vblade-17-aio.2.diff

diff -uprN vblade-17.orig/aoe.c vblade-17/aoe.c
--- vblade-17.orig/aoe.c	2008-06-09 10:53:07.000000000 -0400
+++ vblade-17/aoe.c	2008-06-09 11:05:23.000000000 -0400
@@ -8,6 +8,9 @@
 #include <sys/stat.h>
 #include <fcntl.h>
 #include <netinet/in.h>
+#include <errno.h>
+#include <aio.h>
+#include <poll.h>
 #include "dat.h"
 #include "fns.h"
 
@@ -22,6 +25,11 @@ char config[Nconfig];
 int nconfig = 0;
 int maxscnt = 2;
 char *ifname;
+int queuepipe[2];
+int pktlen[Nplaces], pending[Nplaces];
+Ata *pkt[Nplaces];
+Ataregs regs[Nplaces];
+struct aiocb aiocb[Nplaces];
 
 void
 aoead(int fd)			// advertise the virtual blade
@@ -78,32 +86,52 @@ getlba(uchar *p)
 }
 
 int
-aoeata(Ata *p, int pktlen)	// do ATA reqeust
+aoeata(int place)	// do ATA reqeust
 {
-	Ataregs r;
-	int len = 60;
 	int n;
+	int len = 60; // minimum ethernet packet size
 
-	r.lba = getlba(p->lba);
-	r.sectors = p->sectors;
-	r.feature = p->err;
-	r.cmd = p->cmd;
-	if (atacmd(&r, (uchar *)(p+1), maxscnt*512, pktlen - sizeof(*p)) < 0) {
-		p->h.flags |= Error;
-		p->h.error = BadArg;
+	regs[place].lba = getlba(pkt[place]->lba);
+	regs[place].sectors = pkt[place]->sectors;
+	regs[place].feature = pkt[place]->err;
+	regs[place].cmd = pkt[place]->cmd;
+	n = atacmd(regs + place, (uchar *)(pkt[place] + 1), maxscnt*512,
+				pktlen[place] - sizeof(Ata), aiocb + place);
+	if (n < 0) {
+		pkt[place]->h.flags |= Error;
+		pkt[place]->h.error = BadArg;
 		return len;
+	} else if (n > 0) {
+		pending[place] = 1;
+		return 0;
+	}
+	if (!(pkt[place]->aflag & Write) && (n = pkt[place]->sectors)) {
+		n -= regs[place].sectors;
+		len = sizeof (Ata) + (n*512);
 	}
-	if (!(p->aflag & Write))
-	if ((n = p->sectors)) {
-		n -= r.sectors;
+	pkt[place]->sectors = regs[place].sectors;
+	pkt[place]->err = regs[place].err;
+	pkt[place]->cmd = regs[place].status;
+	return len;
+}
+
+int aoeatacomplete(int place, int pktlen)
+{
+	int n;
+	int len = 60; // minimum ethernet packet size
+	atacmdcomplete(regs + place, aiocb + place);
+	if (!(pkt[place]->aflag & Write) && (n = pkt[place]->sectors)) {
+		n -= regs[place].sectors;
 		len = sizeof (Ata) + (n*512);
 	}
-	p->sectors = r.sectors;
-	p->err = r.err;
-	p->cmd = r.status;
+	pkt[place]->sectors = regs[place].sectors;
+	pkt[place]->err = regs[place].err;
+	pkt[place]->cmd = regs[place].status;
+	pending[place] = 0;
 	return len;
 }
 
+
 #define QCMD(x) ((x)->vercmd & 0xf)
 
