search for: io_submit

depends on "autoconf: regenerate configure scripts with 4.4 version" This series removes some of the really old deadwood from the tools build and makes some other things which are on their way out configurable at build time with a default depending on how far down the slope I judge them to be. * nuke in tree copy of libaio * nuke obsolete tools: xsview, miniterm, lomount & sv *

...enabling_ SG merge if indirect descriptors are not there. > For vq capable of indirect, it should be better to not do SG merge at default > because: > > - from hypervisor view, no matter how many segments one req has, all are > submitted to host kernel by one syscall, such as readv/io_submit > > - host kernel still need to do the same merge again Here we agree. Paolo > From my test(virtio-blk over null_blk), looks enabling SG merge may cause > throughput a little drop(~3%).

...enabling_ SG merge if indirect descriptors are not there. > For vq capable of indirect, it should be better to not do SG merge at default > because: > > - from hypervisor view, no matter how many segments one req has, all are > submitted to host kernel by one syscall, such as readv/io_submit > > - host kernel still need to do the same merge again Here we agree. Paolo > From my test(virtio-blk over null_blk), looks enabling SG merge may cause > throughput a little drop(~3%).

...uest Virtio-net driver provides header and payload address through the receive vqueue to Virtio-net backend service. 2) Virtio-net backend service encapsulates multiple vqueue elements into multiple AIO control blocks and composes them into one AIO read request. 3) Virtio-net backend service uses io_submit() syscall to pass the request to the TUN/TAP device. 4) Virtio-net backend service uses io_getevents() syscall to check the completion of the request. 5) The TUN/TAP driver receives packets from the queue pair of NIC, and prepares for Direct I/O. A modified NIC driver may render a skb which he...

...uest Virtio-net driver provides header and payload address through the receive vqueue to Virtio-net backend service. 2) Virtio-net backend service encapsulates multiple vqueue elements into multiple AIO control blocks and composes them into one AIO read request. 3) Virtio-net backend service uses io_submit() syscall to pass the request to the TUN/TAP device. 4) Virtio-net backend service uses io_getevents() syscall to check the completion of the request. 5) The TUN/TAP driver receives packets from the queue pair of NIC, and prepares for Direct I/O. A modified NIC driver may render a skb which he...

On 5/29/19 10:13 PM, Eric Blake wrote: > Benchmark-wise, using the same setup as in commit e897ed70, I see > an order-of-magnitude slowdown: > > Pre-patch, the runs averaged 1.266s, 1.30E+08 bits/s > Post-patch, the runs averaged 11.663s, 1.41E+07 bits/s > > This will need further profiling to determine how much is nbdkit's > fault, and how much is libnbd's. I

...e + google-ext" didn't get good enough performance. I'd expect it to be on par of qemu-nvme with ioeventfd but the question is: why should it be better? For vhost-net, the answer is that more zerocopy can be done if you put the data path in the kernel. But qemu-nvme is already using io_submit for the data path, perhaps there's not much to gain from vhost-nvme... Paolo > Still tuning.

...t get good enough performance. > > I'd expect it to be on par of qemu-nvme with ioeventfd but the question > is: why should it be better? For vhost-net, the answer is that more > zerocopy can be done if you put the data path in the kernel. > > But qemu-nvme is already using io_submit for the data path, perhaps > there's not much to gain from vhost-nvme... What do you think about virtio-nvme+vhost-nvme? I also have patch for vritio-nvme: https://git.kernel.org/cgit/linux/kernel/git/mlin/linux.git/log/?h=nvme-split/virtio Just need to change vhost-nvme to work with it....

...e + google-ext" didn't get good enough performance. I'd expect it to be on par of qemu-nvme with ioeventfd but the question is: why should it be better? For vhost-net, the answer is that more zerocopy can be done if you put the data path in the kernel. But qemu-nvme is already using io_submit for the data path, perhaps there's not much to gain from vhost-nvme... Paolo > Still tuning.

...also have to make libnbd itself run two separate threads per handle. It's a shame that Linux does not yet have true aio for network packets: - glibc's implementation of POSIX aio_* creates a helper thread under the hood, the same as I suggested we might do above - the Linux kernel's io_submit() still performs I/O in order rather than in parallel https://blog.cloudflare.com/io_submit-the-epoll-alternative-youve-never-heard-about/ - the Linux kernel's sendto(MSG_ZEROCOPY) can perform true async network traffic (you can regain control of your single thread prior to the full write goin...

