search for: mmget_not_zero

...er's mm has been bound. So, only when the bus supports user VA > (e.g. vhost-vdpa). > > vdpasim_mm_work_fn work is used to serialize the binding to a new > address space when the .bind_mm callback is invoked, and unbinding > when the .unbind_mm callback is invoked. > > Call mmget_not_zero()/kthread_use_mm() inside the worker function > to pin the address space only as long as needed, following the > documentation of mmget() in include/linux/sched/mm.h: > > * Never use this function to pin this address space for an > * unbounded/indefinite amount of time. > &g...

...So, only when the bus supports user VA >> (e.g. vhost-vdpa). >> >> vdpasim_mm_work_fn work is used to serialize the binding to a new >> address space when the .bind_mm callback is invoked, and unbinding >> when the .unbind_mm callback is invoked. >> >> Call mmget_not_zero()/kthread_use_mm() inside the worker function >> to pin the address space only as long as needed, following the >> documentation of mmget() in include/linux/sched/mm.h: >> >> * Never use this function to pin this address space for an >> * unbounded/indefinite amoun...

...er's mm has been bound. So, only when the bus supports user VA > (e.g. vhost-vdpa). > > vdpasim_mm_work_fn work is used to serialize the binding to a new > address space when the .bind_mm callback is invoked, and unbinding > when the .unbind_mm callback is invoked. > > Call mmget_not_zero()/kthread_use_mm() inside the worker function > to pin the address space only as long as needed, following the > documentation of mmget() in include/linux/sched/mm.h: > > * Never use this function to pin this address space for an > * unbounded/indefinite amount of time. I wonder...

...vhost-vdpa). >> >> >> >> vdpasim_mm_work_fn work is used to serialize the binding to a new >> >> address space when the .bind_mm callback is invoked, and unbinding >> >> when the .unbind_mm callback is invoked. >> >> >> >> Call mmget_not_zero()/kthread_use_mm() inside the worker function >> >> to pin the address space only as long as needed, following the >> >> documentation of mmget() in include/linux/sched/mm.h: >> >> >> >> * Never use this function to pin this address space for an &gt...

...mirror *dmirror = container_of(mni, struct dmirror, notifier); > + struct mm_struct *mm = dmirror->mm; > + > + /* > + * If the process doesn't exist, we don't need to invalidate the > + * device page table since the address space will be torn down. > + */ > + if (!mmget_not_zero(mm)) > + return true; Why? Don't the notifiers provide for this already. mmget_not_zero() is required before calling hmm_range_fault() though > +static int dmirror_fault(struct dmirror *dmirror, unsigned long start, > + unsigned long end, bool write) > +{ > + struct mm_st...

On 3/17/20 5:59 AM, Christoph Hellwig wrote: > On Tue, Mar 17, 2020 at 09:47:55AM -0300, Jason Gunthorpe wrote: >> I've been using v7 of Ralph's tester and it is working well - it has >> DEVICE_PRIVATE support so I think it can test this flow too. Ralph are >> you able? >> >> This hunk seems trivial enough to me, can we include it now? > > I can send

...of(mni, struct dmirror, notifier); >> + struct mm_struct *mm = dmirror->mm; >> + >> + /* >> + * If the process doesn't exist, we don't need to invalidate the >> + * device page table since the address space will be torn down. >> + */ >> + if (!mmget_not_zero(mm)) >> + return true; > > Why? Don't the notifiers provide for this already. > > mmget_not_zero() is required before calling hmm_range_fault() though This is a workaround for a problem I don't quite understand. If you change tools/testing/selftests/vm/hmm-tests.c line...

...start < vma->vm_start)) { > - r = -EFAULT; > - goto out; > - } > - if (unlikely((gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) && > - vma->vm_file)) { > - r = -EPERM; > - goto out; > - } > + mm = mirror->hmm->mmu_notifier.mm; > + if (!mmget_not_zero(mm)) /* Happens during process shutdown */ This works because mirror->hmm->mmu_notifier holds an mmgrab reference to the mm? So the MM will not just go away, but if the mmget refcount is 0, it means the mm is marked for destruction and shouldn't be used any more. > + return -ESRC...

...t; > + struct mm_struct *mm = dmirror->mm; > > > + > > > + /* > > > + * If the process doesn't exist, we don't need to invalidate the > > > + * device page table since the address space will be torn down. > > > + */ > > > + if (!mmget_not_zero(mm)) > > > + return true; > > > > Why? Don't the notifiers provide for this already. > > > > mmget_not_zero() is required before calling hmm_range_fault() though Oh... This is the invalidate_all path during invalidation IMHO you should test the invalidation...

Let's move work management inside the vdpa_sim core. This way we can easily change how we manage the works, without having to change the devices each time. Acked-by: Eugenio P??rez Martin <eperezma at redhat.com> Acked-by: Jason Wang <jasowang at redhat.com> Signed-off-by: Stefano Garzarella <sgarzare at redhat.com> --- drivers/vdpa/vdpa_sim/vdpa_sim.h | 3 ++-

...limit = start + (ARRAY_SIZE(args.phys) << PAGE_SHIFT); + limit = start + PAGE_SIZE; if (start < svmm->unmanaged.limit) limit = min_t(u64, limit, svmm->unmanaged.start); - SVMM_DBG(svmm, "wndw %016llx-%016llx", start, limit); - mm = svmm->notifier.mm; - if (!mmget_not_zero(mm)) { - nouveau_svm_fault_cancel_fault(svm, buffer->fault[fi]); - continue; - } - - /* Intersect fault window with the CPU VMA, cancelling - * the fault if the address is invalid. + /* + * Prepare the GPU-side update of all pages within the + * fault window, determining required pa...

