linux-kernel - Re: [PATCH v3 5/8] mm: make folio_lock_fault indicate the state of mmap

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: <ZJsBEk4OHlp39vEK@x1n>
Date:   Tue, 27 Jun 2023 11:32:34 -0400
From:   Peter Xu <peterx@...hat.com>
To:     Suren Baghdasaryan <surenb@...gle.com>
Cc:     akpm@...ux-foundation.org, willy@...radead.org, hannes@...xchg.org,
        mhocko@...e.com, josef@...icpanda.com, jack@...e.cz,
        ldufour@...ux.ibm.com, laurent.dufour@...ibm.com,
        michel@...pinasse.org, liam.howlett@...cle.com, jglisse@...gle.com,
        vbabka@...e.cz, minchan@...gle.com, dave@...olabs.net,
        punit.agrawal@...edance.com, lstoakes@...il.com, hdanton@...a.com,
        apopple@...dia.com, ying.huang@...el.com, david@...hat.com,
        yuzhao@...gle.com, dhowells@...hat.com, hughd@...gle.com,
        viro@...iv.linux.org.uk, brauner@...nel.org,
        pasha.tatashin@...een.com, linux-mm@...ck.org,
        linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
        kernel-team@...roid.com
Subject: Re: [PATCH v3 5/8] mm: make folio_lock_fault indicate the state of
 mmap_lock upon return

On Mon, Jun 26, 2023 at 09:23:18PM -0700, Suren Baghdasaryan wrote:
> folio_lock_fault might drop mmap_lock before returning and to extend it
> to work with per-VMA locks, the callers will need to know whether the
> lock was dropped or is still held. Introduce new fault_flag to indicate
> whether the lock got dropped and store it inside vm_fault flags.
> 
> Signed-off-by: Suren Baghdasaryan <surenb@...gle.com>
> ---
>  include/linux/mm_types.h | 1 +
>  mm/filemap.c             | 2 ++
>  2 files changed, 3 insertions(+)
> 
> diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
> index 79765e3dd8f3..6f0dbef7aa1f 100644
> --- a/include/linux/mm_types.h
> +++ b/include/linux/mm_types.h
> @@ -1169,6 +1169,7 @@ enum fault_flag {
>  	FAULT_FLAG_UNSHARE =		1 << 10,
>  	FAULT_FLAG_ORIG_PTE_VALID =	1 << 11,
>  	FAULT_FLAG_VMA_LOCK =		1 << 12,
> +	FAULT_FLAG_LOCK_DROPPED =	1 << 13,
>  };
>  
>  typedef unsigned int __bitwise zap_flags_t;
> diff --git a/mm/filemap.c b/mm/filemap.c
> index 87b335a93530..8ad06d69895b 100644
> --- a/mm/filemap.c
> +++ b/mm/filemap.c
> @@ -1723,6 +1723,7 @@ vm_fault_t __folio_lock_fault(struct folio *folio, struct vm_fault *vmf)
>  			return VM_FAULT_RETRY;
>  
>  		mmap_read_unlock(mm);
> +		vmf->flags |= FAULT_FLAG_LOCK_DROPPED;
>  		if (vmf->flags & FAULT_FLAG_KILLABLE)
>  			folio_wait_locked_killable(folio);
>  		else
> @@ -1735,6 +1736,7 @@ vm_fault_t __folio_lock_fault(struct folio *folio, struct vm_fault *vmf)
>  		ret = __folio_lock_killable(folio);
>  		if (ret) {
>  			mmap_read_unlock(mm);
> +			vmf->flags |= FAULT_FLAG_LOCK_DROPPED;
>  			return VM_FAULT_RETRY;
>  		}
>  	} else {

IIRC we've discussed about this bits in previous version, and the consensus
was that we don't need yet another flag?  Just to recap: I think relying on
RETRY|COMPLETE would be enough for vma lock, as NOWAIT is only used by gup
while not affecting vma lockings, no?

As mentioned in the other reply, even COMPLETE won't appear for vma lock
path yet afaict, so mostly only RETRY matters here and it can 100% imply a
lock release happened.  It's just that it's very easy to still cover
COMPLETE altogether in this case, being prepared for any possible shared
support on vma locks, IMHO.

Thanks,

-- 
Peter Xu