linux-kernel - Re: [PATCH] mm/mmap: Define index macros for protection

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <YVKdH4G5Alfwjkix@infradead.org>
Date:   Tue, 28 Sep 2021 05:42:07 +0100
From:   Christoph Hellwig <hch@...radead.org>
To:     Anshuman Khandual <anshuman.khandual@....com>
Cc:     Christoph Hellwig <hch@...radead.org>, linux-mm@...ck.org,
        Thomas Bogendoerfer <tsbogend@...ha.franken.de>,
        "David S. Miller" <davem@...emloft.net>,
        Andrew Morton <akpm@...ux-foundation.org>,
        linux-mips@...r.kernel.org, sparclinux@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] mm/mmap: Define index macros for protection_map[]

On Tue, Sep 28, 2021 at 08:24:43AM +0530, Anshuman Khandual wrote:
> > simple switch statement provided by each architecture.  See the below
> > WIP which just works for x86 and without pagetable debugging for where I
> > think we should be going.
> 
> Sure, this will work as well but all platforms need to be changed at once.
> Is there any platform that would not subscribe ARCH_HAS_GET_PAGE_PROT and
> export its own vm_get_page_prot() ? AFAICS all platforms are required to
> export __PXXX and __SXXX elements currently.
> 
> This seems to be a better idea than the current proposal. Probably all the
> vm_flags combinations, which will be used in those switch statements can be
> converted into macros just to improve readability. Are you planning to send
> this as a proper patch soon ?

This was just a quіck WIP patch.  If you have some spare time to tackle
it for real I'd sugget the following approach:

 1) Remove the direct references to protection_map in debug_vm_pgtable.c
 2) add the ARCH_HAS_GET_PAGE_PROT symbol that lets architectures
    provide vm_get_page_prot itself and not define protection_map at all
    in this case
 3) convert all architectures that touch protection_map to provide
    vm_get_page_prot themselves
 4) mark protection_map static
 5) convert all architectures that provide arch_filter_pgprot and/or
    arch_vm_get_page_prot to provide vm_get_page_prot directly and
    remove those hooks
 6) remove the __S???/__P??? macros and the generic vm_get_page_prot
    after providing an arch implementation for every architecture.
    This can maybe simplified with a new generic version that directly
    looks at PAGE_* macros, but that will need further investigation
    first.