nuttx-apps/examples/fdpicxip/modules/callback.c
Marco Casaroli e72e737ef2 examples/fdpicxip: Demonstrate FDPIC modules executed in place.
The FDPIC counterpart of examples/nxflatxip, kept separate from it: that
example is about NXFLAT, and mixing the two module formats into one program
would leave neither demonstrating anything clearly.

Four subcommands, each showing one property of the loader.  'qsort' is the
simplest case it has -- one self-contained module, two concurrent instances,
one shared copy of the text in flash and a private copy of the data each,
with the pin count showing the extent held in place while they run and
released after.  'solib' adds a shared library, so two objects are mapped
and each instance still gets its own copy of both objects' data.  'cxx' is
the same in C++, which additionally requires global constructors to have run
in dependency order before main.  'jmprel' is a module whose imports are all
in DT_JMPREL rather than DT_REL.

The modules are embedded as headers rather than built as part of the app:
linking one needs arm-uclinuxfdpiceabi binutils, which the tree does not
require, so the headers are committed and both this example and
testing/fs/xipfs build with a plain toolchain.

Their sources are in modules/, with a makefile that rebuilds every header
from them on an explicit 'make regen NUTTX_DIR=...' and is never invoked by
the application build.  It drives nuttx/tools/fdpic and writes the headers
this example needs alongside the ones testing/fs/xipfs needs, so one source
tree serves both and the two cannot drift apart.  CPU is cortex-m3 there
deliberately: a v7-M module runs on the v8-M targets too, so one set of blobs
serves both the RP2350 and mps2-an500, while a cortex-m33 build produces
blobs the Cortex-M7 cannot execute at all.

This demonstrates; it does not assert.  The assertions are in the fdpic and
reject sections of apps/testing/fs/xipfs.

Verified on a Pimoroni Pico Plus 2: all four subcommands, and nxflatxip
unaffected alongside them.  Also verified on mps2-an500 under QEMU, which is
a Cortex-M7 -- a different core generation from the RP2350's Cortex-M33.

Assisted-by: Claude Code:claude-opus-5
Signed-off-by: Marco Casaroli <marco.casaroli@gmail.com>
2026-07-30 14:05:01 +02:00

