Re: [Devel] [PATCH rh7] ms/net: don't wait for order-3 page allocation

On 02/13/2018 07:52 PM, Alexey Kuznetsov wrote: > Hello! > > BTW, check this a few lines above: > > gfp_head = gfp_mask; > if (gfp_head & __GFP_WAIT) > gfp_head |= __GFP_REPEAT; > > What is this?? > Can't tell you why this was added and what purpose does it

[Devel] [PATCH RHEL7 COMMIT] ms/x86/kvm/vmx: do not use vm-exit instruction length for fast MMIO when running nested

The commit is pushed to "branch-rh7-3.10.0-693.17.1.vz7.43.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-693.17.1.vz7.43.4 --> commit 988f48689f68fe80186f2ec994e4f135769c1d1b Author: Vitaly Kuznetsov Date: Wed Feb 14 13:43:38

Re: [Devel] [PATCH RHEL7 COMMIT] ms/x86/kvm/vmx: do not use vm-exit instruction length for fast MMIO when running nested

On Wed, Feb 14, 2018 at 01:47:42PM +0300, Konstantin Khorenko wrote: > Makes sense to release it as a ReadyKernel patch? I don't see any point in it: this only helps VZ nested in ESXi or Hyper-V, which is probably used only in evaluation or demo scenarios. I'm pretty confident we have zero

[Devel] [PATCH] inotify: Extend ioctl to allow to request id of new watch descriptor

[This patch is accepted in ms and is going to main tree. As there is no idr_set_cursor() in 3.10 I've directly used data->idr.cur on port]. Watch descriptor is id of the watch created by inotify_add_watch(). It is allocated in inotify_add_to_idr(), and takes the numbers starting from 1. Every

Re: [Devel] [PATCH] inotify: Extend ioctl to allow to request id of new watch descriptor

https://jira.sw.ru/browse/PSBM-81411 On 14.02.2018 14:34, Kirill Tkhai wrote: > [This patch is accepted in ms and is going to main tree. > As there is no idr_set_cursor() in 3.10 I've directly > used data->idr.cur on port]. > > Watch descriptor is id of the watch created by

[Devel] [PATCH RHEL7 COMMIT] ms/net: don't wait for order-3 page allocation

The commit is pushed to "branch-rh7-3.10.0-693.17.1.vz7.43.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-693.17.1.vz7.43.4 --> commit 497a0449ec62f49bc2f29f2e1a90154c6b536fa3 Author: Shaohua Li Date: Wed Feb 14 13:17:33 2018 +0300

Re: [Devel] [PATCH rh7]: fs/pipe: Use kvmalloc for pipe buffers

On 02/13/2018 11:17 AM, Cyrill Gorcunov wrote: To restore big slabs of memory (more than 1G) via pipes we need pipes to be able to carry such sizes, for this sake Andrey propose to use kvmalloc calls. Looking into the code I didn't find a reason why we can't do so, thus here is a patch.

[Devel] [PATCH RHEL7 COMMIT] fs/pipe: Use kvmalloc for pipe buffers

The commit is pushed to "branch-rh7-3.10.0-693.17.1.vz7.43.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-693.17.1.vz7.43.4 --> commit effeb5d21ed02dbb6c155965cb83e8641286451b Author: Cyrill Gorcunov Date: Wed Feb 14 13:37:55

Re: [Devel] [PATCH RHEL7 COMMIT] ms/x86/kvm/vmx: do not use vm-exit instruction length for fast MMIO when running nested

Makes sense to release it as a ReadyKernel patch? Or the code is executed too early, so modules won't be loaded yet? -- Best regards, Konstantin Khorenko, Virtuozzo Linux Kernel Team On 02/14/2018 01:43 PM, Konstantin Khorenko wrote: The commit is pushed to

Re: [Devel] [PATCH rh7] configs.OpenVZ: add minimal kernel config for a KVM VM

On 02/13/2018 05:25 PM, Roman Kagan wrote: On Tue, Feb 13, 2018 at 05:02:33PM +0300, Konstantin Khorenko wrote: Add a minimal kernel config: - localmodconfig done on a KVM VM - all options are set to "=y" where possible What is it going to be use{d,ful} for? It should be OK for most

[Devel] [PATCH rh7 1/3] ve/kstat/alloc_lat: Don't separate GFP_HIGHMEM and !GFP_HIGHMEM allocation latencies

We use mostly 64-bit systems this days. Since they don't have higmem it's better to not segregate GFP_HIGHMEM and !GFP_HIGHMEM latencies. https://jira.sw.ru/browse/PSBM-81395 Signed-off-by: Andrey Ryabinin --- include/linux/vzstat.h | 6 ++ kernel/ve/vzstat.c |

[Devel] [PATCH rh7 2/3] ve/vzstat.h: move some kstat definitions into new header.

Move some definitions into kstat.h, so we could use later in other headers (sched.h) https://jira.sw.ru/browse/PSBM-81395 Signed-off-by: Andrey Ryabinin --- include/linux/kstat.h | 45 + include/linux/vzstat.h | 42

[Devel] [PATCH rh7 3/3] ve/page_alloc, kstat: account allocation latencies per-task

Vstorage wants per-task allocation latencies. This adds /proc//vz_latency which outputs max/total latencies and count of allocation attempts. https://jira.sw.ru/browse/PSBM-81395 Signed-off-by: Andrey Ryabinin --- fs/proc/base.c| 46

[Devel] [PATCH RHEL7 COMMIT] ms/inotify: Extend ioctl to allow to request id of new watch descriptor

The commit is pushed to "branch-rh7-3.10.0-693.17.1.vz7.43.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-693.17.1.vz7.43.5 --> commit 46e66f7faa187d0b4d266f8ba7ffc998239a43de Author: Kirill Tkhai Date: Wed Feb 14 16:19:59 2018

[Devel] [PATCH 1/4] fuse: add a new async operation to unmap regions

[This sender failed our fraud detection checks and may not be who they appear to be. Learn about spoofing at http://aka.ms/LearnAboutSpoofing] The fuse interface allows to run any operation asynchronously, because the kernel redirect all operations to an user daemon and then waits an answer. In

[Devel] [PATCH 2/4] ploop: handle discard requests via fallocate

[This sender failed our fraud detection checks and may not be who they appear to be. Learn about spoofing at http://aka.ms/LearnAboutSpoofing] Currently ploop can be compacted, but in this case discard requests are hanled with help of a userspace tool. This patch adds a native support for

[Devel] [PATCH 4/4] ploop: give aligned regions into fuse fallocate()

[This sender failed our fraud detection checks and may not be who they appear to be. Learn about spoofing at http://aka.ms/LearnAboutSpoofing] fuse fallocate() can have requirements about granularity and alignment for regions. For exmple, vstorage requires that all regions have to be aligned

[Devel] [PATCH] ploop: don't forget to dec preq->io_count on a error path

[This sender failed our fraud detection checks and may not be who they appear to be. Learn about spoofing at http://aka.ms/LearnAboutSpoofing] preq->io_count is incrimented before calling aio_kernel_submit() and it is decrimented in kaio_rw_aio_complete(). But if aio_kernel_submit() failed ,

[Devel] [PATCH rh7] target/sbc: Add LBPRZ attribute + control CDB emulation

[This sender failed our fraud detection checks and may not be who they appear to be. Learn about spoofing at http://aka.ms/LearnAboutSpoofing] From: Jamie Pocas This patch is back-ported from the upstream kernel: ML e6f41633cb79b55ead84b023c02035322c7827e7 We need it to