From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751658Ab3GPFrf (ORCPT ); Tue, 16 Jul 2013 01:47:35 -0400 Received: from e23smtp09.au.ibm.com ([202.81.31.142]:55849 "EHLO e23smtp09.au.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751400Ab3GPFrd (ORCPT ); Tue, 16 Jul 2013 01:47:33 -0400 From: "Aneesh Kumar K.V" To: Joonsoo Kim Cc: Andrew Morton , Rik van Riel , Mel Gorman , Michal Hocko , KAMEZAWA Hiroyuki , Hugh Dickins , Davidlohr Bueso , David Gibson , linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH 7/9] mm, hugetlb: add VM_NORESERVE check in vma_has_reserves() In-Reply-To: <20130716021245.GI2430@lge.com> References: <1373881967-16153-1-git-send-email-iamjoonsoo.kim@lge.com> <1373881967-16153-8-git-send-email-iamjoonsoo.kim@lge.com> <87li57j1tb.fsf@linux.vnet.ibm.com> <20130716021245.GI2430@lge.com> User-Agent: Notmuch/0.15.2+167~g5306b2b (http://notmuchmail.org) Emacs/24.3.50.1 (x86_64-unknown-linux-gnu) Date: Tue, 16 Jul 2013 11:17:23 +0530 Message-ID: <874nbvhx90.fsf@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: text/plain X-Content-Scanned: Fidelis XPS MAILER x-cbid: 13071616-3568-0000-0000-000003EE831E Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Joonsoo Kim writes: > On Mon, Jul 15, 2013 at 08:41:12PM +0530, Aneesh Kumar K.V wrote: >> Joonsoo Kim writes: >> >> > If we map the region with MAP_NORESERVE and MAP_SHARED, >> > we can skip to check reserve counting and eventually we cannot be ensured >> > to allocate a huge page in fault time. >> > With following example code, you can easily find this situation. >> > >> > Assume 2MB, nr_hugepages = 100 >> > >> > fd = hugetlbfs_unlinked_fd(); >> > if (fd < 0) >> > return 1; >> > >> > size = 200 * MB; >> > flag = MAP_SHARED; >> > p = mmap(NULL, size, PROT_READ|PROT_WRITE, flag, fd, 0); >> > if (p == MAP_FAILED) { >> > fprintf(stderr, "mmap() failed: %s\n", strerror(errno)); >> > return -1; >> > } >> > >> > size = 2 * MB; >> > flag = MAP_ANONYMOUS | MAP_SHARED | MAP_HUGETLB | MAP_NORESERVE; >> > p = mmap(NULL, size, PROT_READ|PROT_WRITE, flag, -1, 0); >> > if (p == MAP_FAILED) { >> > fprintf(stderr, "mmap() failed: %s\n", strerror(errno)); >> > } >> > p[0] = '0'; >> > sleep(10); >> > >> > During executing sleep(10), run 'cat /proc/meminfo' on another process. >> > You'll find a mentioned problem. >> > >> > Solution is simple. We should check VM_NORESERVE in vma_has_reserves(). >> > This prevent to use a pre-allocated huge page if free count is under >> > the reserve count. >> >> You have a problem with this patch, which i guess you are fixing in >> patch 9. Consider two process >> >> a) MAP_SHARED on fd >> b) MAP_SHARED | MAP_NORESERVE on fd >> >> We should allow the (b) to access the page even if VM_NORESERVE is set >> and we are out of reserve space . > > I can't get your point. > Please elaborate more on this. One process mmap with MAP_SHARED and another one with MAP_SHARED | MAP_NORESERVE Now the first process will result in reserving the pages from the hugtlb pool. Now if the second process try to dequeue huge page and we don't have free space we will fail because vma_has_reservers will now return zero because VM_NORESERVE is set and we can have (h->free_huge_pages - h->resv_huge_pages) == 0; The below hunk in your patch 9 handles that + if (vma->vm_flags & VM_NORESERVE) { + /* + * This address is already reserved by other process(chg == 0), + * so, we should decreament reserved count. Without + * decreamenting, reserve count is remained after releasing + * inode, because this allocated page will go into page cache + * and is regarded as coming from reserved pool in releasing + * step. Currently, we don't have any other solution to deal + * with this situation properly, so add work-around here. + */ + if (vma->vm_flags & VM_MAYSHARE && chg == 0) + return 1; + else + return 0; + } so may be both of these should be folded ? -aneesh From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from psmtp.com (na3sys010amx185.postini.com [74.125.245.185]) by kanga.kvack.org (Postfix) with SMTP id 306246B0032 for ; Tue, 16 Jul 2013 01:47:35 -0400 (EDT) Received: from /spool/local by e23smtp07.au.