...limit = start + (ARRAY_SIZE(args.phys) << PAGE_SHIFT); + limit = start + PAGE_SIZE; if (start < svmm->unmanaged.limit) limit = min_t(u64, limit, svmm->unmanaged.start); - SVMM_DBG(svmm, "wndw %016llx-%016llx", start, limit); - mm = svmm->notifier.mm; - if (!mmget_not_zero(mm)) { - nouveau_svm_fault_cancel_fault(svm, buffer->fault[fi]); - continue; - } - - /* Intersect fault window with the CPU VMA, cancelling - * the fault if the address is invalid. + /* + * Prepare the GPU-side update of all pages within the + * fault window, determining required pa...

...a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvmgt.c index dee01c371bf5..92e9b340dbc2 100644 --- a/drivers/gpu/drm/i915/gvt/kvmgt.c +++ b/drivers/gpu/drm/i915/gvt/kvmgt.c @@ -2048,7 +2048,7 @@ static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa, if (kthread) { if (!mmget_not_zero(kvm->mm)) return -EFAULT; - use_mm(kvm->mm); + kthread_use_mm(kvm->mm); } idx = srcu_read_lock(&kvm->srcu); @@ -2057,7 +2057,7 @@ static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa, srcu_read_unlock(&kvm->srcu, idx); if (kthread) { - unuse_mm(k...

...a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvmgt.c index ca1dd6e6f395..f2927575b793 100644 --- a/drivers/gpu/drm/i915/gvt/kvmgt.c +++ b/drivers/gpu/drm/i915/gvt/kvmgt.c @@ -2048,7 +2048,7 @@ static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa, if (kthread) { if (!mmget_not_zero(kvm->mm)) return -EFAULT; - use_mm(kvm->mm); + kthread_use_mm(kvm->mm); } idx = srcu_read_lock(&kvm->srcu); @@ -2057,7 +2057,7 @@ static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa, srcu_read_unlock(&kvm->srcu, idx); if (kthread) { - unuse_mm(k...

...5,6 @@ The usage pattern is:: > range.start = ...; > range.end = ...; > range.pfns = ...; That should be: range.hmm_pfns = ...; > - range.flags = ...; > - range.values = ...; > - range.pfn_shift = ...; > > if (!mmget_not_zero(interval_sub->notifier.mm)) > return -EFAULT; > @@ -229,15 +226,10 @@ The hmm_range struct has 2 fields, default_flags and pfn_flags_mask, that specif > fault or snapshot policy for the whole range instead of having to set them > for each entry in the pfns array. >...

...vma = find_vma(mm, start); - if (unlikely(!vma || start < vma->vm_start)) { - r = -EFAULT; - goto out; - } - if (unlikely((gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) && - vma->vm_file)) { - r = -EPERM; - goto out; - } + mm = mirror->hmm->mmu_notifier.mm; + if (!mmget_not_zero(mm)) /* Happens during process shutdown */ + return -ESRCH; range = kzalloc(sizeof(*range), GFP_KERNEL); if (unlikely(!range)) { @@ -847,6 +837,17 @@ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages) hmm_range_wait_until_valid(range, HMM_RANGE_DEFAULT_TIMEOUT);...

...gt;mm); mmput(worker->mm); worker->mm = NULL; @@ -420,14 +419,11 @@ static void io_wq_switch_mm(struct io_worker *worker, struct io_wq_work *work) mmput(worker->mm); worker->mm = NULL; } - if (!work->mm) { - set_fs(KERNEL_DS); + if (!work->mm) return; - } + if (mmget_not_zero(work->mm)) { kthread_use_mm(work->mm); - if (!worker->mm) - set_fs(USER_DS); worker->mm = work->mm; /* hang on to this mm */ work->mm = NULL; diff --git a/fs/io_uring.c b/fs/io_uring.c index 367406381044..c332a34e8b34 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @...

...gt;mm); mmput(worker->mm); worker->mm = NULL; @@ -421,14 +420,11 @@ static void io_wq_switch_mm(struct io_worker *worker, struct io_wq_work *work) mmput(worker->mm); worker->mm = NULL; } - if (!work->mm) { - set_fs(KERNEL_DS); + if (!work->mm) return; - } + if (mmget_not_zero(work->mm)) { kthread_use_mm(work->mm); - if (!worker->mm) - set_fs(USER_DS); worker->mm = work->mm; /* hang on to this mm */ work->mm = NULL; diff --git a/fs/io_uring.c b/fs/io_uring.c index 8a8148512da7..40f90b98a18a 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @...

Hi all, this series improves the use_mm / unuse_mm interface by better documenting the assumptions, and my taking the set_fs manipulations spread over the callers into the core API.

Hi all, this series improves the use_mm / unuse_mm interface by better documenting the assumptions, and my taking the set_fs manipulations spread over the callers into the core API.