From: Nick Desaulniers <ndesaulniers@google.com>
To: Will Deacon <will@kernel.org>
Cc: Android Kernel Team <kernel-team@android.com>,
Jan Kara <jack@suse.cz>, Minchan Kim <minchan@kernel.org>,
Catalin Marinas <catalin.marinas@arm.com>,
Hugh Dickins <hughd@google.com>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Linux-MM <linux-mm@kvack.org>,
Vinayak Menon <vinmenon@codeaurora.org>,
"Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>,
Andrew Morton <akpm@linux-foundation.org>,
Linus Torvalds <torvalds@linux-foundation.org>,
Linux ARM <linux-arm-kernel@lists.infradead.org>
Subject: Re: [RFC PATCH 4/8] mm: Separate fault info out of 'struct vm_fault'
Date: Thu, 14 Jan 2021 12:09:47 -0800 [thread overview]
Message-ID: <CAKwvOd=8-Ff+bC82pUMUHtZyKWm_1g7AKEO12GJ-kOTQ=0PXkw@mail.gmail.com> (raw)
In-Reply-To: <20210114194129.GA13314@willie-the-truck>
On Thu, Jan 14, 2021 at 11:41 AM Will Deacon <will@kernel.org> wrote:
>
> On Thu, Jan 14, 2021 at 11:09:01AM -0800, Linus Torvalds wrote:
> > On Thu, Jan 14, 2021 at 11:00 AM Will Deacon <will@kernel.org> wrote:
> > >
> > > I tried that initially, but I found that I had to make all of the
> > > members const to get it to work, at which point the anonymous struct
> > > wasn't really adding anything. Did I just botch the syntax?
> >
> > I'm not sure what you tried. But this stupid test-case sure works for me:
> >
> > struct hello {
> > const struct {
> > unsigned long address;
> > };
> > unsigned int flags;
> > };
> >
> > extern int fn(struct hello *);
> >
> > int test(void)
> > {
> > struct hello a = {
> > .address = 1,
> > };
> > a.flags = 0;
> > return fn(&a);
> > }
> >
> > and because "address" is in that unnamed constant struct, you can only
> > set it within that initializer, and cannot do
> >
> > a.address = 0;
> >
> > without an error (the way you _can_ do "a.flags = 0").
> >
> > I don't see naming the struct making a difference - apart from forcing
> > that big rename patch, of course.
> >
> > But maybe we're talking about different issues?
>
> Urgh...
>
> We _are_ both on the same page, and your reply above had me thinking I've
> lost the plot, so I went back to the start. Check out v5.11-rc3 and apply
> this patch:
>
>
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index ecdf8a8cd6ae..1eb950865450 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -514,11 +514,14 @@ static inline bool fault_flag_allow_retry_first(unsigned int flags)
> * pgoff should be used in favour of virtual_address, if possible.
> */
> struct vm_fault {
> - struct vm_area_struct *vma; /* Target VMA */
> + const struct {
> + struct vm_area_struct *vma; /* Target VMA */
> + gfp_t gfp_mask; /* gfp mask to be used for allocations */
> + pgoff_t pgoff; /* Logical page offset based on vma */
> + unsigned long address; /* Faulting virtual address */
> + };
> +
> unsigned int flags; /* FAULT_FLAG_xxx flags */
> - gfp_t gfp_mask; /* gfp mask to be used for allocations */
> - pgoff_t pgoff; /* Logical page offset based on vma */
> - unsigned long address; /* Faulting virtual address */
> pmd_t *pmd; /* Pointer to pmd entry matching
> * the 'address' */
> pud_t *pud; /* Pointer to pud entry matching
>
>
> Sure enough, an arm64 defconfig builds perfectly alright with that change,
> but it really shouldn't. I'm using clang 11.0.5, so I had another go with
> GCC 9.2.1 and bang:
>
> mm/filemap.c: In function ‘filemap_map_pages’:
> mm/filemap.c:2963:16: error: assignment of member ‘address’ in read-only object
> 2963 | vmf->address += (xas.xa_index - last_pgoff) << PAGE_SHIFT;
> | ^~
> make[1]: *** [scripts/Makefile.build:279: mm/filemap.o] Error 1
> make[1]: *** Waiting for unfinished jobs....
> make: *** [Makefile:1805: mm] Error 2
> make: *** Waiting for unfinished jobs....
>
> Nick -- any clue what's happening here? We would like that const anonymous
> struct to behave like a const struct member, as the alternative (naming the
> thing) results in a lot of refactoring churn.
Weird, looks like a bug to me in Clang, filed
https://bugs.llvm.org/show_bug.cgi?id=48755.
--
Thanks,
~Nick Desaulniers
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
next prev parent reply other threads:[~2021-01-14 20:11 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-14 17:59 [PATCH v3 0/8] Create 'old' ptes for faultaround mappings on arm64 with hardware access flag Will Deacon
2021-01-14 17:59 ` [PATCH v3 1/8] mm: Cleanup faultaround and finish_fault() codepaths Will Deacon
2021-02-09 20:24 ` Guenter Roeck
2021-02-10 11:44 ` Will Deacon
2021-02-10 14:57 ` Guenter Roeck
2021-01-14 17:59 ` [PATCH v3 2/8] mm: Allow architectures to request 'old' entries when prefaulting Will Deacon
2021-01-14 17:59 ` [PATCH v3 3/8] arm64: mm: Implement arch_wants_old_prefaulted_pte() Will Deacon
2021-01-14 17:59 ` [RFC PATCH 5/8] mm: Pass 'address' to map to do_set_pte() and drop FAULT_FLAG_PREFAULT Will Deacon
2021-01-14 18:17 ` Linus Torvalds
2021-01-15 10:24 ` Will Deacon
2021-01-14 17:59 ` [RFC PATCH 6/8] mm: Avoid modifying vmf.info.address in __collapse_huge_page_swapin() Will Deacon
2021-01-14 17:59 ` [RFC PATCH 7/8] mm: Use static initialisers for 'info' field of 'struct vm_fault' Will Deacon
2021-01-14 17:59 ` [RFC PATCH 8/8] mm: Mark 'info' field of 'struct vm_fault' as 'const' Will Deacon
[not found] ` <20210114175934.13070-5-will@kernel.org>
2021-01-14 18:16 ` [RFC PATCH 4/8] mm: Separate fault info out of 'struct vm_fault' Linus Torvalds
2021-01-14 19:00 ` Will Deacon
2021-01-14 19:09 ` Linus Torvalds
2021-01-14 19:41 ` Will Deacon
2021-01-14 20:09 ` Nick Desaulniers [this message]
2021-01-14 21:11 ` Linus Torvalds
2021-01-15 9:23 ` Will Deacon
2021-01-15 21:32 ` Linus Torvalds
2021-01-20 0:00 ` Nick Desaulniers
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAKwvOd=8-Ff+bC82pUMUHtZyKWm_1g7AKEO12GJ-kOTQ=0PXkw@mail.gmail.com' \
--to=ndesaulniers@google.com \
--cc=akpm@linux-foundation.org \
--cc=catalin.marinas@arm.com \
--cc=hughd@google.com \
--cc=jack@suse.cz \
--cc=kernel-team@android.com \
--cc=kirill.shutemov@linux.intel.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=minchan@kernel.org \
--cc=torvalds@linux-foundation.org \
--cc=vinmenon@codeaurora.org \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).