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Tue, 16 Jul 2013 15:35:16 +1000 Received: from d23relay04.au.ibm.com (d23relay04.au.ibm.com [9.190.234.120]) by d23dlp03.au.ibm.com (Postfix) with ESMTP id DA6AF3578053 for ; Tue, 16 Jul 2013 15:47:28 +1000 (EST) Received: from d23av03.au.ibm.com (d23av03.au.ibm.com [9.190.234.97]) by d23relay04.au.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id r6G5WDak12714028 for ; Tue, 16 Jul 2013 15:32:14 +1000 Received: from d23av03.au.ibm.com (loopback [127.0.0.1]) by d23av03.au.ibm.com (8.14.4/8.13.1/NCO v10.0 AVout) with ESMTP id r6G5lQUn029483 for ; Tue, 16 Jul 2013 15:47:27 +1000 From: "Aneesh Kumar K.V" Subject: Re: [PATCH 7/9] mm, hugetlb: add VM_NORESERVE check in vma_has_reserves() In-Reply-To: <20130716021245.GI2430@lge.com> References: <1373881967-16153-1-git-send-email-iamjoonsoo.kim@lge.com> <1373881967-16153-8-git-send-email-iamjoonsoo.kim@lge.com> <87li57j1tb.fsf@linux.vnet.ibm.com> <20130716021245.GI2430@lge.com> Date: Tue, 16 Jul 2013 11:17:23 +0530 Message-ID: <874nbvhx90.fsf@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: text/plain Sender: owner-linux-mm@kvack.org List-ID: To: Joonsoo Kim Cc: Andrew Morton , Rik van Riel , Mel Gorman , Michal Hocko , KAMEZAWA Hiroyuki , Hugh Dickins , Davidlohr Bueso , David Gibson , linux-mm@kvack.org, linux-kernel@vger.kernel.org Joonsoo Kim writes: > On Mon, Jul 15, 2013 at 08:41:12PM +0530, Aneesh Kumar K.V wrote: >> Joonsoo Kim writes: >> >> > If we map the region with MAP_NORESERVE and MAP_SHARED, >> > we can skip to check reserve counting and eventually we cannot be ensured >> > to allocate a huge page in fault time. >> > With following example code, you can easily find this situation. >> > >> > Assume 2MB, nr_hugepages = 100 >> > >> > fd = hugetlbfs_unlinked_fd(); >> > if (fd < 0) >> > return 1; >> > >> > size = 200 * MB; >> > flag = MAP_SHARED; >> > p = mmap(NULL, size, PROT_READ|PROT_WRITE, flag, fd, 0); >> > if (p == MAP_FAILED) { >> > fprintf(stderr, "mmap() failed: %s\n", strerror(errno)); >> > return -1; >> > } >> > >> > size = 2 * MB; >> > flag = MAP_ANONYMOUS | MAP_SHARED | MAP_HUGETLB | MAP_NORESERVE; >> > p = mmap(NULL, size, PROT_READ|PROT_WRITE, flag, -1, 0); >> > if (p == MAP_FAILED) { >> > fprintf(stderr, "mmap() failed: %s\n", strerror(errno)); >> > } >> > p[0] = '0'; >> > sleep(10); >> > >> > During executing sleep(10), run 'cat /proc/meminfo' on another process. >> > You'll find a mentioned problem. >> > >> > Solution is simple. We should check VM_NORESERVE in vma_has_reserves(). >> > This prevent to use a pre-allocated huge page if free count is under >> > the reserve count. >> >> You have a problem with this patch, which i guess you are fixing in >> patch 9. Consider two process >> >> a) MAP_SHARED on fd >> b) MAP_SHARED | MAP_NORESERVE on fd >> >> We should allow the (b) to access the page even if VM_NORESERVE is set >> and we are out of reserve space . > > I can't get your point. > Please elaborate more on this. One process mmap with MAP_SHARED and another one with MAP_SHARED | MAP_NORESERVE Now the first process will result in reserving the pages from the hugtlb pool. Now if the second process try to dequeue huge page and we don't have free space we will fail because vma_has_reservers will now return zero because VM_NORESERVE is set and we can have (h->free_huge_pages - h->resv_huge_pages) == 0; The below hunk in your patch 9 handles that + if (vma->vm_flags & VM_NORESERVE) { + /* + * This address is already reserved by other process(chg == 0), + * so, we should decreament reserved count. Without + * decreamenting, reserve count is remained after releasing + * inode, because this allocated page will go into page cache + * and is regarded as coming from reserved pool in releasing + * step. Currently, we don't have any other solution to deal + * with this situation properly, so add work-around here. + */ + if (vma->vm_flags & VM_MAYSHARE && chg == 0) + return 1; + else + return 0; + } so may be both of these should be folded ? -aneesh -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org