From 4e8a3561652ebcda6a126b3162fc545573889dc4 Mon Sep 17 00:00:00 2001
From: Rich Felker <dalias@aerifal.cx>
Date: Fri, 13 Feb 2015 00:27:45 -0500
Subject: overhaul aio implementation for correctness

previously, aio operations were not tracked by file descriptor; each
operation was completely independent. this resulted in non-conforming
behavior for non-seekable/append-mode writes (which are required to be
ordered) and made it impossible to implement aio_cancel, which in turn
made closing file descriptors with outstanding aio operations unsafe.

the new implementation is significantly heavier (roughly twice the
size, and seems to be slightly slower) and presently aims mainly at
correctness, not performance.

most of the public interfaces have been moved into a single file,
aio.c, because there is little benefit to be had from splitting them.
whenever any aio functions are used, aio_cancel and the internal
queue lifetime management and fd-to-queue mapping code must be linked,
and these functions make up the bulk of the code size.

the close function's interaction with aio is implemented with weak
alias magic, to avoid pulling in heavy aio cancellation code in
programs that don't use aio, and the expensive cancellation path
(which includes signal blocking) is optimized out when there are no
active aio queues.
---
 src/aio/aio_readwrite.c | 110 ------------------------------------------------
 1 file changed, 110 deletions(-)
 delete mode 100644 src/aio/aio_readwrite.c

(limited to 'src/aio/aio_readwrite.c')

diff --git a/src/aio/aio_readwrite.c b/src/aio/aio_readwrite.c
deleted file mode 100644
index 8753ffda..00000000
--- a/src/aio/aio_readwrite.c
+++ /dev/null
@@ -1,110 +0,0 @@
-#include <aio.h>
-#include <fcntl.h>
-#include <errno.h>
-#include <unistd.h>
-#include "pthread_impl.h"
-#include "libc.h"
-
-static void dummy(void)
-{
-}
-
-weak_alias(dummy, __aio_wake);
-
-static void notify_signal(struct sigevent *sev)
-{
-	siginfo_t si = {
-		.si_signo = sev->sigev_signo,
-		.si_value = sev->sigev_value,
-		.si_code = SI_ASYNCIO,
-		.si_pid = getpid(),
-		.si_uid = getuid()
-	};
-	__syscall(SYS_rt_sigqueueinfo, si.si_pid, si.si_signo, &si);
-}
-
-static void *io_thread(void *p)
-{
-	struct aiocb *cb = p;
-	int fd = cb->aio_fildes;
-	void *buf = (void *)cb->aio_buf;
-	size_t len = cb->aio_nbytes;
-	off_t off = cb->aio_offset;
-	int op = cb->aio_lio_opcode;
-	struct sigevent sev = cb->aio_sigevent;
-	ssize_t ret;
-
-	if (op == LIO_WRITE) {
-		if (  (fcntl(fd, F_GETFL) & O_APPEND)
-		    ||((ret = pwrite(fd, buf, len, off))<0 && errno==ESPIPE) )
-			ret = write(fd, buf, len);
-	} else if (op == LIO_READ) {
-		if ( (ret = pread(fd, buf, len, off))<0 && errno==ESPIPE )
-			ret = read(fd, buf, len);
-	} else {
-		ret = 0;
-	}
-	cb->__ret = ret;
-
-	if (ret < 0) a_store(&cb->__err, errno);
-	else a_store(&cb->__err, 0);
-
-	__aio_wake();
-
-	switch (sev.sigev_notify) {
-	case SIGEV_SIGNAL:
-		notify_signal(&sev);
-		break;
-	case SIGEV_THREAD:
-		sev.sigev_notify_function(sev.sigev_value);
-		break;
-	}
-
-	return 0;
-}
-
-static int new_req(struct aiocb *cb)
-{
-	int ret = 0;
-	pthread_attr_t a;
-	sigset_t set;
-	pthread_t td;
-
-	if (cb->aio_sigevent.sigev_notify == SIGEV_THREAD) {
-		if (cb->aio_sigevent.sigev_notify_attributes)
-			a = *cb->aio_sigevent.sigev_notify_attributes;
-		else
-			pthread_attr_init(&a);
-	} else {
-		pthread_attr_init(&a);
-		pthread_attr_setstacksize(&a, PAGE_SIZE);
-		pthread_attr_setguardsize(&a, 0);
-	}
-	pthread_attr_setdetachstate(&a, PTHREAD_CREATE_DETACHED);
-	sigfillset(&set);
-	pthread_sigmask(SIG_BLOCK, &set, &set);
-	cb->__err = EINPROGRESS;
-	if (pthread_create(&td, &a, io_thread, cb)) {
-		errno = EAGAIN;
-		ret = -1;
-	}
-	pthread_sigmask(SIG_SETMASK, &set, 0);
-	cb->__td = td;
-
-	return ret;
-}
-
-int aio_read(struct aiocb *cb)
-{
-	cb->aio_lio_opcode = LIO_READ;
-	return new_req(cb);
-}
-
-int aio_write(struct aiocb *cb)
-{
-	cb->aio_lio_opcode = LIO_WRITE;
-	return new_req(cb);
-}
-
-LFS64(aio_read);
-LFS64(aio_write);
-- 
cgit v1.2.1