For shared mappings, the pointer to the hugetlb_cgroup to uncharge lives in the resv_map entries, in file_region->reservation_counter.
After a call to region_chg, we charge the approprate hugetlb_cgroup, and if successful, we pass on the hugetlb_cgroup info to a follow up region_add call. When a file_region entry is added to the resv_map via region_add, we put the pointer to that cgroup in file_region->reservation_counter. If charging doesn't succeed, we report the error to the caller, so that the kernel fails the reservation.
On region_del, which is when the hugetlb memory is unreserved, we also uncharge the file_region->reservation_counter.
Signed-off-by: Mina Almasry almasrymina@google.com
--- mm/hugetlb.c | 151 ++++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 119 insertions(+), 32 deletions(-)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 3d98e1b771390..cef15e68626bd 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -242,6 +242,15 @@ struct file_region { struct list_head link; long from; long to; +#ifdef CONFIG_CGROUP_HUGETLB + /* + * On shared mappings, each reserved region appears as a struct + * file_region in resv_map. These fields hold the info needed to + * uncharge each reservation. + */ + struct page_counter *reservation_counter; + unsigned long pages_per_hpage; +#endif };
/* Helper that removes a struct file_region from the resv_map cache and returns @@ -250,6 +259,23 @@ struct file_region { static struct file_region * get_file_region_entry_from_cache(struct resv_map *resv, long from, long to);
+/* Helper that records hugetlb_cgroup uncharge info. */ +static void record_hugetlb_cgroup_uncharge_info(struct hugetlb_cgroup *h_cg, + struct file_region *nrg, + struct hstate *h) +{ +#ifdef CONFIG_CGROUP_HUGETLB + if (h_cg) { + nrg->reservation_counter = + &h_cg->reserved_hugepage[hstate_index(h)]; + nrg->pages_per_hpage = pages_per_huge_page(h); + } else { + nrg->reservation_counter = NULL; + nrg->pages_per_hpage = 0; + } +#endif +} + /* Must be called with resv->lock held. Calling this with count_only == true * will count the number of pages to be added but will not modify the linked * list. If regions_needed != NULL and count_only == true, then regions_needed @@ -257,7 +283,9 @@ get_file_region_entry_from_cache(struct resv_map *resv, long from, long to); * add the regions for this range. */ static long add_reservation_in_range(struct resv_map *resv, long f, long t, - long *regions_needed, bool count_only) + struct hugetlb_cgroup *h_cg, + struct hstate *h, long *regions_needed, + bool count_only) { long add = 0; struct list_head *head = &resv->regions; @@ -296,6 +324,8 @@ static long add_reservation_in_range(struct resv_map *resv, long f, long t, if (!count_only) { nrg = get_file_region_entry_from_cache( resv, last_accounted_offset, rg->from); + record_hugetlb_cgroup_uncharge_info(h_cg, nrg, + h); list_add(&nrg->link, rg->link.prev); } else if (regions_needed) *regions_needed += 1; @@ -312,12 +342,14 @@ static long add_reservation_in_range(struct resv_map *resv, long f, long t, if (!count_only) { nrg = get_file_region_entry_from_cache( resv, last_accounted_offset, t); + record_hugetlb_cgroup_uncharge_info(h_cg, nrg, h); list_add(&nrg->link, rg->link.prev); } else if (regions_needed) *regions_needed += 1; last_accounted_offset = t; }
+ VM_BUG_ON(add < 0); return add; }
@@ -336,7 +368,8 @@ static long add_reservation_in_range(struct resv_map *resv, long f, long t, * region_add of regions of length 1 never allocate file_regions and cannot * fail. */ -static long region_add(struct resv_map *resv, long f, long t, +static long region_add(struct hstate *h, struct hugetlb_cgroup *h_cg, + struct resv_map *resv, long f, long t, long in_regions_needed) { long add = 0, actual_regions_needed = 0; @@ -346,7 +379,8 @@ static long region_add(struct resv_map *resv, long f, long t, retry:
