Documentation/filesystems/proc.rst | 6 + Documentation/mm/balloon.rst | 138 ++++++++++++++++++ MAINTAINERS | 4 +- arch/powerpc/platforms/pseries/cmm.c | 2 +- drivers/hv/hv_balloon.c | 12 ++ drivers/misc/vmw_balloon.c | 3 +- drivers/virtio/virtio_balloon.c | 7 +- fs/proc/meminfo.c | 10 ++ .../linux/{balloon_compaction.h => balloon.h} | 18 ++- lib/show_mem.c | 8 + mm/Makefile | 2 +- mm/{balloon_compaction.c => balloon.c} | 19 ++- mm/migrate.c | 1 - mm/vmscan.c | 1 - 14 files changed, 213 insertions(+), 18 deletions(-) create mode 100644 Documentation/mm/balloon.rst rename include/linux/{balloon_compaction.h => balloon.h} (91%) rename mm/{balloon_compaction.c => balloon.c} (94%)
Currently balloon drivers (Virtio,XEN, HyperV, VMWare, ...) inflate and deflate the guest memory size but there is no way to know how much the memory size is changed by them. Make it possible for the drivers to report the values to mm core. Display reported InflatedTotal and InflatedFree in /proc/meminfo and print these values on OOM and sysrq from show_mem(). The two values are the result of the two modes the drivers work with using adjust_managed_page_count or without. In earlier versions, there was a notifier for these changes but after discussion - it is better to implement it in separate patch series. Since it came out as larger work than initially expected. Amount of inflated memory can be used: - totalram_pages() users working with drivers not using adjust_managed_page_count - si_meminfo(..) users can improve calculations - by userspace software that monitors memory pressure Alexander Atanasov (8): mm: Make a place for a common balloon code mm: Enable balloon drivers to report inflated memory mm: Display inflated memory to users mm: Display inflated memory in logs drivers: virtio: balloon - report inflated memory drivers: vmware: balloon - report inflated memory drivers: hyperv: balloon - report inflated memory documentation: create a document about how balloon drivers operate Documentation/filesystems/proc.rst | 6 + Documentation/mm/balloon.rst | 138 ++++++++++++++++++ MAINTAINERS | 4 +- arch/powerpc/platforms/pseries/cmm.c | 2 +- drivers/hv/hv_balloon.c | 12 ++ drivers/misc/vmw_balloon.c | 3 +- drivers/virtio/virtio_balloon.c | 7 +- fs/proc/meminfo.c | 10 ++ .../linux/{balloon_compaction.h => balloon.h} | 18 ++- lib/show_mem.c | 8 + mm/Makefile | 2 +- mm/{balloon_compaction.c => balloon.c} | 19 ++- mm/migrate.c | 1 - mm/vmscan.c | 1 - 14 files changed, 213 insertions(+), 18 deletions(-) create mode 100644 Documentation/mm/balloon.rst rename include/linux/{balloon_compaction.h => balloon.h} (91%) rename mm/{balloon_compaction.c => balloon.c} (94%) v4->v5: - removed notifier - added documentation - vmware update after op is done , outside of the mutex v3->v4: - add support in hyperV and vmware balloon drivers - display balloon memory in show_mem so it is logged on OOM and on sysrq v2->v3: - added missed EXPORT_SYMBOLS Reported-by: kernel test robot <lkp@intel.com> - instead of balloon_common.h just use balloon.h (yes, naming is hard) - cleaned up balloon.h - remove from files that do not use it and remove externs from function declarations v1->v2: - reworked from simple /proc/meminfo addition Cc: Michael S. Tsirkin <mst@redhat.com> Cc: David Hildenbrand <david@redhat.com> Cc: Wei Liu <wei.liu@kernel.org> Cc: Nadav Amit <namit@vmware.com> Cc: pv-drivers@vmware.com Cc: Jason Wang <jasowang@redhat.com> Cc: virtualization@lists.linux-foundation.org Cc: "K. Y. Srinivasan" <kys@microsoft.com> Cc: Haiyang Zhang <haiyangz@microsoft.com> Cc: Stephen Hemminger <sthemmin@microsoft.com> Cc: Dexuan Cui <decui@microsoft.com> Cc: linux-hyperv@vger.kernel.org Cc: Juergen Gross <jgross@suse.com> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: xen-devel@lists.xenproject.org base-commit: 9abf2313adc1ca1b6180c508c25f22f9395cc780 -- 2.31.1
