From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id EB5B71C29 for ; Wed, 23 Nov 2022 09:30:02 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id C799EC433C1; Wed, 23 Nov 2022 09:30:01 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1669195802; bh=PyvGfMNO4eGVZEKytnxM1XElbL0QDh3Vv3c9PMIsKIE=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=grit8xvfaXM4qoUM3vK+Df6VZA8iiYcaKmxm0xwtV63FAPM72d2TViaLEMHB8iz99 iAa1oGRMFE7h0tkZlC3kWnl8Q/foBY2Km8wQS0XGOFqvMfVQTHoG7M6ii1ObHdZSk5 hVUKP1nqHspXqUUYpjTnvA212209UP/9EcH9PN24= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, James Houghton , Mike Kravetz , Naoya Horiguchi , Yang Shi , Axel Rasmussen , Miaohe Lin , Muchun Song , Andrew Morton , Sasha Levin Subject: [PATCH 5.15 037/181] hugetlbfs: dont delete error page from pagecache Date: Wed, 23 Nov 2022 09:50:00 +0100 Message-Id: <20221123084603.965798988@linuxfoundation.org> X-Mailer: git-send-email 2.38.1 In-Reply-To: <20221123084602.707860461@linuxfoundation.org> References: <20221123084602.707860461@linuxfoundation.org> User-Agent: quilt/0.67 Precedence: bulk X-Mailing-List: patches@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit From: James Houghton [ Upstream commit 8625147cafaa9ba74713d682f5185eb62cb2aedb ] This change is very similar to the change that was made for shmem [1], and it solves the same problem but for HugeTLBFS instead. Currently, when poison is found in a HugeTLB page, the page is removed from the page cache. That means that attempting to map or read that hugepage in the future will result in a new hugepage being allocated instead of notifying the user that the page was poisoned. As [1] states, this is effectively memory corruption. The fix is to leave the page in the page cache. If the user attempts to use a poisoned HugeTLB page with a syscall, the syscall will fail with EIO, the same error code that shmem uses. For attempts to map the page, the thread will get a BUS_MCEERR_AR SIGBUS. [1]: commit a76054266661 ("mm: shmem: don't truncate page if memory failure happens") Link: https://lkml.kernel.org/r/20221018200125.848471-1-jthoughton@google.com Signed-off-by: James Houghton Reviewed-by: Mike Kravetz Reviewed-by: Naoya Horiguchi Tested-by: Naoya Horiguchi Reviewed-by: Yang Shi Cc: Axel Rasmussen Cc: James Houghton Cc: Miaohe Lin Cc: Muchun Song Cc: Signed-off-by: Andrew Morton Signed-off-by: Sasha Levin --- fs/hugetlbfs/inode.c | 13 ++++++------- mm/hugetlb.c | 4 ++++ mm/memory-failure.c | 5 ++++- 3 files changed, 14 insertions(+), 8 deletions(-) diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c index d74a49b188c2..be8deec29ebe 100644 --- a/fs/hugetlbfs/inode.c +++ b/fs/hugetlbfs/inode.c @@ -361,6 +361,12 @@ static ssize_t hugetlbfs_read_iter(struct kiocb *iocb, struct iov_iter *to) } else { unlock_page(page); + if (PageHWPoison(page)) { + put_page(page); + retval = -EIO; + break; + } + /* * We have the page, copy it to user space buffer. */ @@ -984,13 +990,6 @@ static int hugetlbfs_migrate_page(struct address_space *mapping, static int hugetlbfs_error_remove_page(struct address_space *mapping, struct page *page) { - struct inode *inode = mapping->host; - pgoff_t index = page->index; - - remove_huge_page(page); - if (unlikely(hugetlb_unreserve_pages(inode, index, index + 1, 1))) - hugetlb_fix_reserve_counts(inode); - return 0; } diff --git a/mm/hugetlb.c b/mm/hugetlb.c index dbb63ec3b5fa..e7bd42f23667 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -5350,6 +5350,10 @@ int hugetlb_mcopy_atomic_pte(struct mm_struct *dst_mm, ptl = huge_pte_lockptr(h, dst_mm, dst_pte); spin_lock(ptl); + ret = -EIO; + if (PageHWPoison(page)) + goto out_release_unlock; + /* * Recheck the i_size after holding PT lock to make sure not * to leave any page mapped (as page_mapped()) beyond the end diff --git a/mm/memory-failure.c b/mm/memory-failure.c index 85b1a77e3a99..2ad0f4580091 100644 --- a/mm/memory-failure.c +++ b/mm/memory-failure.c @@ -1040,6 +1040,7 @@ static int me_huge_page(struct page_state *ps, struct page *p) int res; struct page *hpage = compound_head(p); struct address_space *mapping; + bool extra_pins = false; if (!PageHuge(hpage)) return MF_DELAYED; @@ -1047,6 +1048,8 @@ static int me_huge_page(struct page_state *ps, struct page *p) mapping = page_mapping(hpage); if (mapping) { res = truncate_error_page(hpage, page_to_pfn(p), mapping); + /* The page is kept in page cache. */ + extra_pins = true; unlock_page(hpage); } else { res = MF_FAILED; @@ -1064,7 +1067,7 @@ static int me_huge_page(struct page_state *ps, struct page *p) } } - if (has_extra_refcount(ps, p, false)) + if (has_extra_refcount(ps, p, extra_pins)) res = MF_FAILED; return res; -- 2.35.1