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 kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by smtp.lore.kernel.org (Postfix) with ESMTP id 46DEDC433EF for ; Mon, 25 Apr 2022 11:34:22 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id C9D456B0074; Mon, 25 Apr 2022 07:34:21 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id C4C676B0075; Mon, 25 Apr 2022 07:34:21 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id AECDC6B0078; Mon, 25 Apr 2022 07:34:21 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (relay.a.hostedemail.com [64.99.140.24]) by kanga.kvack.org (Postfix) with ESMTP id 9D2EA6B0074 for ; Mon, 25 Apr 2022 07:34:21 -0400 (EDT) Received: from smtpin05.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay09.hostedemail.com (Postfix) with ESMTP id 4AC9826E01 for ; Mon, 25 Apr 2022 11:34:21 +0000 (UTC) X-FDA: 79395193122.05.CDEDABC Received: from szxga02-in.huawei.com (szxga02-in.huawei.com [45.249.212.188]) by imf22.hostedemail.com (Postfix) with ESMTP id 8A0D6C002B for ; Mon, 25 Apr 2022 11:34:19 +0000 (UTC) Received: from kwepemi500024.china.huawei.com (unknown [172.30.72.57]) by szxga02-in.huawei.com (SkyGuard) with ESMTP id 4Kn2w93PKFzhYjF; Mon, 25 Apr 2022 19:34:01 +0800 (CST) Received: from kwepemm600017.china.huawei.com (7.193.23.234) by kwepemi500024.china.huawei.com (7.221.188.100) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2375.24; Mon, 25 Apr 2022 19:34:16 +0800 Received: from [10.174.179.234] (10.174.179.234) by kwepemm600017.china.huawei.com (7.193.23.234) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2375.24; Mon, 25 Apr 2022 19:34:15 +0800 Message-ID: <8ce3eb46-2b18-e268-1013-d713c4c8f90b@huawei.com> Date: Mon, 25 Apr 2022 19:34:14 +0800 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101 Thunderbird/91.8.0 Subject: Re: [PATCH -next v5 3/5] mm: page_table_check: add hooks to public helpers To: Anshuman Khandual , Thomas Gleixner , Ingo Molnar , Borislav Petkov , Dave Hansen , , "H. Peter Anvin" , Pasha Tatashin , Andrew Morton , Catalin Marinas , Will Deacon , Paul Walmsley , Palmer Dabbelt , Albert Ou CC: , , , , Kefeng Wang , Guohanjun References: <20220421082042.1167967-1-tongtiangen@huawei.com> <20220421082042.1167967-4-tongtiangen@huawei.com> <7fc56f41-a3c2-76be-7a20-dda392f3c4fc@arm.com> <0f4de9db-1e36-b9d8-bd94-6e3ec3842940@arm.com> From: Tong Tiangen In-Reply-To: <0f4de9db-1e36-b9d8-bd94-6e3ec3842940@arm.com> Content-Type: text/plain; charset="UTF-8"; format=flowed Content-Transfer-Encoding: 8bit X-Originating-IP: [10.174.179.234] X-ClientProxiedBy: dggems706-chm.china.huawei.com (10.3.19.183) To kwepemm600017.china.huawei.com (7.193.23.234) X-CFilter-Loop: Reflected X-Rspamd-Server: rspam04 X-Rspamd-Queue-Id: 8A0D6C002B X-Stat-Signature: n19r6azs3qw1cje6yhp4bjtf1oxpwp5y X-Rspam-User: Authentication-Results: imf22.hostedemail.com; dkim=none; dmarc=pass (policy=quarantine) header.from=huawei.com; spf=pass (imf22.hostedemail.com: domain of tongtiangen@huawei.com designates 45.249.212.188 as permitted sender) smtp.mailfrom=tongtiangen@huawei.com X-HE-Tag: 1650886459-539426 X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: 在 2022/4/25 13:52, Anshuman