517 lines
13 KiB
C

/****************************************************************************
* apps/examples/fdpicxip/modules/callback.c
*
* SPDX-License-Identifier: Apache-2.0
*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership. The
* ASF licenses this file to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance with the
* License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
****************************************************************************/
/* Hands a module function to the firmware every way the firmware will
* accept one.
*
* In FDPIC a function pointer is the address of a two-word descriptor, not
* a code address, and that descriptor lives in the module's writable
* segment in RAM. Firmware that stores such a pointer and later branches to
* it therefore jumps into the module's *data* and faults -- unless the entry
* point it came through resolved the descriptor first.
*
* qsort and bsearch have always done that. pthread_create, signal and
* task_create did not, which made them look usable and then take the board
* down; pthread_once, task_spawn, scandir's filter and sigaction followed
* later. This module is what keeps all of them honest.
*
* scandir is one worth noting: its filter is called by scandir itself and
* must be resolved there, while its comparison function is handed to qsort,
* which resolves it. Resolving both would be a bug, so both are passed here
* at once.
*
* mq_notify and timer_create with SIGEV_THREAD are the interesting ones:
* their callback runs on a work-queue worker that carries no module data
* base. The base is captured at registration and installed around the call,
* so the callback still reaches the module's own globals. Each is checked
* by having the callback write a distinctive value into a module global.
*
* Everything is checked by observing a side effect the callback itself
* produces, because a callback that never runs is the failure being tested
* for -- and on this target the alternative failure is a HardFault, which
* reports nothing at all.
*/
/****************************************************************************
* Included Files
****************************************************************************/
#include <dirent.h>
#include <fcntl.h>
#include <mqueue.h>
#include <pthread.h>
#include <sched.h>
#include <signal.h>
#include <spawn.h>
#include <stdlib.h>
#include <string.h>
#include <syslog.h>
#include <time.h>
#include <unistd.h>
/****************************************************************************
* Pre-processor Definitions
****************************************************************************/
#define FAIL_QSORT 0x01
#define FAIL_PTHREAD 0x02
#define FAIL_SIGNAL 0x04
#define FAIL_TASK 0x08
#define FAIL_ONCE 0x10
#define FAIL_SPAWN 0x20
#define FAIL_SCANDIR 0x40
#define FAIL_SIGACTION 0x80
/* The mq checks run under "callback mq" and report in their own byte, so the
* exit status -- only eight bits survive waitpid -- never has to hold more
* than eight flags at once.
*/
#define FAIL_MQ_THREAD 0x01
#define FAIL_MQ_SIGNAL 0x02
#define FAIL_TIMER_THREAD 0x04
/****************************************************************************
* Private Data
****************************************************************************/
/* Written by the callbacks, read by main. These live in the module's
* writable segment, which is what the callback can only reach if it was
* entered with this module's data base in the FDPIC register.
*/
static volatile int g_qsort_calls;
static volatile int g_thread_ran;
static volatile int g_signal_ran;
static volatile int g_task_ran;
static volatile int g_once_ran;
static volatile int g_spawn_ran;
static volatile int g_filter_calls;
static volatile int g_compar_calls;
static volatile int g_sigaction_ran;
static volatile int g_mq_signal_ran;
static volatile int g_mq_thread_ran;
static volatile int g_timer_thread_ran;