On Wed, 19 Oct 2022 at 12:57, Alexander Atanasov < alexander.atanasov@virtuozzo.com> wrote: > Currently balloon drivers (Virtio,XEN, HyperV, VMWare, ...) > inflate and deflate the guest memory size but there is no > way to know how much the memory size is changed by them. > > Make it possible for the drivers to report the values to mm core. > > Display reported InflatedTotal and InflatedFree in /proc/meminfo > and print these values on OOM and sysrq from show_mem(). > > The two values are the result of the two modes the drivers work > with using adjust_managed_page_count or without. > > In earlier versions, there was a notifier for these changes > but after discussion - it is better to implement it in separate > patch series. Since it came out as larger work than initially expected. > > Amount of inflated memory can be used: > - totalram_pages() users working with drivers not using > adjust_managed_page_count > - si_meminfo(..) users can improve calculations > - by userspace software that monitors memory pressure > Sorry, I see no reason for that series. Balloon inflation adjusts totalram_pages. That's enough. There is no reason to know the amount of non-existent ballooned memory inside. Management software which works outside should care about that. For debugging you could get current balloon size from /proc/vmstat (balloon_inflate - balloon_deflate). Also (I guess) /proc/kpageflags has a bit for that. Anyway it's easy to monitor balloon inflation by seeing changes of total memory size. > > Alexander Atanasov (8): > mm: Make a place for a common balloon code > mm: Enable balloon drivers to report inflated memory > mm: Display inflated memory to users > mm: Display inflated memory in logs > drivers: virtio: balloon - report inflated memory > drivers: vmware: balloon - report inflated memory > drivers: hyperv: balloon - report inflated memory > documentation: create a document about how balloon drivers operate > > Documentation/filesystems/proc.rst | 6 + > Documentation/mm/balloon.rst | 138 ++++++++++++++++++ > MAINTAINERS | 4 +- > arch/powerpc/platforms/pseries/cmm.c | 2 +- > drivers/hv/hv_balloon.c | 12 ++ > drivers/misc/vmw_balloon.c | 3 +- > drivers/virtio/virtio_balloon.c | 7 +- > fs/proc/meminfo.c | 10 ++ > .../linux/{balloon_compaction.h => balloon.h} | 18 ++- > lib/show_mem.c | 8 + > mm/Makefile | 2 +- > mm/{balloon_compaction.c => balloon.c} | 19 ++- > mm/migrate.c | 1 - > mm/vmscan.c | 1 - > 14 files changed, 213 insertions(+), 18 deletions(-) > create mode 100644 Documentation/mm/balloon.rst > rename include/linux/{balloon_compaction.h => balloon.h} (91%) > rename mm/{balloon_compaction.c => balloon.c} (94%) > > v4->v5: > - removed notifier > - added documentation > - vmware update after op is done , outside of the mutex > v3->v4: > - add support in hyperV and vmware balloon drivers > - display balloon memory in show_mem so it is logged on OOM and on sysrq > v2->v3: > - added missed EXPORT_SYMBOLS > Reported-by: kernel test robot <lkp@intel.com> > - instead of balloon_common.h just use balloon.h (yes, naming is hard) > - cleaned up balloon.h - remove from files that do not use it and > remove externs from function declarations > v1->v2: > - reworked from simple /proc/meminfo addition > > Cc: Michael S. Tsirkin <mst@redhat.com> > Cc: David Hildenbrand <david@redhat.com> > Cc: Wei Liu <wei.liu@kernel.org> > Cc: Nadav Amit <namit@vmware.com> > Cc: pv-drivers@vmware.com > Cc: Jason Wang <jasowang@redhat.com> > Cc: virtualization@lists.linux-foundation.org > Cc: "K. Y. Srinivasan" <kys@microsoft.com> > Cc: Haiyang Zhang <haiyangz@microsoft.com> > Cc: Stephen Hemminger <sthemmin@microsoft.com> > Cc: Dexuan Cui <decui@microsoft.com> > Cc: linux-hyperv@vger.kernel.org > Cc: Juergen Gross <jgross@suse.com> > Cc: Stefano Stabellini <sstabellini@kernel.org> > Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> > Cc: xen-devel@lists.xenproject.org > > base-commit: 9abf2313adc1ca1b6180c508c25f22f9395cc780 > -- > 2.31.1 > >