This is a continuation of a discussion we were having on IRC. The problems with IRC are it's not recorded and it's hard to have deep technical conversations. I hope this is a decent summary. Problem simply stated: Certain NBD servers (qemu-nbd in particular) are able to simultaneously read and write on a socket. ie. They can be simultaneously reading a request and writing the reply to

...e Repro:false} mmap(&(0x7f0000000000/0x24000)=nil, 0x24000, 0x3, 0x32, 0xffffffffffffffff, 0x0) r0 = openat(0xffffffffffffff9c, &(0x7f0000019000-0x8)='./file0\x00', 0x42, 0x0) fcntl$setstatus(r0, 0x4, 0x10000) ftruncate(r0, 0x400) io_setup(0x1f, &(0x7f0000018000)=<r1=>0x0) io_submit(r1, 0x1, &(0x7f000001d000-0x28)=[&(0x7f000001b000)={0x0, 0x0, 0x0, 0x1, 0x0, r0, &(0x7f0000022000-0x1000)="0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000...

...e Repro:false} mmap(&(0x7f0000000000/0x24000)=nil, 0x24000, 0x3, 0x32, 0xffffffffffffffff, 0x0) r0 = openat(0xffffffffffffff9c, &(0x7f0000019000-0x8)='./file0\x00', 0x42, 0x0) fcntl$setstatus(r0, 0x4, 0x10000) ftruncate(r0, 0x400) io_setup(0x1f, &(0x7f0000018000)=<r1=>0x0) io_submit(r1, 0x1, &(0x7f000001d000-0x28)=[&(0x7f000001b000)={0x0, 0x0, 0x0, 0x1, 0x0, r0, &(0x7f0000022000-0x1000)="0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000...

...oughput, so it's worth a try. On Mon, Jun 6, 2016 at 10:49 AM, Rafael Espíndola < rafael.espindola at gmail.com> wrote: > Thanks a lot for running the experiment. One thing I want to try one day > is relocating one section at a time in anonymous memory and then using > async io(io_submit) to write the final bits. That way the kernel can do io > while we relocate other sections. > > Cheers, > Rafael > On Jun 5, 2016 4:19 PM, "Rui Ueyama via llvm-dev" <llvm-dev at lists.llvm.org> > wrote: > >> This is a short summary of an experiment that I...

Hello, Am 09.07.2016 um 09:14 schrieb Rowland penny: >> What is the purpose of the option >> * >> **--with-**systemd** >> ****Enable****systemd****integration* >> >> To configure Samba (build). >> > > It is there so that there is also the '--without-systemd' option. > > one turns on systemd integration, the other (thank your deity)

...(0x7f0000000000/0x24000)=nil, 0x24000, 0x3, 0x32, 0xffffffffffffffff, 0x0) > r0 = openat(0xffffffffffffff9c, &(0x7f0000019000-0x8)='./file0\x00', 0x42, 0x0) > fcntl$setstatus(r0, 0x4, 0x10000) > ftruncate(r0, 0x400) > io_setup(0x1f, &(0x7f0000018000)=<r1=>0x0) > io_submit(r1, 0x1, &(0x7f000001d000-0x28)=[&(0x7f000001b000)={0x0, 0x0, 0x0, 0x1, 0x0, r0, &(0x7f0000022000-0x1000)="0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000...

This is a short summary of an experiment that I did for the linker. One of the major tasks of the linker is to copy file contents from input object files to an output file. I was wondering what's the fastest way to copy data from one file to another, so I conducted an experiment. Currently, LLD copies file contents using memcpy (input files and an output file are mapped to memory.)

Most virtio setups have a fairly limited number of ring entries available. Enable S/G entry merging by default to fit into less of them. This restores the behavior at time of the virtio-blk blk-mq conversion, which was changed by commit "block: add queue flag for disabling SG merging" which made the behavior optional, but didn't update the existing drivers to keep their previous

Most virtio setups have a fairly limited number of ring entries available. Enable S/G entry merging by default to fit into less of them. This restores the behavior at time of the virtio-blk blk-mq conversion, which was changed by commit "block: add queue flag for disabling SG merging" which made the behavior optional, but didn't update the existing drivers to keep their previous

...yBoundingSet=CAP_CHOWN CAP_SETGID CAP_SETUID CAP_DAC_OVERRIDE CAP_KILL CAP_NET_BIND_SERVICE CAP_IPC_LOCK CAP_SYS_CHROOT SystemCallFilter=~acct modify_ldt add_key adjtimex clock_adjtime delete_module fanotify_init finit_module get_mempolicy init_module io_destroy io_getevents iopl ioperm io_setup io_submit io_cancel kcmp kexec_load keyctl lookup_dcookie mbind migrate_pages mount move_pages open_by_handle_at perf_event_open pivot_root process_vm_readv process_vm_writev ptrace remap_file_pages request_key set_mempolicy swapoff swapon umount2 uselib vmsplice ReadOnlyDirectories=/etc ReadOnlyDirector...