On 07.09.2016 23:20, Alexei Starovoitov wrote: > On Wed, Sep 07, 2016 at 01:27:35PM +0300, Yauheni Kaliuta wrote: >> The patch instrument different places of resource limits checks with >> reporting using the infrastructure from the previous patch. >> >> Signed-off-by: Yauheni Kaliuta <yauheni.kali...@redhat.com> >> --- >> arch/ia64/kernel/perfmon.c | 4 +++- >> arch/powerpc/kvm/book3s_64_vio.c | 6 ++++-- >> arch/powerpc/mm/mmu_context_iommu.c | 6 ++++-- >> drivers/android/binder.c | 7 ++++++- >> drivers/infiniband/core/umem.c | 1 + >> drivers/infiniband/hw/hfi1/user_pages.c | 5 ++++- >> drivers/infiniband/hw/qib/qib_user_pages.c | 1 + >> drivers/infiniband/hw/usnic/usnic_uiom.c | 1 + >> drivers/misc/mic/scif/scif_rma.c | 1 + >> drivers/vfio/vfio_iommu_spapr_tce.c | 6 ++++-- >> drivers/vfio/vfio_iommu_type1.c | 4 ++++ >> fs/attr.c | 4 +++- >> fs/binfmt_aout.c | 4 +++- >> fs/binfmt_flat.c | 1 + >> fs/coredump.c | 4 +++- >> fs/exec.c | 14 ++++++++++---- >> fs/file.c | 26 +++++++++++++++++++++----- >> fs/select.c | 4 +++- >> include/linux/mm.h | 7 ++++++- >> ipc/mqueue.c | 10 ++++++++-- >> ipc/shm.c | 1 + >> kernel/bpf/syscall.c | 15 ++++++++++++--- >> kernel/events/core.c | 1 + >> kernel/fork.c | 9 ++++++--- >> kernel/sched/core.c | 17 +++++++++++++---- >> kernel/signal.c | 7 ++++--- >> kernel/sys.c | 9 ++++++--- >> kernel/time/posix-cpu-timers.c | 8 ++++++++ >> mm/mlock.c | 14 +++++++++++++- >> mm/mmap.c | 19 +++++++++++++++---- >> mm/mremap.c | 4 +++- >> net/unix/af_unix.c | 9 ++++++--- >> 32 files changed, 179 insertions(+), 50 deletions(-) > > I'm certainly not excited that we'd need to maintain this > rlimit tracking for foreseeable future. > I can be convinced otherwise, but so far I don't see > strong enough use case that warrants these changes all over.
Other operating systems actually allow reporting of those events up till syslog. I does have some value to get to this information at all (how I hate silent errors on iPhones with Jetsam because of resource limitations). Some way to centrally notify about those would actually be a nice thing to have. Bye, Hannes
