler now covers a wider array of functionality, including some asynchronous messaging. This has ramifications for the organisation of work tasks and threads, and leads to a more deterministic memory management. Resource management will be done on a higher level, partially superseding some of the concepts from the early phase of the Lumiera project.
214 lines
6.1 KiB
C
214 lines
6.1 KiB
C
/*
|
|
safe_clib.c - Portable and safe wrapers around some clib functions and some tools
|
|
|
|
Copyright (C) Lumiera.org
|
|
2008, Christian Thaeter <ct@pipapo.org>
|
|
|
|
This program is free software; you can redistribute it and/or
|
|
modify it under the terms of the GNU General Public License as
|
|
published by the Free Software Foundation; either version 2 of
|
|
the License, or (at your option) any later version.
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
GNU General Public License for more details.
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
along with this program; if not, write to the Free Software
|
|
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
|
|
|
* *****************************************************/
|
|
|
|
|
|
/** @file safeclib.c
|
|
** Implementation of error-safe wrappers for some notorious C-Lib functions.
|
|
*/
|
|
|
|
|
|
#include "lib/error.h"
|
|
#include "lib/safeclib.h"
|
|
|
|
#include <string.h>
|
|
#include <stdlib.h>
|
|
#include <pthread.h>
|
|
#include <stdint.h>
|
|
#include <nobug.h>
|
|
|
|
|
|
LUMIERA_ERROR_DEFINE (NO_MEMORY, "Out of Memory!");
|
|
|
|
|
|
/**
|
|
* Resources known to the resource collector
|
|
* @todo 10/2023 this is a left-over of the »resource-collector« plan
|
|
*/
|
|
enum lumiera_resource
|
|
{
|
|
/** memory blocks, context is a pointer to the size_t required **/
|
|
LUMIERA_RESOURCE_MEMORY,
|
|
// /** OS filehandles **/
|
|
// LUMIERA_RESOURCE_FILEHANDLE,
|
|
// /** CPU time, as in threads and such **/
|
|
// LUMIERA_RESOURCE_CPU,
|
|
// /** mmaped regions **/
|
|
// LUMIERA_RESOURCE_MMAP,
|
|
// /** disk space for the storage area, context is a pointer to the filename indication the device **/
|
|
// LUMIERA_RESOURCE_DISKSTORAGE,
|
|
// /** disk bandwidth for the storage area, context is a pointer to the filename indication the device **/
|
|
// LUMIERA_RESOURCE_STORAGEBANDWIDTH,
|
|
// /** disk space for the caching area, context is a pointer to the filename indication the device **/
|
|
// LUMIERA_RESOURCE_DISKCACHE,
|
|
// /** disk bandwidth for the caching area, context is a pointer to the filename indication the device **/
|
|
// LUMIERA_RESOURCE_CACHEBANDWIDTH,
|
|
|
|
LUMIERA_RESOURCE_END /* last entry */
|
|
};
|
|
|
|
|
|
/**
|
|
* Iteration indicator
|
|
* @todo 10/2023 this is a left-over of the »resource-collector« plan
|
|
* Resource collection works iteratively freeing more and more resources.
|
|
* Handlers do not need to obey the request and shall return LUMIERA_RESOURCE_NONE
|
|
* which will then continue with the next handler.
|
|
* This goes through all available handlers until one returns a higher or same value
|
|
* than the current iteration to indicate that it freed enough resources to continue the task.
|
|
* Then control is passed back to the calling loop which retries the resource allocation.
|
|
* LUMIERA_RESOURCE_PANIC is somewhat special since it will always call all registered handlers
|
|
* for all resources, not only the queried one and finally _exit() the application.
|
|
* The exact amounts of resources to be freed for ONE, SOME and MANY in intentionally
|
|
* kept vague the handlers are free to interpret this in some sensible way.
