Some tutorial initiative for beginners
Hi experts, I scribbled thru kvm code as part of my work. I would like to make it useful for beginners , as part of that effort i created blog with some stuff. Would appreciate feedback,so that i can continue enhancing or dropping the effort all together. Once again , really appreciate your time... http://linux-kvm-internals.blogspot.in/2014/08/kvm-linux-kernel-virtual-machine.html Thanks, Ratheesh -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Linux Bridge is slow ?
I have two virtual machine VM1 and VM2 , which are running on
ubuntu. Both VM are running ubunutu. I bridged vnet0 and vnet1 in a
linux bridge on host machine.
brctl addbr virbr0
brctl addif virbr0 vnet0
brctl addif virbr0 vnet1
I did a wget of 1.9G file from VM1 to VM2. it took 1 minute 43 sec at
18.6mbps to transfer the file.
I used the attached program ( which is using epoll ) to read and write
packets. This program will register two network interfaces ( tap77
and tap88 ) with linux stack and will read packets from corresponding
file descriptor in userspace and write it in other. I bridged
tap77 and vnet0 together and tap88 and vnet1 together using a
openvswitch bridge. Her i added an overhead read, write system call
and copy_to_user, copy_from_user copies in both directions. [ from
vm1 to vm2 vm2 to vm1 ] compared to linux bridge implementation
mentioned first.
when i ran wget test. It took less time. 1.9G file from VM1 to VM2.
it took 1 minute 39 sec at 19.3 mbps to transfer the file.
Does that mean that linux bridge is introducing a lot of delay ? i
thought it should be really fast ...any clue or hint is really
appreciated ?
Thanks,
Ratheesh
#include stdio.h
#include sys/types.h
#include sys/stat.h
#include fcntl.h
#include sys/ioctl.h
#include net/if.h
#include linux/if_tun.h
#include unistd.h
#include string.h
#include stdlib.h
#include sys/epoll.h
static int
make_socket_non_blocking (int sfd)
{
int flags, s;
flags = fcntl (sfd, F_GETFL, 0);
if (flags == -1)
{
perror (fcntl);
return -1;
}
flags |= O_NONBLOCK;
s = fcntl (sfd, F_SETFL, flags);
if (s == -1)
{
perror (fcntl);
return -1;
}
return 0;
}
int tun_alloc(char *dev)
{
struct ifreq ifr;
int fd, err;
if( (fd = open(/dev/net/tun, O_RDWR)) 0 ) {
perror(cannot open /dev/net/tun);
return 0;
}
memset(ifr, 0, sizeof(ifr));
/* Flags: IFF_TUN - TUN device (no Ethernet headers)
*IFF_TAP - TAP device
*
*IFF_NO_PI - Do not provide packet information
*/
ifr.ifr_flags = IFF_TAP ;
if( *dev )
strncpy(ifr.ifr_name, dev, IFNAMSIZ);
printf(hello);
//if( err == ioctl(fd, TUNSETIFF, (void *) ifr) ){
err= ioctl(fd, TUNSETIFF, (void *) ifr);
if(err 0 ) {
close(fd);
return err;
}
strcpy(dev, ifr.ifr_name);
return fd;
}
int main()
{
char tun_name1[IFNAMSIZ];
char tun_name2[IFNAMSIZ];
int tun_fd1, tun_fd2;
unsigned char buffer[2048];
int nread, n ;
#define MAX_EVENTS 10
struct epoll_event ev_tap77, ev_tap88, events[MAX_EVENTS];
int nfds, epollfd;
epollfd = epoll_create(10);
if (epollfd == -1) {
perror(epoll_create);
exit(EXIT_FAILURE);
}
/* Connect to the device */
strcpy(tun_name1, tap77);
tun_fd1 = tun_alloc(tun_name1); /* tun interface */
//make_socket_non_blocking(tun_fd1);
strcpy(tun_name2, tap88);
tun_fd2 = tun_alloc(tun_name2); /* tun interface */
//make_socket_non_blocking(tun_fd2);
if(tun_fd1 0 || tun_fd2 0 ){
perror(\nAllocating interface);
printf(\n RAT );
exit(1);
}
ev_tap77.events = EPOLLIN;
ev_tap77.data.fd = tun_fd1;
if (epoll_ctl(epollfd, EPOLL_CTL_ADD, tun_fd1 , ev_tap77) == -1) {
perror(epoll_ctl: listen_sock);
exit(EXIT_FAILURE);
}
ev_tap88.events = EPOLLIN;
ev_tap88.data.fd = tun_fd2;
if (epoll_ctl(epollfd, EPOLL_CTL_ADD, tun_fd2, ev_tap88) == -1) {
perror(epoll_ctl: listen_sock);
exit(EXIT_FAILURE);
}
/* Now read data coming from the kernel */
while(1) {
/* Note that buffer should be at least the MTU size of the interface, eg 1500 bytes */
nfds = epoll_wait(epollfd, events, MAX_EVENTS, -1);
if (nfds == -1) {
perror(epoll_pwait);
exit(EXIT_FAILURE);
}
for (n = 0; n nfds; ++n) {
if (events[n].data.fd == tun_fd1) {
nread = read(tun_fd1,buffer,sizeof(buffer));
if(nread 0) {
perror(Reading from interface);
//close(tun_fd1);
//close(tun_fd2);
//exit(1);
continue;
}
printf(Read %d bytes from device %s\n, nread, tun_name1);
nread = write(tun_fd2,buffer, nread);
//do_use_fd(events[n].data.fd);
}
if (events[n].data.fd == tun_fd2 ) {
nread = read(tun_fd2,buffer,sizeof(buffer));
if(nread 0) {
perror(Reading from interface);
//close(tun_fd1);
//close(tun_fd2);
continue;
//exit(1);
}
printf(Read %d bytes from device %s\n, nread, tun_name2);
nread = write(tun_fd1,buffer, nread);
// do_use_fd(events[n].data.fd);
}
}
/* Do whatever with the data */
}
printf(hello ///i m done \n);
}
performance same ?
