public inbox for linux-s390@vger.kernel.org
 help / color / mirror / Atom feed
From: Eric Farman <farman@linux.ibm.com>
To: Matthew Rosato <mjrosato@linux.ibm.com>,
	Halil Pasic <pasic@linux.ibm.com>
Cc: Heiko Carstens <hca@linux.ibm.com>,
	Vasily Gorbik <gor@linux.ibm.com>,
	Alexander Gordeev <agordeev@linux.ibm.com>,
	Christian Borntraeger <borntraeger@linux.ibm.com>,
	Sven Schnelle <svens@linux.ibm.com>,
	Vineeth Vijayan <vneethv@linux.ibm.com>,
	Peter Oberparleiter <oberpar@linux.ibm.com>,
	linux-s390@vger.kernel.org, kvm@vger.kernel.org
Subject: Re: [PATCH v1 13/16] vfio/ccw: allocate/populate the guest idal
Date: Mon, 19 Dec 2022 16:00:09 -0500	[thread overview]
Message-ID: <44d2f997494d39a9a4f529e676faa082fbfbc577.camel@linux.ibm.com> (raw)
In-Reply-To: <b6542dba-5645-f1f2-12a0-203e1187dd31@linux.ibm.com>

On Mon, 2022-12-19 at 15:14 -0500, Matthew Rosato wrote:
> On 11/21/22 4:40 PM, Eric Farman wrote:
> > Today, we allocate memory for a list of IDAWs, and if the CCW
> > being processed contains an IDAL we read that data from the guest
> > into that space. We then copy each IDAW into the pa_iova array,
> > or fabricate that pa_iova array with a list of addresses based
> > on a direct-addressed CCW.
> > 
> > Combine the reading of the guest IDAL with the creation of a
> > pseudo-IDAL for direct-addressed CCWs, so that both CCW types
> > have a "guest" IDAL that can be populated straight into the
> > pa_iova array.
> > 
> > Signed-off-by: Eric Farman <farman@linux.ibm.com>
> > ---
> >  drivers/s390/cio/vfio_ccw_cp.c | 72 +++++++++++++++++++++++-------
> > ----
> >  1 file changed, 50 insertions(+), 22 deletions(-)
> > 
> > diff --git a/drivers/s390/cio/vfio_ccw_cp.c
> > b/drivers/s390/cio/vfio_ccw_cp.c
> > index 6839e7195182..90685cee85db 100644
> > --- a/drivers/s390/cio/vfio_ccw_cp.c
> > +++ b/drivers/s390/cio/vfio_ccw_cp.c
> > @@ -192,11 +192,12 @@ static inline void
> > page_array_idal_create_words(struct page_array *pa,
> >          * idaw.
> >          */
> >  
> > -       for (i = 0; i < pa->pa_nr; i++)
> > +       for (i = 0; i < pa->pa_nr; i++) {
> >                 idaws[i] = page_to_phys(pa->pa_page[i]);
> >  
> > -       /* Adjust the first IDAW, since it may not start on a page
> > boundary */
> > -       idaws[0] += pa->pa_iova[0] & (PAGE_SIZE - 1);
> > +               /* Incorporate any offset from each starting
> > address */
> > +               idaws[i] += pa->pa_iova[i] & (PAGE_SIZE - 1);
> > +       }
> >  }
> >  
> >  static void convert_ccw0_to_ccw1(struct ccw1 *source, unsigned
> > long len)
> > @@ -496,6 +497,44 @@ static int ccwchain_fetch_tic(struct ccw1
> > *ccw,
> >         return -EFAULT;
> >  }
> >  
> > +static unsigned long *get_guest_idal(struct ccw1 *ccw,
> > +                                    struct channel_program *cp,
> > +                                    int idaw_nr)
> > +{
> > +       struct vfio_device *vdev =
> > +               &container_of(cp, struct vfio_ccw_private, cp)-
> > >vdev;
> > +       unsigned long *idaws;
> > +       int idal_len = idaw_nr * sizeof(*idaws);
> > +       int idaw_size = PAGE_SIZE;
> > +       int idaw_mask = ~(idaw_size - 1);
> > +       int i, ret;
> > +
> > +       idaws = kcalloc(idaw_nr, sizeof(*idaws), GFP_DMA |
> > GFP_KERNEL);
> > +       if (!idaws)
> > +               return NULL;
> > +
> > +       if (ccw_is_idal(ccw)) {
> > +               /* Copy IDAL from guest */
> > +               ret = vfio_dma_rw(vdev, ccw->cda, idaws, idal_len,
> > false);
> > +               if (ret) {
> > +                       kfree(idaws);
> > +                       return NULL;
> 
> As discussed off-list, for debug purposes consider using something
> like ERR_PTR of the vfio_dma_rw error return here rather than NULL. 