/****************************************************************************
* Private Functions
****************************************************************************/
static int cmp_int(const void *a, const void *b)
{
g_qsort_calls++;
return *(const int *)a - *(const int *)b;
}
static void *thread_main(void *arg)
{
g_thread_ran = (int)(intptr_t)arg;
return NULL;
}
static void sig_handler(int signo)
{
g_signal_ran = signo;
}
static void sigaction_handler(int signo, siginfo_t *info, void *context)
{
g_sigaction_ran = signo;
}
static void mq_signal_handler(int signo, siginfo_t *info, void *context)
{
g_mq_signal_ran = signo;
}
static void mq_thread_notify(union sigval value)
{
/* Runs on a work-queue worker, which carries no module data base of its
* own. Writing this module global proves the base was installed around
* the call -- without it this store would fault or land nowhere.
*/
g_mq_thread_ran = value.sival_int;
}
static void timer_thread_notify(union sigval value)
{
g_timer_thread_ran = value.sival_int;
}
static int task_main(int argc, char *argv[])
{
g_task_ran = 1;
return 0;
}
static void once_routine(void)
{
g_once_ran++;
}
static int spawn_main(int argc, char *argv[])
{
g_spawn_ran = 1;
return 0;
}
static int dir_filter(const struct dirent *entry)
{
g_filter_calls++;
return 1;
}
static int dir_compar(const struct dirent **a, const struct dirent **b)
{
g_compar_calls++;
return strcmp((*a)->d_name, (*b)->d_name);
}
/* The message-queue checks run as a second invocation, "callback mq", so
* their two flags report in a byte of their own rather than overflowing the
* eight-bit exit status the other checks already fill.
*/
static int run_mq(void)
{
struct mq_attr attr;
struct sigaction sa;
struct sigevent ev;
struct itimerspec its;
timer_t timer;
mqd_t mqd;
int fails = 0;
int i;
attr.mq_maxmsg = 1;
attr.mq_msgsize = 1;
attr.mq_flags = 0;
mqd = mq_open("/cbmq", O_CREAT | O_RDWR, 0644, &attr);
if (mqd == (mqd_t)-1)
{
return FAIL_MQ_THREAD | FAIL_MQ_SIGNAL | FAIL_TIMER_THREAD;
}
/* mq_notify with SIGEV_THREAD. The callback runs on a work-queue worker
* that carries no module data base of its own; the base captured at
* registration is installed around the call, so the callback reaches this
* module's globals. A distinctive sival marks that the right callback
* ran with the right argument.
*/
memset(&ev, 0, sizeof(ev));
ev.sigev_notify = SIGEV_THREAD;
ev.sigev_notify_function = mq_thread_notify;
ev.sigev_value.sival_int = 0x5a;
if (mq_notify(mqd, &ev) < 0)
{
fails |= FAIL_MQ_THREAD;
}
else
{
char buf[1];
/* The notification fires on a send to an empty queue; drain the
* message afterwards so the queue is empty for the next test and the
* next send does not block on a full queue.
*/
mq_send(mqd, "x", 1, 0);
for (i = 0; i < 50 && g_mq_thread_ran == 0; i++)
{
usleep(10000);
}
mq_receive(mqd, buf, sizeof(buf), NULL);
if (g_mq_thread_ran != 0x5a)
{
fails |= FAIL_MQ_THREAD;
}
}
/* mq_notify with SIGEV_SIGNAL, delivered to this task and caught by a
* handler installed through sigaction.
*/
memset(&sa, 0, sizeof(sa));
sa.sa_sigaction = mq_signal_handler;
sa.sa_flags = SA_SIGINFO;
sigaction(SIGUSR1, &sa, NULL);
memset(&ev, 0, sizeof(ev));
ev.sigev_notify = SIGEV_SIGNAL;
ev.sigev_signo = SIGUSR1;
if (mq_notify(mqd, &ev) < 0)
{
fails |= FAIL_MQ_SIGNAL;
}
else
{
char buf[1];
mq_send(mqd, "x", 1, 0);
for (i = 0; i < 50 && g_mq_signal_ran == 0; i++)
{
usleep(10000);
}
mq_receive(mqd, buf, sizeof(buf), NULL);
if (g_mq_signal_ran != SIGUSR1)
{
fails |= FAIL_MQ_SIGNAL;
}
}
mq_close(mqd);
mq_unlink("/cbmq");
/* timer_create with SIGEV_THREAD reaches the module the same way: the
* expiry callback runs on the worker with the module's base installed.
*/
memset(&ev, 0, sizeof(ev));
ev.sigev_notify = SIGEV_THREAD;
ev.sigev_notify_function = timer_thread_notify;
ev.sigev_value.sival_int = 0xa5;
if (timer_create(CLOCK_MONOTONIC, &ev, &timer) < 0)