On 19.10.22 13:53, Konstantin Khlebnikov wrote: > On Wed, 19 Oct 2022 at 12:57, Alexander Atanasov > <alexander.atanasov@virtuozzo.com > <mailto:alexander.atanasov@virtuozzo.com>> wrote: > > Currently balloon drivers (Virtio,XEN, HyperV, VMWare, ...) > inflate and deflate the guest memory size but there is no > way to know how much the memory size is changed by them. > > Make it possible for the drivers to report the values to mm core. > > Display reported InflatedTotal and InflatedFree in /proc/meminfo > and print these values on OOM and sysrq from show_mem(). > > The two values are the result of the two modes the drivers work > with using adjust_managed_page_count or without. > > In earlier versions, there was a notifier for these changes > but after discussion - it is better to implement it in separate > patch series. Since it came out as larger work than initially expected. > > Amount of inflated memory can be used: > - totalram_pages() users working with drivers not using > adjust_managed_page_count > - si_meminfo(..) users can improve calculations > - by userspace software that monitors memory pressure > > > Sorry, I see no reason for that series. > Balloon inflation adjusts totalram_pages. That's enough. That is not true in all cases - some do some do not. > > There is no reason to know the amount of non-existent ballooned memory > inside. Why? Memory managment is becoming more and more dynamic - to make it easy and accurate you need to know that amount. The kernel itself on boot pre-allocates caches and sets limits based on total ram at boot time when balloon steals half of the memory these initial calculations become very wrong. To redo them correctly these amounts need to be known. The idea of doing this is thru a notifier chain in a separte series. > Management software which works outside should care about that. > > For debugging you could get current balloon size from /proc/vmstat > (balloon_inflate - balloon_deflate). Currently you can do that only if using VMWare balloon. > Also (I guess) /proc/kpageflags has a bit for that. > > Anyway it's easy to monitor balloon inflation by seeing changes of total > memory size. Not all drivers do that - VMWare and virtio (in one case) for example do not. I proposed to unify them but since it can break existing users it was NAKed. > > Alexander Atanasov (8): > mm: Make a place for a common balloon code > mm: Enable balloon drivers to report inflated memory > mm: Display inflated memory to users > mm: Display inflated memory in logs > drivers: virtio: balloon - report inflated memory > drivers: vmware: balloon - report inflated memory > drivers: hyperv: balloon - report inflated memory > documentation: create a document about how balloon drivers operate > > Documentation/filesystems/proc.rst | 6 + > Documentation/mm/balloon.rst | 138 ++++++++++++++++++ > MAINTAINERS | 4 +- > arch/powerpc/platforms/pseries/cmm.c | 2 +- > drivers/hv/hv_balloon.c | 12 ++ > drivers/misc/vmw_balloon.c | 3 +- > drivers/virtio/virtio_balloon.c | 7 +- > fs/proc/meminfo.c | 10 ++ > .../linux/{balloon_compaction.h => balloon.h} | 18 ++- > lib/show_mem.c | 8 + > mm/Makefile | 2 +- > mm/{balloon_compaction.c => balloon.c} | 