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: <ddc6a307-9520-4f5e-bc41-ef380b0d826a@bytedance.com>
Date: Tue, 24 Sep 2024 16:45:03 +0800
From: Qi Zheng <zhengqi.arch@...edance.com>
To: Muchun Song <muchun.song@...ux.dev>
Cc: david@...hat.com, hughd@...gle.com, willy@...radead.org,
 vbabka@...nel.org, akpm@...ux-foundation.org, rppt@...nel.org,
 vishal.moola@...il.com, peterx@...hat.com, ryan.roberts@....com,
 christophe.leroy2@...soprasteria.com, linux-kernel@...r.kernel.org,
 linux-mm@...ck.org, linux-arm-kernel@...ts.infradead.org,
 linuxppc-dev@...ts.ozlabs.org
Subject: Re: [PATCH v4 10/13] mm: page_vma_mapped_walk: map_pte() use
 pte_offset_map_rw_nolock()



On 2024/9/24 16:39, Muchun Song wrote:
> 
> 
>> On Sep 24, 2024, at 16:33, Qi Zheng <zhengqi.arch@...edance.com> wrote:
>>
>>
>>
>> On 2024/9/24 16:25, Muchun Song wrote:
>>>> On Sep 24, 2024, at 14:11, Qi Zheng <zhengqi.arch@...edance.com> wrote:
>>>>
>>>> In the caller of map_pte(), we may modify the pvmw->pte after acquiring
>>>> the pvmw->ptl, so convert it to using pte_offset_map_rw_nolock(). At
>>>> this time, the pte_same() check is not performed after the pvmw->ptl held,
>>>> so we should get pmdval and do pmd_same() check to ensure the stability of
>>>> pvmw->pmd.
>>>>
>>>> Signed-off-by: Qi Zheng <zhengqi.arch@...edance.com>
>>>> ---
>>>> mm/page_vma_mapped.c | 25 ++++++++++++++++++++-----
>>>> 1 file changed, 20 insertions(+), 5 deletions(-)
>>>>
>>>> diff --git a/mm/page_vma_mapped.c b/mm/page_vma_mapped.c
>>>> index ae5cc42aa2087..6410f29b37c1b 100644
>>>> --- a/mm/page_vma_mapped.c
>>>> +++ b/mm/page_vma_mapped.c
>>>> @@ -13,9 +13,11 @@ static inline bool not_found(struct page_vma_mapped_walk *pvmw)
>>>>     return false;
>>>> }
>>>>
>>>> -static bool map_pte(struct page_vma_mapped_walk *pvmw, spinlock_t **ptlp)
>>>> +static bool map_pte(struct page_vma_mapped_walk *pvmw, pmd_t *pmdvalp,
>>>> +            spinlock_t **ptlp)
>>>> {
>>>>     pte_t ptent;
>>>> +    pmd_t pmdval;
>>> Why declare a new variable? Can’t we use *pmdvalp instead?
>>
>> It's just a coding habit, both are fine for me.
> 
> Agree. But sometime it could make code look a little simpler.
> 
>>
>>>>
>>>>     if (pvmw->flags & PVMW_SYNC) {
>>>>         /* Use the stricter lookup */
>>>> @@ -25,6 +27,7 @@ static bool map_pte(struct page_vma_mapped_walk *pvmw, spinlock_t **ptlp)
>>>>         return !!pvmw->pte;
>>>>     }
>>>>
>>>> +again:
>>>>     /*
>>>>      * It is important to return the ptl corresponding to pte,
>>>>      * in case *pvmw->pmd changes underneath us; so we need to
>>>> @@ -32,10 +35,11 @@ static bool map_pte(struct page_vma_mapped_walk *pvmw, spinlock_t **ptlp)
>>>>      * proceeds to loop over next ptes, and finds a match later.
>>>>      * Though, in most cases, page lock already protects this.
>>>>      */
>>>> -    pvmw->pte = pte_offset_map_nolock(pvmw->vma->vm_mm, pvmw->pmd,
>>>> -                      pvmw->address, ptlp);
>>>> +    pvmw->pte = pte_offset_map_rw_nolock(pvmw->vma->vm_mm, pvmw->pmd,
>>>> +                         pvmw->address, &pmdval, ptlp);
>>>>     if (!pvmw->pte)
>>>>         return false;
>>>> +    *pmdvalp = pmdval;
> 
> For instance, here, it is unnecessary if pmdvalp is passed directly to
> pte_offset_map_rw_nolock.

OK, will use pmdvalp directly. ;)

> 
>>>>
>>>>     ptent = ptep_get(pvmw->pte);
>>>>
>>>> @@ -67,8 +71,13 @@ static bool map_pte(struct page_vma_mapped_walk *pvmw, spinlock_t **ptlp)
>>>>     } else if (!pte_present(ptent)) {
>>>>         return false;
>>>>     }
>>>> +    spin_lock(*ptlp);
>>>> +    if (unlikely(!pmd_same(pmdval, pmdp_get_lockless(pvmw->pmd)))) {
>>>> +        pte_unmap_unlock(pvmw->pte, *ptlp);
>>>> +        goto again;
>>>> +    }
>>>>     pvmw->ptl = *ptlp;
>>>> -    spin_lock(pvmw->ptl);
>>>> +
>>>>     return true;
>>>> }
>>>>
>>>> @@ -278,7 +287,7 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
>>>>             step_forward(pvmw, PMD_SIZE);
>>>>             continue;
>>>>         }
>>>> -        if (!map_pte(pvmw, &ptl)) {
>>>> +        if (!map_pte(pvmw, &pmde, &ptl)) {
>>>>             if (!pvmw->pte)
>>>>                 goto restart;
>>>>             goto next_pte;
>>>> @@ -307,6 +316,12 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
>>>>         if (!pvmw->ptl) {
>>>>             pvmw->ptl = ptl;
>>>>             spin_lock(pvmw->ptl);
>>>> +            if (unlikely(!pmd_same(pmde, pmdp_get_lockless(pvmw->pmd)))) {
>>>> +                pte_unmap_unlock(pvmw->pte, pvmw->ptl);
>>>> +                pvmw->ptl = NULL;
>>>> +                pvmw->pte = NULL;
>>>> +                goto restart;
>>>> +            }
>>>>         }
>>>>         goto this_pte;
>>>>     } while (pvmw->address < end);
>>>> --
>>>> 2.20.1
> 
> 

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