Yes, good idea.

> 
> > +               }
> > +       } else {
> > +               /* Fabricate an IDAL based off CCW data address */
> > +               if (cp->orb.cmd.c64) {
> > +                       idaws[0] = ccw->cda;
> > +                       for (i = 1; i < idaw_nr; i++)
> > +                               idaws[i] = (idaws[i - 1] +
> > idaw_size) & idaw_mask;
> > +               } else {
> 
> If anyone else is reviewing and stumbles on this, I was initially
> wondering why we bail here with no obvious explanation - was going to
> ask for a comment here but it looks like this else gets replaced next
> patch with implementation for format-1.

As you note, this goes away in the next patch so I didn't put a comment
in place. But while putting in the ERR_PTR/PTR_ERR stuff, I opted to
return EOPNOTSUPP here instead of NULL, so we get an error equal to the
current fence instead of an ENOMEM. So that'll be in v2.

Thanks,
Eric

> 
> > +                       kfree(idaws);
> > +                       return NULL;
> > +               }
> > +       }
> > +
> > +       return idaws;
> > +}
> > +
> >  /*
> >   * ccw_count_idaws() - Calculate the number of IDAWs needed to
> > transfer
> >   * a specified amount of data
> > @@ -555,7 +594,7 @@ static int ccwchain_fetch_ccw(struct ccw1 *ccw,
> >                 &container_of(cp, struct vfio_ccw_private, cp)-
> > >vdev;
> >         unsigned long *idaws;
> >         int ret;
> > -       int idaw_nr, idal_len;
> > +       int idaw_nr;
> >         int i;
> >  
> >         /* Calculate size of IDAL */
> > @@ -563,10 +602,8 @@ static int ccwchain_fetch_ccw(struct ccw1
> > *ccw,
> >         if (idaw_nr < 0)
> >                 return idaw_nr;
> >  
> > -       idal_len = idaw_nr * sizeof(*idaws);
> > -
> >         /* Allocate an IDAL from host storage */
> > -       idaws = kcalloc(idaw_nr, sizeof(*idaws), GFP_DMA |
> > GFP_KERNEL);
> > +       idaws = get_guest_idal(ccw, cp, idaw_nr);
> >         if (!idaws) {
> >                 ret = -ENOMEM;
> >                 goto out_init;
> > @@ -582,22 +619,13 @@ static int ccwchain_fetch_ccw(struct ccw1
> > *ccw,
> >         if (ret < 0)
> >                 goto out_free_idaws;
> >  
> > -       if (ccw_is_idal(ccw)) {
> > -               /* Copy guest IDAL into host IDAL */
> > -               ret = vfio_dma_rw(vdev, ccw->cda, idaws, idal_len,
> > false);
> > -               if (ret)
> > -                       goto out_unpin;
> > -
> > -               /*
> > -                * Copy guest IDAWs into page_array, in case the
> > memory they
> > -                * occupy is not contiguous.
> > -                */
> > -               for (i = 0; i < idaw_nr; i++)
> > +       /*
> > +        * Copy guest IDAWs into page_array, in case the memory
> > they
> > +        * occupy is not contiguous.
> > +        */
> > +       for (i = 0; i < idaw_nr; i++) {
> > +               if (cp->orb.cmd.c64)
> >                         pa->pa_iova[i] = idaws[i];
> > -       } else {
> > -               pa->pa_iova[0] = ccw->cda;
> > -               for (i = 1; i < pa->pa_nr; i++)
> > -                       pa->pa_iova[i] = pa->pa_iova[i - 1] +
> > PAGE_SIZE;
> >         }
> >  
> >         if (ccw_does_data_transfer(ccw)) {
> 


  reply	other threads:[~2022-12-19 21:00 UTC|newest]