 // yes, this makes unnecessary copies.
@@ -156,8 +184,9 @@ confcmd(Conf *p, int payload)	// process
 }
 
 void
-doaoe(Aoehdr *p, int n)
+doaoe(int place)
 {
+	Aoehdr *p = (Aoehdr *) pkt[place];
 	int len;
 	enum {	// config query header size
 		CHDR_SIZ = sizeof(Conf) - sizeof(((Conf *)0)->data),
@@ -165,14 +194,16 @@ doaoe(Aoehdr *p, int n)
 
 	switch (p->cmd) {
 	case ATAcmd:
-		if (n < sizeof(Ata))
+		if (pktlen[place] < sizeof(Ata))
+			return;
+		len = aoeata(place);
+		if (len == 0)
 			return;
-		len = aoeata((Ata*)p, n);
 		break;
 	case Config:
-		if (n < CHDR_SIZ)
+		if (pktlen[place] < CHDR_SIZ)
 			return;
-		len = confcmd((Conf *)p, n - CHDR_SIZ);
+		len = confcmd((Conf *)p, pktlen[place] - CHDR_SIZ);
 		if (len == 0)
 			return;
 		break;
@@ -193,25 +224,129 @@ doaoe(Aoehdr *p, int n)
 }
 
 void
+doaoecomplete(int place)
+{
+	Aoehdr *p = (Aoehdr *) pkt[place];
+	int len = aoeatacomplete(place, pktlen[place]);
+	memmove(p->dst, p->src, 6);
+	memmove(p->src, mac, 6);
+	p->maj = htons(shelf);
+	p->min = slot;
+	p->flags |= Resp;
+	if (putpkt(sfd, (uchar *) p, len) == -1) {
+		perror("write to network");
+		exit(1);
+	}
+
+}
+
+// allocate the buffer so that the ata data area
+// is page aligned for o_direct on linux
+
+void *
+bufalloc(void **buf, long len)
+{
+	long psize;
+	unsigned long n;
+
+	psize = sysconf(_SC_PAGESIZE);
+	if (psize == -1) {
+		perror("sysconf");
+		exit(EXIT_FAILURE);
+	}
+	n = len/psize + 3;
+	*buf = malloc(psize * n);
+	if (!*buf) {
+		perror("malloc");
+		exit(EXIT_FAILURE);
+	}
+	n = (unsigned long) *buf;
+	n += psize * 2;
+	n &= ~(psize - 1);
+	return (void *) (n - sizeof (Ata));
+}
+
+void
+sigio(int signo) 
+{
+	const char dummy = 0;
+	write(queuepipe[1], &dummy, 1);
+}
+
+void
 aoe(void)
 {
 	Aoehdr *p;
-	uchar *buf;
-	int n, sh;
+	char dummy;
+	int n, place, sh;
 	enum { bufsz = 1<<16, };
-
-	buf = malloc(bufsz);
+	sigset_t mask, oldmask;
+	struct sigaction sigact;
+	struct pollfd pollfds[2];
+	void *freeme[Nplaces];
+
+	for (n = 0; n < Nplaces; n++) {
+		pkt[n] = bufalloc(freeme + n, bufsz);
+		pending[n] = 0;
+	}
 	aoead(sfd);
 
