qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Thomas Huth <thuth@redhat.com>
To: Halil Pasic <pasic@linux.ibm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>,
	qemu-s390x@nongnu.org, Cornelia Huck <cohuck@redhat.com>,
	qemu-devel@nongnu.org, Peter Maydell <peter.maydell@linaro.org>
Subject: Re: [PATCH] pc-bios/s390-ccw: Use memory barriers in virtio code
Date: Tue, 16 Feb 2021 17:15:19 +0100	[thread overview]
Message-ID: <72e9a5b3-dd88-85de-e4a8-88a6a9c45099@redhat.com> (raw)
In-Reply-To: <20210216154010.3691880f.pasic@linux.ibm.com>

On 16/02/2021 15.40, Halil Pasic wrote:
> On Tue, 16 Feb 2021 12:00:56 +0100
> Thomas Huth <thuth@redhat.com> wrote:
> 
>> According to the virtio specification, a memory barrier should be
>> used before incrementing the idx field in the "available" ring.
>> So far, we did not do this in the s390-ccw bios yet, but recently
>> Peter Maydell saw problems with the s390-ccw bios when running
>> the qtests on an aarch64 host (the bios panic'ed with the message:
>> "SCSI cannot report LUNs: response VS RESP=09"), which could
>> maybe be related to the missing memory barriers. Thus let's add
>> those barriers now. Since we've only seen the problem on TCG so far,
>> a "bcr 14,0" should be sufficient here to trigger the tcg_gen_mb()
>> in the TCG translate code.
>>
>> (Note: The virtio spec also talks about using a memory barrier
>> *after* incrementing the idx field, but if I understood correctly
>> this is only required when using notification suppression - which
>> we don't use in the s390-ccw bios here)
> 
> I suggest to the barrier after incrementing the idx field for two
> reasons. First: If the device were to see the notification, but
> not see the incremented idx field, it would effectively loose
> initiative. That is pretty straight forward, because the
> notification just says 'check out that queue', and if we don't
> see the incremented index, miss the buffer that was made available
> by incrementing idx.

I was just about to reply that this is certainly not necessary, since
the DIAGNOSE instruction that we use for the notification hypercall
should be serializing anyway ... but after looking at the PoP, it
actually is not marked as a serializing instruction! (while e.g.
SVC - supervisor call - is explicitly marked as serializing)

So maybe that's worth a try: Peter, could you please apply this patch
on top an see whether it makes a difference?

diff --git a/pc-bios/s390-ccw/virtio.c b/pc-bios/s390-ccw/virtio.c
--- a/pc-bios/s390-ccw/virtio.c
+++ b/pc-bios/s390-ccw/virtio.c
@@ -54,6 +54,7 @@ static long kvm_hypercall(unsigned long nr, unsigned long param1,
      register ulong r_param3 asm("4") = param3;
      register long retval asm("2");
  
+    virtio_mb();
      asm volatile ("diag 2,4,0x500"
                    : "=d" (retval)
                    : "d" (r_nr), "0" (r_param1), "r"(r_param2), "d"(r_param3)

  Thanks,
   Thomas



  parent reply	other threads:[~2021-02-16 16:17 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-02-16 11:00 [PATCH] pc-bios/s390-ccw: Use memory barriers in virtio code Thomas Huth
2021-02-16 11:43 ` Peter Maydell
2021-02-16 11:47   ` Thomas Huth
2021-02-16 11:47 ` Cornelia Huck
2021-02-16 14:21   ` Thomas Huth
2021-02-16 14:30     ` Cornelia Huck
2021-02-16 14:32       ` Peter Maydell
2021-02-16 14:35         ` Thomas Huth
2021-02-16 14:37           ` Peter Maydell
2021-02-16 14:49             ` Cornelia Huck
2021-02-16 14:40 ` Halil Pasic
2021-02-16 15:17   ` Cornelia Huck
2021-02-16 16:15   ` Thomas Huth [this message]
2021-02-16 16:44     ` Peter Maydell
2021-02-16 17:11     ` Halil Pasic
2021-02-17  4:31     ` Richard Henderson
2021-02-17 11:15       ` Peter Maydell
2021-02-17 15:11         ` Richard Henderson

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=72e9a5b3-dd88-85de-e4a8-88a6a9c45099@redhat.com \
    --to=thuth@redhat.com \
    --cc=borntraeger@de.ibm.com \
    --cc=cohuck@redhat.com \
    --cc=pasic@linux.ibm.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-s390x@nongnu.org \
    /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;
as well as URLs for NNTP newsgroup(s).