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 431BC137F for ; Fri, 1 Apr 2022 18:21:22 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 0AFB4C340EE; Fri, 1 Apr 2022 18:21:22 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linux-foundation.org; s=korg; t=1648837282; bh=vYYaHwg9vpXslMSd0ZFZlQDo++62jKaYBzliYeEXWkg=; h=Date:To:From:In-Reply-To:Subject:From; b=GozFqdhKKYkIw3R8FIKYuzDZb+VyC/g7gF8GjTzTU3JwO2TKFilRiJ/5wrHg3ZNEK aGvgbavKakWxT16JOePUjSEk5tzQ5IFDET9W3w1rVNesBx8cIV5wKdEDY8VR0SpcCZ /t2zpNaqolx1WwTy9VVfZfkAd7+rBvTx1dG6NiiY= Date: Fri, 01 Apr 2022 11:21:21 -0700 To: stable@vger.kernel.org,osalvador@suse.de,naoya.horiguchi@nec.com,mgorman@suse.de,linmiaohe@huawei.com,hannes@cmpxchg.org,riel@surriel.com,akpm@linux-foundation.org,patches@lists.linux.dev,linux-mm@kvack.org,mm-commits@vger.kernel.org,torvalds@linux-foundation.org,akpm@linux-foundation.org From: Andrew Morton In-Reply-To: Subject: [patch 11/16] mm,hwpoison: unmap poisoned page before invalidation Message-Id: <20220401182122.0AFB4C340EE@smtp.kernel.org> Precedence: bulk X-Mailing-List: patches@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: From: Rik van Riel Subject: mm,hwpoison: unmap poisoned page before invalidation In some cases it appears the invalidation of a hwpoisoned page fails because the page is still mapped in another process. This can cause a program to be continuously restarted and die when it page faults on the page that was not invalidated. Avoid that problem by unmapping the hwpoisoned page when we find it. Another issue is that sometimes we end up oopsing in finish_fault, if the code tries to do something with the now-NULL vmf->page. I did not hit this error when submitting the previous patch because there are several opportunities for alloc_set_pte to bail out before accessing vmf->page, and that apparently happened on those systems, and most of the time on other systems, too. However, across several million systems that error does occur a handful of times a day. It can be avoided by returning VM_FAULT_NOPAGE which will cause do_read_fault to return before calling finish_fault. Link: https://lkml.kernel.org/r/20220325161428.5068d97e@imladris.surriel.com Fixes: e53ac7374e64 ("mm: invalidate hwpoison page cache page in fault path") Signed-off-by: Rik van Riel Reviewed-by: Miaohe Lin Tested-by: Naoya Horiguchi Reviewed-by: Oscar Salvador Cc: Mel Gorman Cc: Johannes Weiner Cc: Signed-off-by: Andrew Morton --- mm/memory.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) --- a/mm/memory.c~mmhwpoison-unmap-poisoned-page-before-invalidation +++ a/mm/memory.c @@ -3918,14 +3918,18 @@ static vm_fault_t __do_fault(struct vm_f return ret; if (unlikely(PageHWPoison(vmf->page))) { + struct page *page = vmf->page; vm_fault_t poisonret = VM_FAULT_HWPOISON; if (ret & VM_FAULT_LOCKED) { + if (page_mapped(page)) + unmap_mapping_pages(page_mapping(page), + page->index, 1, false); /* Retry if a clean page was removed from the cache. */ - if (invalidate_inode_page(vmf->page)) - poisonret = 0; - unlock_page(vmf->page); + if (invalidate_inode_page(page)) + poisonret = VM_FAULT_NOPAGE; + unlock_page(page); } - put_page(vmf->page); + put_page(page); vmf->page = NULL; return poisonret; } _