Message-ID: <d8656a67a875b72dfdf5826ec4ff7154022f1ef6.camel@linux.ibm.com>
Subject: Re: [PATCH v6 4/6] s390/pci: Use dma-iommu layer
From:   Niklas Schnelle <schnelle@linux.ibm.com>
To:     Matthew Rosato <mjrosato@linux.ibm.com>,
        Joerg Roedel <joro@8bytes.org>, Will Deacon <will@kernel.org>,
        Robin Murphy <robin.murphy@arm.com>,
        Jason Gunthorpe <jgg@nvidia.com>,
        Wenjia Zhang <wenjia@linux.ibm.com>
Cc:     Gerd Bayer <gbayer@linux.ibm.com>,
        Pierre Morel <pmorel@linux.ibm.com>, iommu@lists.linux.dev,
        linux-s390@vger.kernel.org, borntraeger@linux.ibm.com,
        hca@linux.ibm.com, gor@linux.ibm.com,
        gerald.schaefer@linux.ibm.com, agordeev@linux.ibm.com,
        svens@linux.ibm.com, linux-kernel@vger.kernel.org,
        Julian Ruess <julianr@linux.ibm.com>
Date:   Fri, 17 Feb 2023 17:34:43 +0100
In-Reply-To: <7dc31552-1b7a-aeaa-eb88-a60d96e28f3c@linux.ibm.com>
References: <20230215120327.947336-1-schnelle@linux.ibm.com>
         <20230215120327.947336-5-schnelle@linux.ibm.com>
         <4ad0d2b9-2465-b42d-c0f9-b48caadfd72c@linux.ibm.com>
         <59a61553a81282fed88c03af096b2a0830b94302.camel@linux.ibm.com>
         <7dc31552-1b7a-aeaa-eb88-a60d96e28f3c@linux.ibm.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.46.3 (3.46.3-1.fc37) 
MIME-Version: 1.0
Precedence: bulk

On Fri, 2023-02-17 at 09:56 -0500, Matthew Rosato wrote:
> On 2/17/23 3:51 AM, Niklas Schnelle wrote:
> > On Wed, 2023-02-15 at 13:00 -0500, Matthew Rosato wrote:
> > > On 2/15/23 7:03 AM, Niklas Schnelle wrote:
> > > > While s390 already has a standard IOMMU driver and previous changes=
 have
> > > > added I/O TLB flushing operations this driver is currently only use=
d for
> > > > user-space PCI access such as vfio-pci. For the DMA API s390 instea=
d
> > > > utilizes its own implementation in arch/s390/pci/pci_dma.c which dr=
ives
> > > > the same hardware and shares some code but requires a complex and
> > > > fragile hand over between DMA API and IOMMU API use of a device and
> > > > despite code sharing still leads to significant duplication and
> > > > maintenance effort. Let's utilize the common code DMAP API
> > > > implementation from drivers/iommu/dma-iommu.c instead allowing us t=
o
> > > > get rid of arch/s390/pci/pci_dma.c.
> > > >=20
> > > > Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
> > > > ---
> > >=20
> > > FYI, this patch doesn't fit on top of iommu-next, I'd guess at least =
due to baolu's 'Retire detach_dev callback' series, which removed .detach_d=
ev and added .set_platform_dma_ops for s390-iommu.  That's relevant here, b=
ecause now that this patch enables dma-iommu for s390 and removes the platf=
orm DMA ops it must now remove .set_platform_dma_ops/s390_iommu_set_platfor=
m_dma for s390-iommu.
> > >=20
> > > Matt
> >=20
> >=20
> > Ok, yes this series is currently against v6.2-rc8. Should I rebase
> > against iommu-next and send a v7 before further review or after?
> >=20
>=20
> So, overall I'm fine with the code in this patch (and this series) at thi=
s point; however I'd like to do one more pass of testing rebased on top of =
iommu-next / with the set_platform_dma collision handled, so I'm going to h=
old off on tagging my review of this one until v7.  That really only impact=
s this patch so if you want to give others a chance to review the rest of t=
he series before rolling out v7 that's OK by me. =20
>=20
> Thanks,
> Matt


Ok, I've got a version on top of iommu-next running locally and will
send that as v7 next week. Thankfully the conflict resolution is pretty
simple. We get rid of set_platform_dma() completely and the detach is
only used internally when attaching to a different domain. Also got rid
of a missed superfluous zdev->dma_table assignment that I missed as
feedback.

Thanks,
Niklas