+	pipe(queuepipe);
+	fcntl(queuepipe[0], F_SETFL, O_NONBLOCK);
+	fcntl(queuepipe[1], F_SETFL, O_NONBLOCK);
+
+	sigemptyset(&sigact.sa_mask);
+	sigact.sa_flags = 0;
+	sigact.sa_sigaction = (void *) sigio;
+	sigaction(SIGIO, &sigact, NULL);
+
+	sigemptyset(&mask);
+	sigaddset(&mask, SIGIO);
+	sigprocmask(SIG_BLOCK, &mask, &oldmask);
+
+	pollfds[0].fd = queuepipe[0];
+	pollfds[1].fd = sfd;
+	pollfds[0].events = pollfds[1].events = POLLIN;
+
 	for (;;) {
-		n = getpkt(sfd, buf, bufsz);
-		if (n < 0) {
+		sigprocmask(SIG_SETMASK, &oldmask, NULL);
+		n = poll(pollfds, 2, 1000);
+		sigprocmask(SIG_BLOCK, &mask, NULL);
+
+		if (n < 0 && errno != EINTR) {
+			perror("poll");
+			continue;
+		} else if (n == 0 || pollfds[0].revents & POLLIN) {
+			while(read(queuepipe[0], &dummy, 1) > 0);
+			for (place = 0; place < Nplaces; place++) {
+				if (!pending[place])
+					continue;
+				if (aio_error(aiocb + place) == EINPROGRESS)
+					continue;
+				doaoecomplete(place);
+				pollfds[1].events = POLLIN;
+			}
+		}
+
+		if ((pollfds[1].revents & POLLIN) == 0)
+			continue;
+			
+		for (place = 0; pending[place] && place < Nplaces; place++);
+		if (place >= Nplaces) {
+			pollfds[1].events = 0;
+			continue;
+		}
+
+		pktlen[place] = getpkt(sfd, (uchar *) pkt[place], bufsz);
+		if (pktlen[place] < 0) {
+			if (errno == EINTR)
+				continue;
 			perror("read network");
 			exit(1);
 		}
-		if (n < sizeof(Aoehdr))
+		if (pktlen[place] < sizeof(Aoehdr))
 			continue;
-		p = (Aoehdr *) buf;
+		p = (Aoehdr *) pkt[place];
 		if (ntohs(p->type) != 0x88a2)
 			continue;
 		if (p->flags & Resp)
@@ -223,9 +358,10 @@ aoe(void)
 			continue;
 		if (nmasks && !maskok(p->src))
 			continue;
-		doaoe(p, n);
+		doaoe(place);
 	}
-	free(buf);
+	for (place = 0; place < Nplaces; place++)
+		free(freeme[place]);
 }
 
 void
@@ -317,7 +453,7 @@ main(int argc, char **argv)
 	}
 	if (s.st_mode & (S_IWUSR|S_IWGRP|S_IWOTH))
 		omode = O_RDWR;
-	bfd = open(argv[3], omode);
+	bfd = opendisk(argv[3], omode);
 	if (bfd == -1) {
 		perror("open");
 		exit(1);
diff -uprN vblade-17.orig/ata.c vblade-17/ata.c
--- vblade-17.orig/ata.c	2008-06-09 10:53:07.000000000 -0400
+++ vblade-17/ata.c	2008-06-09 11:05:23.000000000 -0400
@@ -3,6 +3,8 @@
 #include <string.h>
 #include <stdio.h>
 #include <sys/types.h>
+#include <errno.h>
+#include <aio.h>
 #include "dat.h"
 #include "fns.h"
 
@@ -98,7 +100,7 @@ atainit(void)
  * check for that.
  */
 int
-atacmd(Ataregs *p, uchar *dp, int ndp, int payload) // do the ata cmd
+atacmd(Ataregs *p, uchar *dp, int ndp, int payload, struct aiocb *aiocb) // do the ata cmd
 {
 	vlong lba;
 	ushort *ip;
@@ -155,14 +157,29 @@ atacmd(Ataregs *p, uchar *dp, int ndp, i
 		return 0;
 	}
 	if (p->cmd == 0x20 || p->cmd == 0x24)
-		n = getsec(bfd, dp, lba, p->sectors);
+		n = getsec(bfd, dp, lba, p->sectors, aiocb);
 	else {
 		// packet should be big enough to contain the data
 		if (payload < 512 * p->sectors)
 			return -1;
-		n = putsec(bfd, dp, lba, p->sectors);
+		n = putsec(bfd, dp, lba, p->sectors, aiocb);
 	}
-	n /= 512;
+	if (n < 0) {
+		p->err = ABRT;
+		p->status = ERR|DRDY;
+		p->lba += n;
+		p->sectors -= n;
+		return 0;
+	}
+	return 1; // callback expected
+}
+
+
+int
+atacmdcomplete(Ataregs *p, struct aiocb *aiocb) // complete the ata cmd
+{
+	int n;
+	n = aio_return(aiocb) / 512;
 	if (n != p->sectors) {
 		p->err = ABRT;
 		p->status = ERR;
@@ -173,4 +190,3 @@ atacmd(Ataregs *p, uchar *dp, int ndp, i
 	p->sectors -= n;
 	return 0;
 }
-
diff -uprN vblade-17.orig/dat.h vblade-17/dat.h
--- vblade-17.orig/dat.h	2008-06-09 10:53:07.000000000 -0400
+++ vblade-17/dat.h	2008-06-09 11:05:23.000000000 -0400
@@ -111,6 +111,8 @@ enum {
 	Nconfig = 1024,
 