19 ++- > mm/migrate.c | 1 - > mm/vmscan.c | 1 - > 14 files changed, 213 insertions(+), 18 deletions(-) > create mode 100644 Documentation/mm/balloon.rst > rename include/linux/{balloon_compaction.h => balloon.h} (91%) > rename mm/{balloon_compaction.c => balloon.c} (94%) > > v4->v5: > - removed notifier > - added documentation > - vmware update after op is done , outside of the mutex > v3->v4: > - add support in hyperV and vmware balloon drivers > - display balloon memory in show_mem so it is logged on OOM and on > sysrq > v2->v3: > - added missed EXPORT_SYMBOLS > Reported-by: kernel test robot <lkp@intel.com <mailto:lkp@intel.com>> > - instead of balloon_common.h just use balloon.h (yes, naming is hard) > - cleaned up balloon.h - remove from files that do not use it and > remove externs from function declarations > v1->v2: > - reworked from simple /proc/meminfo addition > > Cc: Michael S. Tsirkin <mst@redhat.com <mailto:mst@redhat.com>> > Cc: David Hildenbrand <david@redhat.com <mailto:david@redhat.com>> > Cc: Wei Liu <wei.liu@kernel.org <mailto:wei.liu@kernel.org>> > Cc: Nadav Amit <namit@vmware.com <mailto:namit@vmware.com>> > Cc: pv-drivers@vmware.com <mailto:pv-drivers@vmware.com> > Cc: Jason Wang <jasowang@redhat.com <mailto:jasowang@redhat.com>> > Cc: virtualization@lists.linux-foundation.org > <mailto:virtualization@lists.linux-foundation.org> > Cc: "K. Y. Srinivasan" <kys@microsoft.com <mailto:kys@microsoft.com>> > Cc: Haiyang Zhang <haiyangz@microsoft.com > <mailto:haiyangz@microsoft.com>> > Cc: Stephen Hemminger <sthemmin@microsoft.com > <mailto:sthemmin@microsoft.com>> > Cc: Dexuan Cui <decui@microsoft.com <mailto:decui@microsoft.com>> > Cc: linux-hyperv@vger.kernel.org <mailto:linux-hyperv@vger.kernel.org> > Cc: Juergen Gross <jgross@suse.com <mailto:jgross@suse.com>> > Cc: Stefano Stabellini <sstabellini@kernel.org > <mailto:sstabellini@kernel.org>> > Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com > <mailto:oleksandr_tyshchenko@epam.com>> > Cc: xen-devel@lists.xenproject.org > <mailto:xen-devel@lists.xenproject.org> > > base-commit: 9abf2313adc1ca1b6180c508c25f22f9395cc780 > -- > 2.31.1 > -- Regards, Alexander Atanasov
On 10/19/22 12:53, Konstantin Khlebnikov wrote: > On Wed, 19 Oct 2022 at 12:57, Alexander Atanasov > <alexander.atanasov@virtuozzo.com> wrote: > > Currently balloon drivers (Virtio,XEN, HyperV, VMWare, ...) > inflate and deflate the guest memory size but there is no > way to know how much the memory size is changed by them. > > Make it possible for the drivers to report the values to mm core. > > Display reported InflatedTotal and InflatedFree in /proc/meminfo > and print these values on OOM and sysrq from show_mem(). > > The two values are the result of the two modes the drivers work > with using adjust_managed_page_count or without. > > In earlier versions, there was a notifier for these changes > but after discussion - it is better to implement it in separate > patch series. Since it came out as larger work than initially > expected. > > Amount of inflated memory can be used: > - totalram_pages() users working with drivers not using > adjust_managed_page_count > - si_meminfo(..) users can improve calculations > - by userspace software that monitors memory pressure > > > Sorry, I see no reason for that series. > Balloon inflation adjusts totalram_pages. That's enough. > no, they are not at least under some circumstances, f.e. virtio balloon does not do that with VIRTIO_BALLOON_F_DEFLATE_ON_OOM set > There is no reason to know the amount of non-existent ballooned memory > inside. > Management software which works outside should care about that. > The problem comes at the moment when