/* Count how many regions are actually needed to execute this add. */ - add_reservation_in_range(resv, f, t, &actual_regions_needed, true); + add_reservation_in_range(resv, f, t, NULL, NULL, &actual_regions_needed, + true);
/* * Check for sufficient descriptors in the cache to accommodate @@ -372,7 +406,7 @@ static long region_add(struct resv_map *resv, long f, long t, goto retry; }
- add = add_reservation_in_range(resv, f, t, NULL, false); + add = add_reservation_in_range(resv, f, t, h_cg, h, NULL, false);
resv->adds_in_progress -= in_regions_needed;
@@ -410,7 +444,8 @@ static long region_chg(struct resv_map *resv, long f, long t, spin_lock(&resv->lock);
/* Count how many hugepages in this range are NOT respresented. */ - chg = add_reservation_in_range(resv, f, t, out_regions_needed, true); + chg = add_reservation_in_range(resv, f, t, NULL, NULL, + out_regions_needed, true);
if (*out_regions_needed < 1) *out_regions_needed = 1; @@ -459,6 +494,24 @@ static void region_abort(struct resv_map *resv, long f, long t, spin_unlock(&resv->lock); }
+static void uncharge_cgroup_if_shared_mapping(struct resv_map *resv, + struct file_region *rg, + unsigned long nr_pages) +{ +#ifdef CONFIG_CGROUP_HUGETLB + /* + * If resv->reservation_counter is NULL, then this is shared + * reservation, and the reserved memory is tracked in the file_struct + * entries inside of resv_map. So we need to uncharge the memory here. + */ + if (rg->reservation_counter && rg->pages_per_hpage && nr_pages > 0 && + !resv->reservation_counter) { + hugetlb_cgroup_uncharge_counter(rg->reservation_counter, + nr_pages * rg->pages_per_hpage); + } +#endif +} + /* * Delete the specified range [f, t) from the reserve map. If the * t parameter is LONG_MAX, this indicates that ALL regions after f @@ -528,6 +581,9 @@ static long region_del(struct resv_map *resv, long f, long t) /* Original entry is trimmed */ rg->to = f;
+ uncharge_cgroup_if_shared_mapping(resv, rg, + nrg->to - nrg->from); + list_add(&nrg->link, &rg->link); nrg = NULL; break; @@ -535,6 +591,8 @@ static long region_del(struct resv_map *resv, long f, long t)
if (f <= rg->from && t >= rg->to) { /* Remove entire region */ del += rg->to - rg->from; + uncharge_cgroup_if_shared_mapping(resv, rg, + rg->to - rg->from); list_del(&rg->link); kfree(rg); continue; @@ -543,14 +601,20 @@ static long region_del(struct resv_map *resv, long f, long t) if (f <= rg->from) { /* Trim beginning of region */ del += t - rg->from; rg->from = t; + + uncharge_cgroup_if_shared_mapping(resv, rg, + t - rg->from); } else { /* Trim end of region */ del += rg->to - f; rg->to = f; + + uncharge_cgroup_if_shared_mapping(resv, rg, rg->to - f); } }
spin_unlock(&resv->lock); kfree(nrg); + return del; }
@@ -1956,7 +2020,8 @@ __vma_reservation_common(struct hstate *h, struct vm_area_struct *vma, break; case VMA_COMMIT_RESV: VM_BUG_ON(in_regions_needed == -1); - ret = region_add(resv, idx, idx + 1, in_regions_needed); + ret = region_add(NULL, NULL, resv, idx, idx + 1, + in_regions_needed); /* region_add calls of range 1 should never fail. */ VM_BUG_ON(ret < 0); break; @@ -1968,7 +2033,8 @@ __vma_reservation_common(struct hstate *h, struct vm_area_struct *vma, case VMA_ADD_RESV: VM_BUG_ON(in_regions_needed == -1); if (vma->vm_flags & VM_MAYSHARE) { - ret = region_add(resv, idx, idx + 1, in_regions_needed); + ret = region_add(NULL, NULL, resv, idx, idx + 1, + in_regions_needed); /* region_add calls of range 1 should never fail. */ VM_BUG_ON(ret < 0); } else { @@ -4640,7 +4706,7 @@ int hugetlb_reserve_pages(struct inode *inode, struct hstate *h = hstate_inode(inode); struct hugepage_subpool *spool = subpool_inode(inode); struct resv_map *resv_map; - struct hugetlb_cgroup *h_cg; + struct hugetlb_cgroup *h_cg = NULL; long gbl_reserve, regions_needed = 0;
