[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <c299567a-5be8-f65f-d8ec-ffd3fa183b03@intel.com>
Date: Mon, 9 Oct 2023 12:13:52 +0800
From: Yi Liu <yi.l.liu@...el.com>
To: Yan Zhao <yan.y.zhao@...el.com>
CC: <joro@...tes.org>, <alex.williamson@...hat.com>, <jgg@...dia.com>,
<kevin.tian@...el.com>, <robin.murphy@....com>,
<baolu.lu@...ux.intel.com>, <cohuck@...hat.com>,
<eric.auger@...hat.com>, <nicolinc@...dia.com>,
<kvm@...r.kernel.org>, <mjrosato@...ux.ibm.com>,
<chao.p.peng@...ux.intel.com>, <yi.y.sun@...ux.intel.com>,
<peterx@...hat.com>, <jasowang@...hat.com>,
<shameerali.kolothum.thodi@...wei.com>, <lulu@...hat.com>,
<suravee.suthikulpanit@....com>, <iommu@...ts.linux.dev>,
<linux-kernel@...r.kernel.org>, <linux-kselftest@...r.kernel.org>,
<zhenzhong.duan@...el.com>, <joao.m.martins@...cle.com>
Subject: Re: [PATCH v4 03/17] iommufd: Unite all kernel-managed members into a
struct
On 2023/10/7 18:08, Yan Zhao wrote:
> On Thu, Sep 21, 2023 at 12:51:24AM -0700, Yi Liu wrote:
>> From: Nicolin Chen <nicolinc@...dia.com>
>>
>> The struct iommufd_hw_pagetable has been representing a kernel-managed
>> HWPT, yet soon will be reused to represent a user-managed HWPT. These
>> two types of HWPTs has the same IOMMUFD object type and an iommu_domain
>> object, but have quite different attributes/members.
>>
>> Add a union in struct iommufd_hw_pagetable and group all the existing
>> kernel-managed members. One of the following patches will add another
>> struct for user-managed members.
>>
>> Signed-off-by: Nicolin Chen <nicolinc@...dia.com>
>> Signed-off-by: Yi Liu <yi.l.liu@...el.com>
>> ---
>> drivers/iommu/iommufd/iommufd_private.h | 17 +++++++++++------
>> 1 file changed, 11 insertions(+), 6 deletions(-)
>>
>> diff --git a/drivers/iommu/iommufd/iommufd_private.h b/drivers/iommu/iommufd/iommufd_private.h
>> index 3064997a0181..947a797536e3 100644
>> --- a/drivers/iommu/iommufd/iommufd_private.h
>> +++ b/drivers/iommu/iommufd/iommufd_private.h
>> @@ -231,13 +231,18 @@ int iommufd_vfio_ioas(struct iommufd_ucmd *ucmd);
>> */
>> struct iommufd_hw_pagetable {
>> struct iommufd_object obj;
>> - struct iommufd_ioas *ioas;
>> struct iommu_domain *domain;
>> - bool auto_domain : 1;
>> - bool enforce_cache_coherency : 1;
>> - bool msi_cookie : 1;
>> - /* Head at iommufd_ioas::hwpt_list */
>> - struct list_head hwpt_item;
>> +
>> + union {
>> + struct { /* kernel-managed */
>> + struct iommufd_ioas *ioas;
>> + bool auto_domain : 1;
> Will iommufd_hw_pagetable_put() also be called on non-kernel-managed domain?
yes.
> If yes, hwpt->user_managed needs to be checked in iommufd_hw_pagetable_put(),
> (e.g. as below).
> Otherwise, this union will lead to hwpt->ioas and hwpt->auto_domain still being
> accessible though invalid.
not quite get this sentence.
>
> static inline void iommufd_hw_pagetable_put(struct iommufd_ctx *ictx,
> struct iommufd_hw_pagetable *hwpt)
> {
> - lockdep_assert_not_held(&hwpt->ioas->mutex);
> - if (hwpt->auto_domain)
> + if (!hwpt->user_managed)
> + lockdep_assert_not_held(&hwpt->ioas->mutex);
this is true. this assert is not needed when hwpt is not kernel managed domain.
> +
> + if (!hwpt->user_managed && hwpt->auto_domain)
actually, checking auto_domain is more precise. There is hwpt which is
neither user managed nor auto.
> iommufd_object_deref_user(ictx, &hwpt->obj);
> else
> refcount_dec(&hwpt->obj.users);
> }
>
>> + bool enforce_cache_coherency : 1;
>> + bool msi_cookie : 1;
>> + /* Head at iommufd_ioas::hwpt_list */
>> + struct list_head hwpt_item;
>> + };
>> + };
>> };
>>
>> struct iommufd_hw_pagetable *
>> --
>> 2.34.1
>>
--
Regards,
Yi Liu
Powered by blists - more mailing lists