Message-ID: <CY4PR21MB158664748760672446BFA075D7C19@CY4PR21MB1586.namprd21.prod.outlook.com>
Date:   Fri, 20 Aug 2021 16:08:32 +0000
From:   Michael Kelley <mikelley@...rosoft.com>
To:     Tianyu Lan <ltykernel@...il.com>,
        KY Srinivasan <kys@...rosoft.com>,
        Haiyang Zhang <haiyangz@...rosoft.com>,
        Stephen Hemminger <sthemmin@...rosoft.com>,
        "wei.liu@...nel.org" <wei.liu@...nel.org>,
        Dexuan Cui <decui@...rosoft.com>,
        "tglx@...utronix.de" <tglx@...utronix.de>,
        "mingo@...hat.com" <mingo@...hat.com>,
        "bp@...en8.de" <bp@...en8.de>, "x86@...nel.org" <x86@...nel.org>,
        "hpa@...or.com" <hpa@...or.com>,
        "dave.hansen@...ux.intel.com" <dave.hansen@...ux.intel.com>,
        "luto@...nel.org" <luto@...nel.org>,
        "peterz@...radead.org" <peterz@...radead.org>,
        "konrad.wilk@...cle.com" <konrad.wilk@...cle.com>,
        "boris.ostrovsky@...cle.com" <boris.ostrovsky@...cle.com>,
        "jgross@...e.com" <jgross@...e.com>,
        "sstabellini@...nel.org" <sstabellini@...nel.org>,
        "joro@...tes.org" <joro@...tes.org>,
        "will@...nel.org" <will@...nel.org>,
        "davem@...emloft.net" <davem@...emloft.net>,
        "kuba@...nel.org" <kuba@...nel.org>,
        "jejb@...ux.ibm.com" <jejb@...ux.ibm.com>,
        "martin.petersen@...cle.com" <martin.petersen@...cle.com>,
        "arnd@...db.de" <arnd@...db.de>, "hch@....de" <hch@....de>,
        "m.szyprowski@...sung.com" <m.szyprowski@...sung.com>,
        "robin.murphy@....com" <robin.murphy@....com>,
        "thomas.lendacky@....com" <thomas.lendacky@....com>,
        "brijesh.singh@....com" <brijesh.singh@....com>,
        "ardb@...nel.org" <ardb@...nel.org>,
        Tianyu Lan <Tianyu.Lan@...rosoft.com>,
        "pgonda@...gle.com" <pgonda@...gle.com>,
        "martin.b.radev@...il.com" <martin.b.radev@...il.com>,
        "akpm@...ux-foundation.org" <akpm@...ux-foundation.org>,
        "kirill.shutemov@...ux.intel.com" <kirill.shutemov@...ux.intel.com>,
        "rppt@...nel.org" <rppt@...nel.org>,
        "sfr@...b.auug.org.au" <sfr@...b.auug.org.au>,
        "saravanand@...com" <saravanand@...com>,
        "krish.sadhukhan@...cle.com" <krish.sadhukhan@...cle.com>,
        "aneesh.kumar@...ux.ibm.com" <aneesh.kumar@...ux.ibm.com>,
        "xen-devel@...ts.xenproject.org" <xen-devel@...ts.xenproject.org>,
        "rientjes@...gle.com" <rientjes@...gle.com>,
        "hannes@...xchg.org" <hannes@...xchg.org>,
        "tj@...nel.org" <tj@...nel.org>
CC:     "iommu@...ts.linux-foundation.org" <iommu@...ts.linux-foundation.org>,
        "linux-arch@...r.kernel.org" <linux-arch@...r.kernel.org>,
        "linux-hyperv@...r.kernel.org" <linux-hyperv@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-scsi@...r.kernel.org" <linux-scsi@...r.kernel.org>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        vkuznets <vkuznets@...hat.com>,
        "parri.andrea@...il.com" <parri.andrea@...il.com>,
        "dave.hansen@...el.com" <dave.hansen@...el.com>
Subject: RE: [PATCH V3 13/13] HV/Storvsc: Add Isolation VM support for storvsc
 driver

From: Tianyu Lan <ltykernel@...il.com> Sent: Friday, August 20, 2021 8:20 AM
> 
> On 8/20/2021 2:17 AM, Michael Kelley wrote:
> > From: Tianyu Lan <ltykernel@...il.com> Sent: Monday, August 9, 2021 10:56 AM
> >
> > I'm not clear on why payload->range.offset needs to be set again.
> > Even after the dma mapping is done, doesn't the offset in the first
> > page have to be the same?  If it wasn't the same, Hyper-V wouldn't
> > be able to process the PFN list correctly.  In fact, couldn't the above
> > code just always set offset_in_hvpg = 0?
> 
> The offset will be changed. The swiotlb bounce buffer is allocated with
> IO_TLB_SIZE(2K) as unit. So the offset here may be changed.
> 

We need to prevent the offset from changing.  The storvsc driver passes
just a PFN list to Hyper-V, plus an overall starting offset and length.  Unlike
the netvsc driver, each entry in the PFN list does *not* have its own offset
and length.  Hyper-V assumes that the list is "dense" and that there are
no holes (i.e., unused memory areas).

For example, consider an original buffer passed into storvsc_queuecommand()
of 8 Kbytes, but aligned with 1 Kbytes at the end of the first page, then
4 Kbytes in the second page, and 3 Kbytes in the beginning of the third page.
The offset of that first 1 Kbytes has to remain as 3 Kbytes.  If bounce buffering
moves it to a different offset, there's no way to tell Hyper-V to ignore the
remaining bytes in the first page (at least not without using a different
method to communicate with Hyper-V).   In such a case, the wrong
data will get transferred.  Presumably the easier solution is to set the
min_align_mask field as Christop suggested.

> 
> >
> >>   	}
> >
> > The whole approach here is to do dma remapping on each individual page
> > of the I/O buffer.  But wouldn't it be possible to use dma_map_sg() to map
> > each scatterlist entry as a unit?  Each scatterlist entry describes a range of
> > physically contiguous memory.  After dma_map_sg(), the resulting dma
> > address must also refer to a physically contiguous range in the swiotlb
> > bounce buffer memory.   So at the top of the "for" loop over the scatterlist
> > entries, do dma_map_sg() if we're in an isolated VM.  Then compute the
> > hvpfn value based on the dma address instead of sg_page().  But everything
> > else is the same, and the inner loop for populating the pfn_arry is unmodified.
> > Furthermore, the dma_range array that you've added is not needed, since
> > scatterlist entries already have a dma_address field for saving the mapped
> > address, and dma_unmap_sg() uses that field.
> 
> I don't use dma_map_sg() here in order to avoid introducing one more
> loop(e,g dma_map_sg()). We already have a loop to populate
> cmd_request->dma_range[] and so do the dma map in the same loop.
> 

I'm not seeing where the additional loop comes from.  Storvsc
already has a loop through the sgl entries.  Retain that loop and call
dma_map_sg() with nents set to 1.  Then the sequence is
dma_map_sg() --> dma_map_sg_attrs() --> dma_direct_map_sg() ->
dma_direct_map_page().  The latter function will call swiotlb_map()
to map all pages of the sgl entry as a single operation.

Michael

Powered by blists - more mailing lists