/* This should never happen */ @@ -4681,23 +4747,6 @@ int hugetlb_reserve_pages(struct inode *inode,
chg = to - from;
- if (hugetlb_cgroup_charge_cgroup(hstate_index(h), - chg * pages_per_huge_page(h), - &h_cg, true)) { - kref_put(&resv_map->refs, resv_map_release); - return -ENOMEM; - } - -#ifdef CONFIG_CGROUP_HUGETLB - /* - * Since this branch handles private mappings, we attach the - * counter to uncharge for this reservation off resv_map. - */ - resv_map->reservation_counter = - &h_cg->reserved_hugepage[hstate_index(h)]; - resv_map->pages_per_hpage = pages_per_huge_page(h); -#endif - set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } @@ -4707,6 +4756,25 @@ int hugetlb_reserve_pages(struct inode *inode, goto out_err; }
+ ret = hugetlb_cgroup_charge_cgroup( + hstate_index(h), chg * pages_per_huge_page(h), &h_cg, true); + + if (ret < 0) { + ret = -ENOMEM; + goto out_err; + } + +#ifdef CONFIG_CGROUP_HUGETLB + if (vma && !(vma->vm_flags & VM_MAYSHARE)) { + /* For private mappings, the hugetlb_cgroup uncharge info hangs + * of the resv_map. + */ + resv_map->reservation_counter = + &h_cg->reserved_hugepage[hstate_index(h)]; + resv_map->pages_per_hpage = pages_per_huge_page(h); + } +#endif + /* * There must be enough pages in the subpool for the mapping. If * the subpool has a minimum size, there may be some global @@ -4715,7 +4783,7 @@ int hugetlb_reserve_pages(struct inode *inode, gbl_reserve = hugepage_subpool_get_pages(spool, chg); if (gbl_reserve < 0) { ret = -ENOSPC; - goto out_err; + goto out_uncharge_cgroup; }
/* @@ -4724,9 +4792,7 @@ int hugetlb_reserve_pages(struct inode *inode, */ ret = hugetlb_acct_memory(h, gbl_reserve); if (ret < 0) { - /* put back original number of pages, chg */ - (void)hugepage_subpool_put_pages(spool, chg); - goto out_err; + goto out_put_pages; }
/* @@ -4741,7 +4807,7 @@ int hugetlb_reserve_pages(struct inode *inode, * else has to be done for private mappings here */ if (!vma || vma->vm_flags & VM_MAYSHARE) { - add = region_add(resv_map, from, to, regions_needed); + add = region_add(h, h_cg, resv_map, from, to, regions_needed);
if (unlikely(add < 0)) { hugetlb_acct_memory(h, -gbl_reserve); @@ -4758,12 +4824,33 @@ int hugetlb_reserve_pages(struct inode *inode, */ long rsv_adjust;
- rsv_adjust = hugepage_subpool_put_pages(spool, - chg - add); + hugetlb_cgroup_uncharge_cgroup( + hstate_index(h), + (chg - add) * pages_per_huge_page(h), h_cg, + true); + + rsv_adjust = + hugepage_subpool_put_pages(spool, chg - add); hugetlb_acct_memory(h, -rsv_adjust); } + } else { +#ifdef CONFIG_CGROUP_HUGETLB + /* + * Since this branch handles private mappings, we attach the + * counter to uncharge for this reservation off resv_map. + */ + resv_map->reservation_counter = + &h_cg->reserved_hugepage[hstate_index(h)]; + resv_map->pages_per_hpage = pages_per_huge_page(h); +#endif } return 0; +out_put_pages: + /* put back original number of pages, chg */ + (void)hugepage_subpool_put_pages(spool, chg); +out_uncharge_cgroup: + hugetlb_cgroup_uncharge_cgroup( + hstate_index(h), chg * pages_per_huge_page(h), h_cg, true); out_err: if (!vma || vma->vm_flags & VM_MAYSHARE) /* Only call region_abort if the region_chg succeeded but the -- 2.24.0.rc0.303.g954a862665-goog