nuttx/fs/vfs/fs_select.c
SPRESENSE 0cc0c3d503 fs/vfs: Fix bug that select() did not return
Fix bug that select() did not return when the nfds argument
was set to a negative value. The specification is that -1 is
set to the return value and EINVAL is set to errno.
2021-05-19 06:43:22 -05:00

280 lines
7.4 KiB
C

/****************************************************************************
* fs/vfs/fs_select.c
*
* 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.
*
****************************************************************************/
/****************************************************************************
* Included Files
****************************************************************************/
#include <nuttx/config.h>
#include <sys/select.h>
#include <sys/time.h>
#include <string.h>
#include <poll.h>
#include <errno.h>
#include <assert.h>
#include <debug.h>
#include <nuttx/kmalloc.h>
#include <nuttx/cancelpt.h>
#include <nuttx/fs/fs.h>
#include "inode/inode.h"
/****************************************************************************
* Public Functions
****************************************************************************/
/****************************************************************************
* Name: select
*
* Description:
* select() allows a program to monitor multiple file descriptors, waiting
* until one or more of the file descriptors become "ready" for some class
* of I/O operation (e.g., input possible). A file descriptor is
* considered ready if it is possible to perform the corresponding I/O
* operation (e.g., read(2)) without blocking.
*
* NOTE: poll() is the fundamental API for performing such monitoring
* operation under NuttX. select() is provided for compatibility and
* is simply a layer of added logic on top of poll(). As such, select()
* is more wasteful of resources and poll() is the recommended API to be
* used.
*
* Input Parameters:
* nfds - the maximum fd number (+1) of any descriptor in any of the
* three sets.
* readfds - the set of descriptions to monitor for read-ready events
* writefds - the set of descriptions to monitor for write-ready events
* exceptfds - the set of descriptions to monitor for error events
* timeout - Return at this time if none of these events of interest
* occur.
*
* Returned Value:
* 0: Timer expired
* >0: The number of bits set in the three sets of descriptors
* -1: An error occurred (errno will be set appropriately)
*
****************************************************************************/
int select(int nfds, FAR fd_set *readfds, FAR fd_set *writefds,
FAR fd_set *exceptfds, FAR struct timeval *timeout)
{
struct pollfd *pollset = NULL;
int errcode = OK;
int fd;
int npfds;
int msec;
int ndx;
int ret;
/* select() is cancellation point */
enter_cancellation_point();
if (nfds < 0)
{
errcode = EINVAL;
goto errout;
}
/* How many pollfd structures do we need to allocate? */
/* Initialize the descriptor list for poll() */
for (fd = 0, npfds = 0; fd < nfds; fd++)
{
/* Check if any monitor operation is requested on this fd */
if ((readfds && FD_ISSET(fd, readfds)) ||
(writefds && FD_ISSET(fd, writefds)) ||
(exceptfds && FD_ISSET(fd, exceptfds)))
{
/* Yes.. increment the count of pollfds structures needed */
npfds++;
}
}
/* Allocate the descriptor list for poll() */
if (npfds > 0)
{
pollset = (FAR struct pollfd *)
kmm_zalloc(npfds * sizeof(struct pollfd));
if (pollset == NULL)
{
errcode = ENOMEM;
goto errout;
}
}
/* Initialize the descriptor list for poll() */
for (fd = 0, ndx = 0; fd < nfds; fd++)
{
int incr = 0;
/* The readfs set holds the set of FDs that the caller can be assured
* of reading from without blocking. Note that POLLHUP is included as
* a read-able condition. POLLHUP will be reported at the end-of-file
* or when a connection is lost. In either case, the read() can then
* be performed without blocking.
*/
if (readfds && FD_ISSET(fd, readfds))
{
pollset[ndx].fd = fd;
pollset[ndx].events |= POLLIN;
incr = 1;
}
/* The writefds set holds the set of FDs that the caller can be assured
* of writing to without blocking.
*/
if (writefds && FD_ISSET(fd, writefds))
{
pollset[ndx].fd = fd;
pollset[ndx].events |= POLLOUT;
incr = 1;
}
/* The exceptfds set holds the set of FDs that are watched for
* exceptions
*/
if (exceptfds && FD_ISSET(fd, exceptfds))
{
pollset[ndx].fd = fd;
incr = 1;
}
ndx += incr;
}
DEBUGASSERT(ndx == npfds);
/* Convert the timeout to milliseconds */
if (timeout)
{
/* Calculate the timeout in milliseconds */
msec = timeout->tv_sec * 1000 + timeout->tv_usec / 1000;
}
else
{
/* Any negative value of msec means no timeout */
msec = -1;
}
/* Then let poll do all of the real work. */
ret = nx_poll(pollset, npfds, msec);
if (ret < 0)
{
/* poll() failed! Save the errno value */
errcode = -ret;
}
/* Now set up the return values */
if (readfds)
{
memset(readfds, 0, sizeof(fd_set));
}
if (writefds)
{
memset(writefds, 0, sizeof(fd_set));
}
if (exceptfds)
{
memset(exceptfds, 0, sizeof(fd_set));
}
/* Convert the poll descriptor list back into selects 3 bitsets */
if (ret > 0)
{
ret = 0;
for (ndx = 0; ndx < npfds; ndx++)
{
/* Check for read conditions. Note that POLLHUP is included as a
* read condition. POLLHUP will be reported when no more data will
* be available (such as when a connection is lost). In either
* case, the read() can then be performed without blocking.
*/
if (readfds)
{
if (pollset[ndx].revents & (POLLIN | POLLHUP))
{
FD_SET(pollset[ndx].fd, readfds);
ret++;
}
}
/* Check for write conditions */
if (writefds)
{
if (pollset[ndx].revents & POLLOUT)
{
FD_SET(pollset[ndx].fd, writefds);
ret++;
}
}
/* Check for exceptions */
if (exceptfds)
{
if (pollset[ndx].revents & POLLERR)
{
FD_SET(pollset[ndx].fd, exceptfds);
ret++;
}
}
}
}
kmm_free(pollset);
/* Did poll() fail above? */
if (ret >= 0)
{
leave_cancellation_point();
return ret;
}
errout:
set_errno(errcode);
leave_cancellation_point();
return ERROR;
}