i have two ubunutu vm started using virt-manager say vm1 and vm2. Both has one interface has one network interface ( vnet0 and vnet1 respectively ) and they are bridged using virbr0 ( using brctl command ). i excuted command netperf -H vm2 ip address from vm1 ubuntu machine. i got result = 164. I wrote a usespace program - which creates two tap devices ( tap1 and tap2) and will read from tap1 and write to tap2 ( and vice versa ) using epoll system call. i removed vibr0 brdige. i bridged ( using brctl ) tap1 and vnet0 together and tap2 and vnet1 together. i repeated above netperf test. I am getting same 164. In the first test, i avoided 2 userspace copies and 2 read system call and 2 write system call. why performance is same in both case ? any hint is really appreciated ? Thanks, Ratheesh -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
host_fx_image, guest_fx_image ; what are they ?
Hi List, Could you please help me understand below variables ? i found it in struct kvm_vcpu char *host_fx_image; char *guest_fx_image; -Ratheesh -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: host_fx_image, guest_fx_image ; what are they ?
any hint ? or clue ? On Tue, Mar 4, 2014 at 11:35 PM, ratheesh kannoth ratheesh@gmail.com wrote: Hi List, Could you please help me understand below variables ? i found it in struct kvm_vcpu char *host_fx_image; char *guest_fx_image; -Ratheesh -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
where is the function paging32_page_fault() ?
I downloaded kvm-17 ( http://downloads.sourceforge.net/kvm/kvm-17.tar.gz ). I could not find the function definition in tar ball. where is it defined. ??? Thanks, Ratheesh -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: where is the function paging32_page_fault() ?
On Fri, Feb 28, 2014 at 7:37 PM, Paolo Bonzini pbonz...@redhat.com wrote: Il 28/02/2014 09:24, ratheesh kannoth ha scritto: I downloaded kvm-17 ( http://downloads.sourceforge.net/kvm/kvm-17.tar.gz ). I could not find the function definition in tar ball. That's quite old. You should just look in the Linux source code for KVM these days. It is in virt/kvm and arch/x86/kvm. Paolo Ok. Could you tell me , which kernel version will contain the function () ? -Ratheesh -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: where is the function paging32_page_fault() ?
On Fri, Feb 28, 2014 at 9:17 PM, Paolo Bonzini pbonz...@redhat.com wrote: Il 28/02/2014 16:45, ratheesh kannoth ha scritto: On Fri, Feb 28, 2014 at 7:37 PM, Paolo Bonzini pbonz...@redhat.com wrote: Il 28/02/2014 09:24, ratheesh kannoth ha scritto: I downloaded kvm-17 ( http://downloads.sourceforge.net/kvm/kvm-17.tar.gz ). I could not find the function definition in tar ball. That's quite old. You should just look in the Linux source code for KVM these days. It is in virt/kvm and arch/x86/kvm. Paolo Ok. Could you tell me , which kernel version will contain the function () ? All of them, in arch/x86/kvm/paging_tmpl.h. Paolo Paolo, Thanks a ton. I was doing a simple grep so missed the same. -Ratheesh -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Redhat hypervisor default config
As far as i know, redhat kvm hypervisor is a thin linux kernel where kvm is built as part of kernel. This is a licensed one. I can create a thin linux kernel by taking only what i need and adding kvm as part of kernel. But this tedious as there is no defconfig option. I would like to know :- is there any plan to have a default hypervisor config which we can choose a build a hypervor ? similar to redhat hypervisor, any utility to connect and configure the hypervisor ? Thanks, Ratheesh -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