Khandual 写道: > > > On 4/24/22 09:40, Tong Tiangen wrote: >> >> >> 在 2022/4/22 14:05, Anshuman Khandual 写道: >>> >>> >>> On 4/21/22 13:50, Tong Tiangen wrote: >>>> Move ptep_clear() to the include/linux/pgtable.h and add page table check >>>> relate hooks to some helpers, it's prepare for support page table check >>>> feature on new architecture. >>> >>> Could instrumenting generic page table helpers (fallback instances when its >>> corresponding __HAVE_ARCH_XXX is not defined on the platform), might add all >>> the page table check hooks into paths on platforms which have not subscribed >>> ARCH_SUPPORTS_PAGE_TABLE_CHECK in the first place ? Although these looks have >>> !CONFIG_PAGE_TABLE_CHECK fallback stubs in the header, hence a build problem >>> gets avoided. >> >> Right, build problems are avoided by fallback stubs in the header file. > > Although there might not be a build problem as such, but should non subscribing > platforms get their page table helpers instrumented with page table check hooks > in the first place ? The commit message should address these questions. > Add description in commit message to explain that: Non subscription platforms will call a fallback ptc stubs when getting their page table helpers[1] in include/linux/pgtable.h. [1] ptep_clear/ptep_get_and_clear/pmdp_huge_get_and_clear/pudp_huge_get_and_clear Am i right? :) Thanks, Tong. >> >>> >>>> >>>> Signed-off-by: Tong Tiangen >>>> Acked-by: Pasha Tatashin >>>> --- >>>>   arch/x86/include/asm/pgtable.h | 10 ---------- >>>>   include/linux/pgtable.h        | 26 ++++++++++++++++++-------- >>>>   2 files changed, 18 insertions(+), 18 deletions(-) >>>> >>>> diff --git a/arch/x86/include/asm/pgtable.h b/arch/x86/include/asm/pgtable.h >>>> index 564abe42b0f7..51cd39858f81 100644 >>>> --- a/arch/x86/include/asm/pgtable.h >>>> +++ b/arch/x86/include/asm/pgtable.h >>>> @@ -1073,16 +1073,6 @@ static inline pte_t ptep_get_and_clear_full(struct mm_struct *mm, >>>>       return pte; >>>>   } >>>>   -#define __HAVE_ARCH_PTEP_CLEAR >>> >>> AFICS X86 is the only platform subscribing __HAVE_ARCH_PTEP_CLEAR. Hence if >>> this is getting dropped for generic ptep_clear(), then no need to add back >>> #ifnded __HAVE_ARCH_PTEP_CLEAR construct. Generic ptep_clear() is the only >>> definition for all platforms ? >>> >>> Also if this patch is trying to drop off __HAVE_ARCH_PTEP_CLEAR along with >>> other page table check related changes, it needs to be done via a separate >>> patch instead. >> >> Agreed. >> IMO, this fix can be patched later. >> >>> >>>> -static inline void ptep_clear(struct mm_struct *mm, unsigned long addr, >>>> -                  pte_t *ptep) >>>> -{ >>>> -    if (IS_ENABLED(CONFIG_PAGE_TABLE_CHECK)) >>>> -        ptep_get_and_clear(mm, addr, ptep); >>>> -    else >>>> -        pte_clear(mm, addr, ptep); >>>> -} >>>> - >>>>   #define __HAVE_ARCH_PTEP_SET_WRPROTECT >>>>   static inline void ptep_set_wrprotect(struct mm_struct *mm, >>>>                         unsigned long addr, pte_t *ptep) >>>> diff --git a/include/linux/pgtable.h b/include/linux/pgtable.h >>>> index 49ab8ee2d6d7..10d2d91edf20 100644 >>>> --- a/include/linux/pgtable.h >>>> +++ b/include/linux/pgtable.h >>>> @@ -12,6 +12,7 @@ >>>>   #include >>>>   #include >>>>   #include >>>> +#include >>>>     #if 5 - defined(__PAGETABLE_P4D_FOLDED) - defined(__PAGETABLE_PUD_FOLDED) - \ >>>>       defined(__PAGETABLE_PMD_FOLDED) != CONFIG_PGTABLE_LEVELS >>>> @@ -272,14 +273,6 @@ static inline bool arch_has_hw_pte_young(void) >>>>   } >>>>   #endif >>>>   -#ifndef __HAVE_ARCH_PTEP_CLEAR >>>> -static inline void ptep_clear(struct mm_struct *mm, unsigned long addr, >>>> -                  pte_t *ptep) >>>> -{ >>>> -    pte_clear(mm, addr, ptep); >>>> -} >>>> -#endif >>>> - >>>>   #ifndef __HAVE_ARCH_PTEP_GET_AND_CLEAR >>>>   static inline pte_t ptep_get_and_clear(struct mm_struct *mm, >>>>                          unsigned long address, >>>> @@ -287,10 +280,22 @@ static inline pte_t ptep_get_and_clear(struct mm_struct *mm, >>>>   { >>>>       pte_t pte = *ptep; >>>>       pte_clear(mm, address, ptep); >>>> +    page_table_check_pte_clear(mm, address, pte); >>>>       return pte; >>>>   } >>>>   #endif >>>>   +#ifndef __HAVE_ARCH_PTEP_CLEAR >>>> +static inline void ptep_clear(struct mm_struct *mm, unsigned long addr, >>>> +                  pte_t *ptep) >>>> +{ >>>> +    if (IS_ENABLED(CONFIG_PAGE_TABLE_CHECK)) >>>> +        ptep_get_and_clear(mm, addr, ptep); >>>> +    else >>>> +        pte_clear(mm, addr, ptep); >>> >>> Could not this be reworked to avoid IS_ENABLED() ? This is confusing. If the page >>> table hooks can be added to all potential page table paths via generic helpers, >>> irrespective of CONFIG_PAGE_TABLE_CHECK option, there is no rationale for doing >>> a IS_ENABLED() check here. >>> >> >> From the perspective of code logic, we need to check the pte before being cleared. Whether pte check is required depends on IS_ENABLED(). >> >> Are there any suggestions for better implementation? > > But other generic page table helpers already have page table check hooks > instrumented without IS_ENABLED() checks, then why this is any different. > Maybe i understand what you said, the more reasonable implement is: static inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long address, pte_t *ptep) { pte_t pte = *ptep; pte_clear(mm, address, ptep); page_table_check_pte_clear(mm, address, pte); return pte; } static inline void ptep_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep) { ptep_get_and_clear(mm, address, ptep); } >> >> Thank you, >> Tong. >> >>>> +} >>>> +#endif >>>> + >>>>   #ifndef __HAVE_ARCH_PTEP_GET >>>>   static inline pte_t ptep_get(pte_t *ptep) >>>>   { >>>> @@ -360,7 +365,10 @@ static inline pmd_t pmdp_huge_get_and_clear(struct mm_struct *mm, >>>>                           pmd_t *pmdp) >>>>   { >>>>       pmd_t pmd = *pmdp; >>>> + >>>>       pmd_clear(pmdp); >>>> +    page_table_check_pmd_clear(mm, address, pmd); >>>> + >>>>       return pmd; >>>>   } >>>>   #endif /* __HAVE_ARCH_PMDP_HUGE_GET_AND_CLEAR */ >>>> @@ -372,6 +380,8 @@ static inline pud_t pudp_huge_get_and_clear(struct mm_struct *mm, >>>>       pud_t pud = *pudp; >>>>         pud_clear(pudp); >>>> +    page_table_check_pud_clear(mm, address, pud); >>>> + >>>>       return pud; >>>>   } >>>>   #endif /* __HAVE_ARCH_PUDP_HUGE_GET_AND_CLEAR */ >>> . > .