we are running our Linux server inside virtual machine and the customer comes with crazy questions "where our memory?". > For debugging you could get current balloon size from /proc/vmstat > (balloon_inflate - balloon_deflate). > Also (I guess) /proc/kpageflags has a bit for that. > > Anyway it's easy to monitor balloon inflation by seeing changes of > total memory size. for monitoring - may be. But in order to report total amount there is no interface so far. > > Alexander Atanasov (8): > mm: Make a place for a common balloon code > mm: Enable balloon drivers to report inflated memory > mm: Display inflated memory to users > mm: Display inflated memory in logs > drivers: virtio: balloon - report inflated memory > drivers: vmware: balloon - report inflated memory > drivers: hyperv: balloon - report inflated memory > documentation: create a document about how balloon drivers operate > > Documentation/filesystems/proc.rst | 6 + > Documentation/mm/balloon.rst | 138 > ++++++++++++++++++ > MAINTAINERS | 4 +- > arch/powerpc/platforms/pseries/cmm.c | 2 +- > drivers/hv/hv_balloon.c | 12 ++ > drivers/misc/vmw_balloon.c | 3 +- > drivers/virtio/virtio_balloon.c | 7 +- > fs/proc/meminfo.c | 10 ++ > .../linux/{balloon_compaction.h => balloon.h} | 18 ++- > lib/show_mem.c | 8 + > mm/Makefile | 2 +- > mm/{balloon_compaction.c => balloon.c} | 19 ++- > mm/migrate.c | 1 - > mm/vmscan.c | 1 - > 14 files changed, 213 insertions(+), 18 deletions(-) > create mode 100644 Documentation/mm/balloon.rst > rename include/linux/{balloon_compaction.h => balloon.h} (91%) > rename mm/{balloon_compaction.c => balloon.c} (94%) > > v4->v5: > - removed notifier > - added documentation > - vmware update after op is done , outside of the mutex > v3->v4: > - add support in hyperV and vmware balloon drivers > - display balloon memory in show_mem so it is logged on OOM and > on sysrq > v2->v3: > - added missed EXPORT_SYMBOLS > Reported-by: kernel test robot <lkp@intel.com> > - instead of balloon_common.h just use balloon.h (yes, naming is > hard) > - cleaned up balloon.h - remove from files that do not use it and > remove externs from function declarations > v1->v2: > - reworked from simple /proc/meminfo addition > > Cc: Michael S. Tsirkin <mst@redhat.com> > Cc: David Hildenbrand <david@redhat.com> > Cc: Wei Liu <wei.liu@kernel.org> > Cc: Nadav Amit <namit@vmware.com> > Cc: pv-drivers@vmware.com > Cc: Jason Wang <jasowang@redhat.com> > Cc: virtualization@lists.linux-foundation.org > Cc: "K. Y. Srinivasan" <kys@microsoft.com> > Cc: Haiyang Zhang <haiyangz@microsoft.com> > Cc: Stephen Hemminger <sthemmin@microsoft.com> > Cc: Dexuan Cui <decui@microsoft.com> > Cc: linux-hyperv@vger.kernel.org > Cc: Juergen Gross <jgross@suse.com> > Cc: Stefano Stabellini <sstabellini@kernel.org> > Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> > Cc: xen-devel@lists.xenproject.org > > base-commit: 9abf2313adc1ca1b6180c508c25f22f9395cc780 > -- > 2.31.1 >
On Wed, 19 Oct 2022 at 14:06, Denis V. Lunev <den@virtuozzo.com> wrote: > On 10/19/22 12:53, Konstantin Khlebnikov wrote: > > On Wed, 19 Oct 2022 at 12:57, Alexander Atanasov > > <alexander.atanasov@virtuozzo.com> wrote: > > > > Currently balloon drivers (Virtio,XEN, HyperV, VMWare, ...) > > inflate and deflate the guest memory size but there is no > > way to know how much the memory size is changed by them. > > > > Make it possible for the drivers to report the values to mm core. > > > > Display reported InflatedTotal and InflatedFree in /proc/meminfo > > and print these values on OOM and sysrq from show_mem(). > > > > The two values are the result of the two modes the drivers work > > with using adjust_managed_page_count or