From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Morton Subject: [patch 099/155] mm/gup: allow VM_FAULT_RETRY for multiple times Date: Wed, 01 Apr 2020 21:08:49 -0700 Message-ID: <20200402040849.ezjm4X9v0%akpm@linux-foundation.org> References: <20200401210155.09e3b9742e1c6e732f5a7250@linux-foundation.org> Reply-To: linux-kernel@vger.kernel.org Return-path: Received: from mail.kernel.org ([198.145.29.99]:60550 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727746AbgDBEIv (ORCPT ); Thu, 2 Apr 2020 00:08:51 -0400 In-Reply-To: <20200401210155.09e3b9742e1c6e732f5a7250@linux-foundation.org> Sender: mm-commits-owner@vger.kernel.org List-Id: mm-commits@vger.kernel.org To: aarcange@redhat.com, akpm@linux-foundation.org, bgeffon@google.com, bobbypowers@gmail.com, cracauer@cons.org, david@redhat.com, dgilbert@redhat.com, dplotnikov@virtuozzo.com, gokhale2@llnl.gov, hannes@cmpxchg.org, hughd@google.com, jglisse@redhat.com, kirill@shutemov.name, linux-mm@kvack.org, mcfadden8@llnl.gov, mgorman@suse.de, mike.kravetz@oracle.com, mm-commits@vger.kernel.org, peterx@redhat.com, rppt@linux.vnet.ibm.com, torvalds@linux-foundation.org, willy@infradead.org, xemul@openvz.org From: Peter Xu Subject: mm/gup: allow VM_FAULT_RETRY for multiple times This is the gup counterpart of the change that allows the VM_FAULT_RETRY to happen for more than once. One thing to mention is that we must check the fatal signal here before retry because the GUP can be interrupted by that, otherwise we can loop forever. Link: http://lkml.kernel.org/r/20200220195357.16371-1-peterx@redhat.com Signed-off-by: Peter Xu Tested-by: Brian Geffon Cc: Andrea Arcangeli Cc: Bobby Powers Cc: David Hildenbrand Cc: Denis Plotnikov Cc: "Dr . David Alan Gilbert" Cc: Hugh Dickins Cc: Jerome Glisse Cc: Johannes Weiner Cc: "Kirill A . Shutemov" Cc: Linus Torvalds Cc: Martin Cracauer Cc: Marty McFadden Cc: Matthew Wilcox Cc: Maya Gokhale Cc: Mel Gorman Cc: Mike Kravetz Cc: Mike Rapoport Cc: Pavel Emelyanov Signed-off-by: Andrew Morton --- mm/gup.c | 27 +++++++++++++++++++++------ mm/hugetlb.c | 6 ++++-- 2 files changed, 25 insertions(+), 8 deletions(-) --- a/mm/gup.c~mm-gup-allow-vm_fault_retry-for-multiple-times +++ a/mm/gup.c @@ -868,7 +868,10 @@ static int faultin_page(struct task_stru if (*flags & FOLL_NOWAIT) fault_flags |= FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_RETRY_NOWAIT; if (*flags & FOLL_TRIED) { - VM_WARN_ON_ONCE(fault_flags & FAULT_FLAG_ALLOW_RETRY); + /* + * Note: FAULT_FLAG_ALLOW_RETRY and FAULT_FLAG_TRIED + * can co-exist + */ fault_flags |= FAULT_FLAG_TRIED; } @@ -1228,7 +1231,6 @@ retry: down_read(&mm->mmap_sem); if (!(fault_flags & FAULT_FLAG_TRIED)) { *unlocked = true; - fault_flags &= ~FAULT_FLAG_ALLOW_RETRY; fault_flags |= FAULT_FLAG_TRIED; goto retry; } @@ -1312,17 +1314,30 @@ static __always_inline long __get_user_p if (likely(pages)) pages += ret; start += ret << PAGE_SHIFT; + lock_dropped = true; +retry: /* * Repeat on the address that fired VM_FAULT_RETRY - * without FAULT_FLAG_ALLOW_RETRY but with - * FAULT_FLAG_TRIED. + * with both FAULT_FLAG_ALLOW_RETRY and + * FAULT_FLAG_TRIED. Note that GUP can be interrupted + * by fatal signals, so we need to check it before we + * start trying again otherwise it can loop forever. */ + + if (fatal_signal_pending(current)) + break; + *locked = 1; - lock_dropped = true; down_read(&mm->mmap_sem); + ret = __get_user_pages(tsk, mm, start, 1, flags | FOLL_TRIED, - pages, NULL, NULL); + pages, NULL, locked); + if (!*locked) { + /* Continue to retry until we succeeded */ + BUG_ON(ret != 0); + goto retry; + } if (ret != 1) { BUG_ON(ret > 1); if (!pages_done) --- a/mm/hugetlb.c~mm-gup-allow-vm_fault_retry-for-multiple-times +++ a/mm/hugetlb.c @@ -4349,8 +4349,10 @@ long follow_hugetlb_page(struct mm_struc fault_flags |= FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_RETRY_NOWAIT; if (flags & FOLL_TRIED) { - VM_WARN_ON_ONCE(fault_flags & - FAULT_FLAG_ALLOW_RETRY); + /* + * Note: FAULT_FLAG_ALLOW_RETRY and + * FAULT_FLAG_TRIED can co-exist + */ fault_flags |= FAULT_FLAG_TRIED; } ret = hugetlb_fault(mm, vma, vaddr, fault_flags); _