 	Bufcount = 16,
+
+	Nplaces = 32,
 };
 
 int	shelf, slot;
diff -uprN vblade-17.orig/fns.h vblade-17/fns.h
--- vblade-17.orig/fns.h	2008-06-09 10:53:07.000000000 -0400
+++ vblade-17/fns.h	2008-06-09 11:07:21.000000000 -0400
@@ -15,7 +15,8 @@ int	maskok(uchar *);
 // ata.c
 
 void	atainit(void);
-int	atacmd(Ataregs *, uchar *, int, int);
+int	atacmd(Ataregs *, uchar *, int, int, struct aiocb *);
+int	atacmdcomplete(Ataregs *, struct aiocb *);
 
 // bpf.c
 
@@ -26,8 +27,9 @@ void	free_bpf_program(void *);
 
 int	dial(char *);
 int	getea(int, char *, uchar *);
-int	putsec(int, uchar *, vlong, int);
-int	getsec(int, uchar *, vlong, int);
+int	opendisk(const char *, int);
+int	putsec(int, uchar *, vlong, int, struct aiocb *);
+int	getsec(int, uchar *, vlong, int, struct aiocb *);
 int	putpkt(int, uchar *, int);
 int	getpkt(int, uchar *, int);
 vlong	getsize(int);
diff -uprN vblade-17.orig/freebsd.c vblade-17/freebsd.c
--- vblade-17.orig/freebsd.c	2008-06-09 10:53:07.000000000 -0400
+++ vblade-17/freebsd.c	2008-06-09 11:05:23.000000000 -0400
@@ -209,19 +209,40 @@ getea(int s, char *eth, uchar *ea)
 	return(0);
 }
 
-
 int
-getsec(int fd, uchar *place, vlong lba, int nsec)
+opendisk(const char *disk, int omode)
 {
-	return pread(fd, place, nsec * 512, lba * 512);
+	return open(disk, omode);
 }
 
 int
-putsec(int fd, uchar *place, vlong lba, int nsec)
-{
-	return pwrite(fd, place, nsec * 512, lba * 512);
+getsec(int fd, uchar *place, vlong lba, int nsec, struct aiocb *aiocb)
+{       
+        bzero((char *) aiocb, sizeof(struct aiocb));
+        aiocb->aio_fildes = fd;
+        aiocb->aio_buf = place;
+        aiocb->aio_nbytes = nsec * 512;
+        aiocb->aio_offset = lba * 512;
+        aiocb->aio_sigevent.sigev_notify = SIGEV_SIGNAL;
+        aiocb->aio_sigevent.sigev_signo = SIGIO;
+        aiocb->aio_sigevent.sigev_value.sival_ptr = aiocb;
+        return aio_read(aiocb);
 }
 
+int
+putsec(int fd, uchar *place, vlong lba, int nsec, struct aiocb *aiocb)
+{       
+        bzero((char *) aiocb, sizeof(struct aiocb));
+        aiocb->aio_fildes = fd;
+        aiocb->aio_buf = place;
+        aiocb->aio_nbytes = nsec * 512;
+        aiocb->aio_offset = lba * 512;
+        aiocb->aio_sigevent.sigev_notify = SIGEV_SIGNAL;
+        aiocb->aio_sigevent.sigev_signo = SIGIO;
+        aiocb->aio_sigevent.sigev_value.sival_ptr = aiocb;
+        return aio_write(aiocb);
+}       
+
 static int pktn = 0;
 static uchar *pktbp = NULL;
 
diff -uprN vblade-17.orig/linux.c vblade-17/linux.c
--- vblade-17.orig/linux.c	2008-06-09 10:53:07.000000000 -0400
+++ vblade-17/linux.c	2008-06-09 11:05:23.000000000 -0400
@@ -1,5 +1,6 @@
 // linux.c: low level access routines for Linux
 #include "config.h"
+#define _GNU_SOURCE
 #include <sys/socket.h>
 #include <stdio.h>
 #include <string.h>
@@ -22,6 +23,9 @@
 #include <netinet/in.h>
 #include <linux/fs.h>
 #include <sys/stat.h>
+#include <fcntl.h>
+#include <errno.h>
+#include <aio.h>
 