without. > > > > In earlier versions, there was a notifier for these changes > > but after discussion - it is better to implement it in separate > > patch series. Since it came out as larger work than initially > > expected. > > > > Amount of inflated memory can be used: > > - totalram_pages() users working with drivers not using > > adjust_managed_page_count > > - si_meminfo(..) users can improve calculations > > - by userspace software that monitors memory pressure > > > > > > Sorry, I see no reason for that series. > > Balloon inflation adjusts totalram_pages. That's enough. > > > no, they are not at least under some circumstances, f.e. > virtio balloon does not do that with VIRTIO_BALLOON_F_DEFLATE_ON_OOM > set > > > There is no reason to know the amount of non-existent ballooned memory > > inside. > > Management software which works outside should care about that. > > > The problem comes at the moment when we are running > our Linux server inside virtual machine and the customer > comes with crazy questions "where our memory?". > Ok. In this case balloon management is partially inside VM. I.e. we could report portion of balloon as potentially available memory. I guess memory pressure could deflate balloon till some threshold set by external hypervisor. So, without knowledge about this threshold there is no correct answer about size of available memory. Showing just size of balloon doesn't gives much. > > > For debugging you could get current balloon size from /proc/vmstat > > (balloon_inflate - balloon_deflate). > > Also (I guess) /proc/kpageflags has a bit for that. > > > > Anyway it's easy to monitor balloon inflation by seeing changes of > > total memory size. > for monitoring - may be. But in order to report total amount > there is no interface so far. > > > > > Alexander Atanasov (8): > > mm: Make a place for a common balloon code > > mm: Enable balloon drivers to report inflated memory > > mm: Display inflated memory to users > > mm: Display inflated memory in logs > > drivers: virtio: balloon - report inflated memory > > drivers: vmware: balloon - report inflated memory > > drivers: hyperv: balloon - report inflated memory > > documentation: create a document about how balloon drivers operate > > > > Documentation/filesystems/proc.rst | 6 + > > Documentation/mm/balloon.rst | 138 > > ++++++++++++++++++ > > MAINTAINERS | 4 +- > > arch/powerpc/platforms/pseries/cmm.c | 2 +- > > drivers/hv/hv_balloon.c | 12 ++ > > drivers/misc/vmw_balloon.c | 3 +- > > drivers/virtio/virtio_balloon.c | 7 +- > > fs/proc/meminfo.c | 10 ++ > > .../linux/{balloon_compaction.h => balloon.h} | 18 ++- > > lib/show_mem.c | 8 + > > mm/Makefile | 2 +- > > mm/{balloon_compaction.c => balloon.c} | 19 ++- > > mm/migrate.c | 1 - > > mm/vmscan.c | 1 - > > 14 files changed, 213 insertions(+), 18 deletions(-) > > create mode 100644 Documentation/mm/balloon.rst > > rename include/linux/{balloon_compaction.h => balloon.h} (91%) > > rename mm/{balloon_compaction.c => balloon.c} (94%) > > > > v4->v5: > > - removed notifier > > - added documentation > > - vmware update after op is done , outside of the mutex > > v3->v4: > > - add support in hyperV and vmware balloon drivers > > - display balloon memory in show_mem so it is logged on OOM and > > on sysrq > > v2->v3: > > - added missed EXPORT_SYMBOLS > > Reported-by: kernel test robot <lkp@intel.com> > > - instead of balloon_common.h just use balloon.h (yes, naming is > > hard) > > - cleaned up balloon.h - remove from files that do not use it and > > remove externs from function declarations > > v1->v2: > > - reworked from simple /proc/meminfo addition > > > > Cc: Michael S. Tsirkin <mst@redhat.com> > > Cc: David