From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 2B65CC43334 for ; Mon, 6 Jun 2022 15:30:10 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S240846AbiFFPaI (ORCPT ); Mon, 6 Jun 2022 11:30:08 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:40434 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S240797AbiFFPaG (ORCPT ); Mon, 6 Jun 2022 11:30:06 -0400 Received: from dfw.source.kernel.org (dfw.source.kernel.org [139.178.84.217]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id E0B6433A26 for ; Mon, 6 Jun 2022 08:30:04 -0700 (PDT) Received: from smtp.kernel.org (relay.kernel.org [52.25.139.140]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dfw.source.kernel.org (Postfix) with ESMTPS id 7B24C612D0 for ; Mon, 6 Jun 2022 15:30:04 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id ACB9BC385A9; Mon, 6 Jun 2022 15:30:02 +0000 (UTC) Date: Mon, 6 Jun 2022 16:29:59 +0100 From: Catalin Marinas To: Patrick Wang Cc: akpm@linux-foundation.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org, yee.lee@mediatek.com Subject: Re: [PATCH v2 4/4] mm: kmemleak: kmemleak_*_phys() set address type and check PA when scan Message-ID: References: <20220603035415.1243913-1-patrick.wang.shcn@gmail.com> <20220603035415.1243913-5-patrick.wang.shcn@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20220603035415.1243913-5-patrick.wang.shcn@gmail.com> Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Fri, Jun 03, 2022 at 11:54:15AM +0800, Patrick Wang wrote: > diff --git a/mm/kmemleak.c b/mm/kmemleak.c > index 246a70b7218f..62d1ad8f8a44 100644 > --- a/mm/kmemleak.c > +++ b/mm/kmemleak.c > @@ -1156,8 +1156,12 @@ EXPORT_SYMBOL(kmemleak_no_scan); > void __ref kmemleak_alloc_phys(phys_addr_t phys, size_t size, int min_count, > gfp_t gfp) > { > - if (PHYS_PFN(phys) >= min_low_pfn && PHYS_PFN(phys) < max_low_pfn) > - kmemleak_alloc(__va(phys), size, min_count, gfp); > + pr_debug("%s(0x%pa, %zu, %d)\n", __func__, &phys, size, min_count); > + > + if (kmemleak_enabled && !min_count) > + /* create object with OBJECT_PHYS flag */ > + create_object((unsigned long)phys, size, min_count, > + gfp, true); > } With an early patch, just drop min_count altogether from this API, assume 0. > EXPORT_SYMBOL(kmemleak_alloc_phys); > > @@ -1170,8 +1174,10 @@ EXPORT_SYMBOL(kmemleak_alloc_phys); > */ > void __ref kmemleak_free_part_phys(phys_addr_t phys, size_t size) > { > - if (PHYS_PFN(phys) >= min_low_pfn && PHYS_PFN(phys) < max_low_pfn) > - kmemleak_free_part(__va(phys), size); > + pr_debug("%s(0x%pa)\n", __func__, &phys); > + > + if (kmemleak_enabled) > + delete_object_part((unsigned long)phys, size, true); > } > EXPORT_SYMBOL(kmemleak_free_part_phys); > > @@ -1182,8 +1188,10 @@ EXPORT_SYMBOL(kmemleak_free_part_phys); > */ > void __ref kmemleak_not_leak_phys(phys_addr_t phys) > { > - if (PHYS_PFN(phys) >= min_low_pfn && PHYS_PFN(phys) < max_low_pfn) > - kmemleak_not_leak(__va(phys)); > + pr_debug("%s(0x%pa)\n", __func__, &phys); > + > + if (kmemleak_enabled) > + make_gray_object((unsigned long)phys, true); > } > EXPORT_SYMBOL(kmemleak_not_leak_phys); This function doesn't have any callers, so please remove it. > @@ -1194,8 +1202,10 @@ EXPORT_SYMBOL(kmemleak_not_leak_phys); > */ > void __ref kmemleak_ignore_phys(phys_addr_t phys) > { > - if (PHYS_PFN(phys) >= min_low_pfn && PHYS_PFN(phys) < max_low_pfn) > - kmemleak_ignore(__va(phys)); > + pr_debug("%s(0x%pa)\n", __func__, &phys); > + > + if (kmemleak_enabled) > + make_black_object((unsigned long)phys, true); > } > EXPORT_SYMBOL(kmemleak_ignore_phys); Ah, that's still in use and we do need make_black_object(), contrary to what I commented on a previous patch. We can still avoid changing make_gray_object(). (we could replace the kmemleak_ignore_phys() calls kmemleak_free_part_phys() but that's not in line with what we do for the virtual objects) > @@ -1468,6 +1478,17 @@ static void kmemleak_scan(void) > dump_object_info(object); > } > #endif > + > + /* ignore objects outside lowmem (paint them black) */ > + if ((object->flags & OBJECT_PHYS) && > + !(object->flags & OBJECT_NO_SCAN)) { > + unsigned long phys = object->pointer; > + > + if (PHYS_PFN(phys) < min_low_pfn || > + PHYS_PFN(phys + object->size) >= max_low_pfn) > + __paint_it(object, KMEMLEAK_BLACK); > + } > + > /* reset the reference count (whiten the object) */ > object->count = 0; > if (color_gray(object) && get_object(object)) This looks fine. -- Catalin