 #include "dat.h"
 #include "fns.h"
@@ -29,8 +33,6 @@
 int	getindx(int, char *);
 int	getea(int, char *, uchar *);
 
-
-
 int
 dial(char *eth)		// get us a raw connection to an interface
 {
@@ -84,7 +86,7 @@ getea(int s, char *name, uchar *ea)
 	struct ifreq xx;
 	int n;
 
-        strcpy(xx.ifr_name, name);
+	strcpy(xx.ifr_name, name);
 	n = ioctl(s, SIOCGIFHWADDR, &xx);
 	if (n == -1) {
 		perror("Can't get hw addr");
@@ -110,17 +112,37 @@ getmtu(int s, char *name)
 }
 
 int
-getsec(int fd, uchar *place, vlong lba, int nsec)
+opendisk(const char *disk, int omode)
+{
+	return open(disk, omode|O_DIRECT);
+}
+
+int
+getsec(int fd, uchar *place, vlong lba, int nsec, struct aiocb *aiocb)
 {
-	lseek(fd, lba * 512, 0);
-	return read(fd, place, nsec * 512);
+	bzero((char *) aiocb, sizeof(struct aiocb));
+	aiocb->aio_fildes = fd;
+	aiocb->aio_buf = place;
+	aiocb->aio_nbytes = nsec * 512;
+	aiocb->aio_offset = lba * 512;
+	aiocb->aio_sigevent.sigev_notify = SIGEV_SIGNAL;
+	aiocb->aio_sigevent.sigev_signo = SIGIO;
+	aiocb->aio_sigevent.sigev_value.sival_ptr = aiocb;
+	return aio_read(aiocb);
 }
 
 int
-putsec(int fd, uchar *place, vlong lba, int nsec)
+putsec(int fd, uchar *place, vlong lba, int nsec, struct aiocb *aiocb)
 {
-	lseek(fd, lba * 512, 0);
-	return write(fd, place, nsec * 512);
+	bzero((char *) aiocb, sizeof(struct aiocb));
+	aiocb->aio_fildes = fd;
+	aiocb->aio_buf = place;
+	aiocb->aio_nbytes = nsec * 512;
+	aiocb->aio_offset = lba * 512;
+	aiocb->aio_sigevent.sigev_notify = SIGEV_SIGNAL;
+	aiocb->aio_sigevent.sigev_signo = SIGIO;
+	aiocb->aio_sigevent.sigev_value.sival_ptr = aiocb;
+	return aio_write(aiocb);
 }
 
 int
diff -uprN vblade-17.orig/linux.h vblade-17/linux.h
--- vblade-17.orig/linux.h	2008-06-09 10:53:07.000000000 -0400
+++ vblade-17/linux.h	2008-06-09 11:05:23.000000000 -0400
@@ -6,6 +6,6 @@ typedef long long vlong;
 int	dial(char *);
 int	getindx(int, char *);
 int	getea(int, char *, uchar *);
-int	getsec(int, uchar *, vlong, int);
-int	putsec(int, uchar *, vlong, int);
+int	getsec(int, uchar *, vlong, int, struct aiocb *);
+int	putsec(int, uchar *, vlong, int, struct aiocb *);
 vlong	getsize(int);
diff -uprN vblade-17.orig/makefile vblade-17/makefile
--- vblade-17.orig/makefile	2008-06-09 10:53:07.000000000 -0400
+++ vblade-17/makefile	2008-06-09 11:05:23.000000000 -0400
@@ -13,7 +13,7 @@ CFLAGS += -Wall -g -O2
 CC = gcc
 
 vblade: $O
-	${CC} -o vblade $O
+	${CC} -lrt -o vblade $O
 
 aoe.o : aoe.c config.h dat.h fns.h makefile
 	${CC} ${CFLAGS} -c $<