Hildenbrand <david@redhat.com> > > Cc: Wei Liu <wei.liu@kernel.org> > > Cc: Nadav Amit <namit@vmware.com> > > Cc: pv-drivers@vmware.com > > Cc: Jason Wang <jasowang@redhat.com> > > Cc: virtualization@lists.linux-foundation.org > > Cc: "K. Y. Srinivasan" <kys@microsoft.com> > > Cc: Haiyang Zhang <haiyangz@microsoft.com> > > Cc: Stephen Hemminger <sthemmin@microsoft.com> > > Cc: Dexuan Cui <decui@microsoft.com> > > Cc: linux-hyperv@vger.kernel.org > > Cc: Juergen Gross <jgross@suse.com> > > Cc: Stefano Stabellini <sstabellini@kernel.org> > > Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> > > Cc: xen-devel@lists.xenproject.org > > > > base-commit: 9abf2313adc1ca1b6180c508c25f22f9395cc780 > > -- > > 2.31.1 > > > >
On 19.10.22 18:39, Konstantin Khlebnikov wrote: > > > On Wed, 19 Oct 2022 at 14:06, Denis V. Lunev <den@virtuozzo.com > <mailto:den@virtuozzo.com>> wrote: > > On 10/19/22 12:53, Konstantin Khlebnikov wrote: > > On Wed, 19 Oct 2022 at 12:57, Alexander Atanasov > > <alexander.atanasov@virtuozzo.com > <mailto:alexander.atanasov@virtuozzo.com>> wrote: > > > > Currently balloon drivers (Virtio,XEN, HyperV, VMWare, ...) > > inflate and deflate the guest memory size but there is no > > way to know how much the memory size is changed by them. > > > > Make it possible for the drivers to report the values to mm core. > > > > Display reported InflatedTotal and InflatedFree in /proc/meminfo > > and print these values on OOM and sysrq from show_mem(). > > > > The two values are the result of the two modes the drivers work > > with using adjust_managed_page_count or without. > > > > In earlier versions, there was a notifier for these changes > > but after discussion - it is better to implement it in separate > > patch series. Since it came out as larger work than initially > > expected. > > > > Amount of inflated memory can be used: > > - totalram_pages() users working with drivers not using > > adjust_managed_page_count > > - si_meminfo(..) users can improve calculations > > - by userspace software that monitors memory pressure > > > > > > Sorry, I see no reason for that series. > > Balloon inflation adjusts totalram_pages. That's enough. > > > no, they are not at least under some circumstances, f.e. > virtio balloon does not do that with VIRTIO_BALLOON_F_DEFLATE_ON_OOM > set > > > > There is no reason to know the amount of non-existent ballooned > memory > > inside. > > Management software which works outside should care about that. > > > The problem comes at the moment when we are running > our Linux server inside virtual machine and the customer > comes with crazy questions "where our memory?". > > > Ok. In this case balloon management is partially inside VM. > I.e. we could report portion of balloon as potentially available memory. > > I guess memory pressure could deflate balloon till some threshold set by > external hypervisor. > So, without knowledge about this threshold there is no correct answer > about size of available memory. > Showing just size of balloon doesn't gives much. You need the current and the adjustment to get the absolute top. If you check only totalram_pages() it is the current. To get the absolute maximum you need to know how much the balloon adjusted it. The drivers that do not adjust totalram_pages() and leave the inflated memory as used assume that this memory can be reclaimed at anytime. But that assumption is not completely true and provides the system with false totalram value. Why - VMWare does not have oom_notifier at all (it is possible to have sone other mechanism, i do not know), Virtio balloon reclaims 1MB on OOM _if_ it can. -- Regards, Alexander Atanasov
© 2016 - 2024 Red Hat, Inc.