{
fails |= FAIL_TIMER_THREAD;
}
else
{
its.it_value.tv_sec = 0;
its.it_value.tv_nsec = 20 * 1000 * 1000;
its.it_interval.tv_sec = 0;
its.it_interval.tv_nsec = 0;
timer_settime(timer, 0, &its, NULL);
for (i = 0; i < 50 && g_timer_thread_ran == 0; i++)
{
usleep(10000);
}
if (g_timer_thread_ran != 0xa5)
{
fails |= FAIL_TIMER_THREAD;
}
timer_delete(timer);
}
syslog(LOG_INFO,
"[callback] mq_thread=%#x mq_signal=%d timer_thread=%#x -- %s\n",
g_mq_thread_ran, g_mq_signal_ran, g_timer_thread_ran,
fails == 0 ? "PASS" : "FAIL");
return fails;
}
/****************************************************************************
* Public Functions
****************************************************************************/
int main(int argc, char *argv[])
{
int v[5] =
{
5, 3, 1, 4, 2
};
struct dirent **namelist;
pthread_once_t once = PTHREAD_ONCE_INIT;
struct sigaction sa;
pthread_t tid;
int fails = 0;
int n;
int i;
if (argc > 1 && strcmp(argv[1], "mq") == 0)
{
return run_mq();
}
/* qsort -- the path that always worked */
qsort(v, 5, sizeof(int), cmp_int);
if (g_qsort_calls == 0 || v[0] != 1 || v[4] != 5)
{
fails |= FAIL_QSORT;
}
/* pthread_create. The thread inherits this module's D-Space, so it can
* reach g_thread_ran only if it was entered at the right address.
*/
if (pthread_create(&tid, NULL, thread_main, (void *)(intptr_t)7) != 0)
{
fails |= FAIL_PTHREAD;
}
else
{
pthread_join(tid, NULL);
if (g_thread_ran != 7)
{
fails |= FAIL_PTHREAD;
}
}
/* signal, delivered to this task */
if (signal(SIGUSR1, sig_handler) == SIG_ERR)
{
fails |= FAIL_SIGNAL;
}
else
{
/* raise() rather than kill(getpid(), ...): getpid is not in the
* firmware's export list, and raise targets the calling thread, which
* is exactly what this is checking.
*/
raise(SIGUSR1);
for (i = 0; i < 50 && g_signal_ran == 0; i++)
{
usleep(10000);
}
if (g_signal_ran != SIGUSR1)
{
fails |= FAIL_SIGNAL;
}
}
/* task_create. A separate task, but the same D-Space. */
if (task_create("cbtask", SCHED_PRIORITY_DEFAULT, 2048, task_main,
NULL) < 0)
{
fails |= FAIL_TASK;
}
else
{
for (i = 0; i < 50 && g_task_ran == 0; i++)
{
usleep(10000);
}
if (g_task_ran == 0)
{
fails |= FAIL_TASK;
}
}
/* sigaction, delivered to this task. signal() above resolves the handler
* before calling sigaction; this reaches sigaction directly, which is the
* path signal() does not cover.
*/
memset(&sa, 0, sizeof(sa));
sa.sa_sigaction = sigaction_handler;
sa.sa_flags = SA_SIGINFO;
if (sigaction(SIGUSR2, &sa, NULL) < 0)
{
fails |= FAIL_SIGACTION;
}
else
{
raise(SIGUSR2);
for (i = 0; i < 50 && g_sigaction_ran == 0; i++)
{
usleep(10000);
}
if (g_sigaction_ran != SIGUSR2)
{
fails |= FAIL_SIGACTION;
}
}
/* pthread_once. Runs on this thread, so only the code address matters. */
if (pthread_once(&once, once_routine) != 0 || g_once_ran != 1)
{
fails |= FAIL_ONCE;
}
/* task_spawn. Like task_create, a separate task with the same D-Space. */
if (task_spawn("cbspawn", spawn_main, NULL, NULL, NULL, NULL) < 0)
{
fails |= FAIL_SPAWN;
}
else
{
for (i = 0; i < 50 && g_spawn_ran == 0; i++)
{
usleep(10000);
}
if (g_spawn_ran == 0)
{
fails |= FAIL_SPAWN;
}
}
/* scandir takes two callbacks that reach the module by different routes:
* the filter is called by scandir itself, while the comparison function is
* handed to qsort. Passing both at once is the point -- it is what would
* catch the comparison function being resolved twice.
*/
namelist = NULL;
n = scandir("/mnt/xipfs", &namelist, dir_filter, dir_compar);
if (n < 0 || g_filter_calls == 0)
{
fails |= FAIL_SCANDIR;
}
if (namelist != NULL)
{
while (n-- > 0)
{
free(namelist[n]);
}
free(namelist);
}
syslog(LOG_INFO,
"[callback] qsort=%d pthread=%d signal=%d task=%d sigaction=%d "
"once=%d spawn=%d filter=%d compar=%d -- %s\n",
g_qsort_calls, g_thread_ran, g_signal_ran, g_task_ran,
g_sigaction_ran, g_once_ran, g_spawn_ran, g_filter_calls,
g_compar_calls, fails == 0 ? "PASS" : "FAIL");
return fails;
}