|
|
*/
|
|
enum lumiera_resource_try
|
|
{
|
|
/** No op, returned by a handler when it did nothing **/
|
|
LUMIERA_RESOURCE_NONE,
|
|
/** try to free one or really few of this resources **/
|
|
LUMIERA_RESOURCE_ONE,
|
|
/** try to free a small reasonable implementation defined amount of resources **/
|
|
LUMIERA_RESOURCE_SOME,
|
|
/** try to free a bigger implementation defined amount of resources **/
|
|
LUMIERA_RESOURCE_MANY,
|
|
/** free as much as possible **/
|
|
LUMIERA_RESOURCE_ALL,
|
|
/** die! **/
|
|
LUMIERA_RESOURCE_PANIC,
|
|
/** When a handler gets unregistered it will be called with this value to give it a chance to clean up the user 'data' **/
|
|
LUMIERA_RESOURCE_UNREGISTER
|
|
};
|
|
|
|
|
|
/**
|
|
* @internal placeholder function in case the resource-collector was not installed
|
|
* @todo 10/2023 in the early stages of the Lumiera project a centralised resource pooling
|
|
* was considered. On resource shortage, the (planned) resource-collector would attempt
|
|
* to reclaim excess resources. This would allow to use resources (e.g. memory) wastefully
|
|
* as trade-off for performance gains.
|
|
* This concept was never developed beyond an interface draft; most notably it was never
|
|
* clarified how excess resources could have been identified on-demand while in-flight.
|
|
* Today I view this concept as an immature plan and remove the remaining draft code.
|
|
*/
|
|
static int
|
|
die_no_mem (enum lumiera_resource which, enum lumiera_resource_try* iteration, void* context)
|
|
{
|
|
(void) which; (void) iteration; (void) context;
|
|
LUMIERA_DIE (NO_MEMORY);
|
|
return 0; /* not reached */
|
|
}
|
|
|
|
|
|
void*
|
|
lumiera_malloc (size_t size)
|
|
{
|
|
enum lumiera_resource_try iteration = LUMIERA_RESOURCE_ONE;
|
|
void* o = NULL;
|
|
|
|
if (size)
|
|
{
|
|
o = malloc (size);
|
|
if (!o)
|
|
die_no_mem (LUMIERA_RESOURCE_MEMORY, &iteration, &size);
|
|
}
|
|
|
|
return o;
|
|
}
|
|
|
|
|
|
void*
|
|
lumiera_calloc (size_t n, size_t size)
|
|
{
|
|
enum lumiera_resource_try iteration = LUMIERA_RESOURCE_ONE;
|
|
void* o = NULL;
|
|
|
|
size_t gross = n*size;
|
|
|
|
if (n&&size)
|
|
{
|
|
o = calloc (n, size);
|
|
if (!o)
|
|
die_no_mem (LUMIERA_RESOURCE_MEMORY, &iteration, &gross);
|
|
}
|
|
|
|
return o;
|
|
}
|
|
|
|
|
|
void*
|
|
lumiera_realloc (void* ptr, size_t size)
|
|
{
|
|
enum lumiera_resource_try iteration = LUMIERA_RESOURCE_ONE;
|
|
void* o = NULL;
|
|
|
|
if (size)
|
|
{
|
|
o = realloc (ptr, size);
|
|
if (!o)
|
|
die_no_mem (LUMIERA_RESOURCE_MEMORY, &iteration, &size);
|
|
}
|
|
|
|
return o;
|
|
}
|
|
|
|
|
|
char*
|
|
lumiera_strndup (const char* str, size_t len)
|
|
{
|
|
enum lumiera_resource_try iteration = LUMIERA_RESOURCE_ONE;
|
|
void* o = NULL;
|
|
|
|
if (str && len)
|
|
o = strndup (str, len);
|
|
else
|
|
o = strdup ("");
|
|
|
|
if (!o)
|
|
die_no_mem (LUMIERA_RESOURCE_MEMORY, &iteration, &len);
|
|
|
|
return o;
|
|
}
|
|
|
|
|
|
int
|
|
lumiera_strncmp (const char* a, const char* b, size_t len)
|
|
{
|
|
return a == b ? 0 : strncmp (a?a:"", b?b:"", len);
|
|
}
|
|
|
|
|
|
int
|
|
lumiera_streq (const char* a, const char* b)
|
|
{
|
|
return !lumiera_strncmp (a, b, SIZE_MAX);
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
// Local Variables:
|
|
// mode: C
|
|
// c-file-style: "gnu"
|
|
// indent-tabs-mode: nil
|
|
// End:
|
|
*/
|