Thread overview: 41+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-21 21:40 [PATCH v1 00/16] vfio/ccw: channel program cleanup Eric Farman
2022-11-21 21:40 ` [PATCH v1 01/16] vfio/ccw: cleanup some of the mdev commentary Eric Farman
2022-11-22 16:12   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 02/16] vfio/ccw: simplify the cp_get_orb interface Eric Farman
2022-11-22 16:13   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 03/16] vfio/ccw: allow non-zero storage keys Eric Farman
2022-12-15 20:55   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 04/16] vfio/ccw: move where IDA flag is set in ORB Eric Farman
2022-12-15 20:55   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 05/16] vfio/ccw: replace copy_from_iova with vfio_dma_rw Eric Farman
2022-11-22  1:41   ` Jason Gunthorpe
2022-12-15 20:59   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 06/16] vfio/ccw: simplify CCW chain fetch routines Eric Farman
2022-12-15 21:18   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 07/16] vfio/ccw: remove unnecessary malloc alignment Eric Farman
2022-12-16 20:10   ` Matthew Rosato
2022-12-19 16:22     ` Eric Farman
2022-11-21 21:40 ` [PATCH v1 08/16] vfio/ccw: pass page count to page_array struct Eric Farman
2022-12-16 19:59   ` Matthew Rosato
2022-12-19 16:22     ` Eric Farman
2022-11-21 21:40 ` [PATCH v1 09/16] vfio/ccw: populate page_array struct inline Eric Farman
2022-12-16 21:05   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 10/16] vfio/ccw: refactor the idaw counter Eric Farman
2022-12-19 19:16   ` Matthew Rosato
2022-12-19 19:31     ` Eric Farman
2022-12-19 19:40       ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 11/16] vfio/ccw: discard second fmt-1 IDAW Eric Farman
2022-12-19 19:27   ` Matthew Rosato
2022-12-19 20:27     ` Eric Farman
2022-11-21 21:40 ` [PATCH v1 12/16] vfio/ccw: calculate number of IDAWs regardless of format Eric Farman
2022-12-19 19:49   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 13/16] vfio/ccw: allocate/populate the guest idal Eric Farman
2022-12-19 20:14   ` Matthew Rosato
2022-12-19 21:00     ` Eric Farman [this message]
2022-11-21 21:40 ` [PATCH v1 14/16] vfio/ccw: handle a guest Format-1 IDAL Eric Farman
2022-12-19 20:29   ` Matthew Rosato
2022-12-19 21:04     ` Eric Farman
2022-11-21 21:40 ` [PATCH v1 15/16] vfio/ccw: don't group contiguous pages on 2K IDAWs Eric Farman
2022-12-19 20:40   ` Matthew Rosato
2022-11-21 21:40 ` [PATCH v1 16/16] vfio/ccw: remove old IDA format restrictions Eric Farman
2022-12-19 20:44   ` Matthew Rosato

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=44d2f997494d39a9a4f529e676faa082fbfbc577.camel@linux.ibm.com \
    --to=farman@linux.ibm.com \
    --cc=agordeev@linux.ibm.com \
    --cc=borntraeger@linux.ibm.com \
    --cc=gor@linux.ibm.com \
    --cc=hca@linux.ibm.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-s390@vger.kernel.org \
    --cc=mjrosato@linux.ibm.com \
    --cc=oberpar@linux.ibm.com \
    --cc=pasic@linux.ibm.com \
    --cc=svens@linux.ibm.com \
    --cc=vneethv@linux.ibm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox