qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Ilya Leoshkevich <iii@linux.ibm.com>
To: David Hildenbrand <david@redhat.com>,
	Richard Henderson <richard.henderson@linaro.org>,
	Cornelia Huck <cohuck@redhat.com>, Thomas Huth <thuth@redhat.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>,
	qemu-s390x@nongnu.org, qemu-devel@nongnu.org
Subject: Re: [PATCH v2 3/5] target/s390x: Fix cc_calc_sla_64() missing overflows
Date: Wed, 12 Jan 2022 13:51:38 +0100	[thread overview]
Message-ID: <add9d2da455bf9b07cf77c942e95183f5bf6f090.camel@linux.ibm.com> (raw)
In-Reply-To: <08948b31-0729-5463-829f-35fb444cf82d@redhat.com>

On Wed, 2022-01-12 at 09:59 +0100, David Hildenbrand wrote:
> On 12.01.22 05:39, Ilya Leoshkevich wrote:
> > An overflow occurs for SLAG when at least one shifted bit is not
> > equal
> > to sign bit. Therefore, we need to check that `shift + 1` bits are
> > neither all 0s nor all 1s. The current code checks only `shift`
> > bits,
> > missing some overflows.
> 
> Right, "shifted + 1" here means, the shifted bits + the sign bit.
> 
> But doesn't the
> 
> if (src & sign) {
>     match = mask;
> } else {
>     match = 0;
> }
> 
> logic handle that?
> 
> If the sign is false, the shifted bits (mask) have to be 0.
> If the sign bit is true, the shifted bits (mask) have to be set.

IIUC this logic handles sign bit + "shift - 1" bits. So if the last
shifted bit is different, the overflow is not detected.

> 
> Do you have an example that would be broken?

sla-2 test covers this. I added a similar one for SLAG in v3 and
it fails as well.

> > Fixes: cbe24bfa91d2 ("target-s390: Convert SHIFT, ROTATE SINGLE")
> > Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
> > ---
> >  target/s390x/tcg/cc_helper.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> > 
> > diff --git a/target/s390x/tcg/cc_helper.c
> > b/target/s390x/tcg/cc_helper.c
> > index c2c96c3a3c..b6acffa3e8 100644
> > --- a/target/s390x/tcg/cc_helper.c
> > +++ b/target/s390x/tcg/cc_helper.c
> > @@ -297,7 +297,8 @@ static uint32_t cc_calc_sla_32(uint32_t src,
> > int shift)
> >  
> >  static uint32_t cc_calc_sla_64(uint64_t src, int shift)
> >  {
> > -    uint64_t mask = ((1ULL << shift) - 1ULL) << (64 - shift);
> > +    /* Do not use (1ULL << (shift + 1)): it triggers UB when shift
> > is 63.  */
> > +    uint64_t mask = ((((1ULL << shift) - 1) << 1) + 1) << (64 -
> > (shift + 1));
> >      uint64_t sign = 1ULL << 63;
> >      uint64_t match;
> >      int64_t r;
> 
> This looks like some black magic :)

Yeah, I felt this way too, but didn't come up with anything better and
just left a comment warning not to simplify.


  reply	other threads:[~2022-01-12 14:01 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-01-12  4:39 [PATCH v2 0/5] target/s390x: Fix shift instructions Ilya Leoshkevich
2022-01-12  4:39 ` [PATCH v2 1/5] target/s390x: Fix SLDA sign bit index Ilya Leoshkevich
2022-01-12  8:33   ` David Hildenbrand
2022-01-12  4:39 ` [PATCH v2 2/5] target/s390x: Fix SRDA CC calculation Ilya Leoshkevich
2022-01-12  8:46   ` David Hildenbrand
2022-01-12  4:39 ` [PATCH v2 3/5] target/s390x: Fix cc_calc_sla_64() missing overflows Ilya Leoshkevich
2022-01-12  8:59   ` David Hildenbrand
2022-01-12 12:51     ` Ilya Leoshkevich [this message]
2022-01-12 15:45       ` David Hildenbrand
2022-01-12 16:38         ` Ilya Leoshkevich
2022-01-12  4:39 ` [PATCH v2 4/5] target/s390x: Fix shifting 32-bit values for more than 31 bits Ilya Leoshkevich
2022-01-12  8:43   ` David Hildenbrand
2022-01-12  4:39 ` [PATCH v2 5/5] tests/tcg/s390x: Test shift instructions Ilya Leoshkevich
2022-01-12  9:07   ` David Hildenbrand

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=add9d2da455bf9b07cf77c942e95183f5bf6f090.camel@linux.ibm.com \
    --to=iii@linux.ibm.com \
    --cc=borntraeger@de.ibm.com \
    --cc=cohuck@redhat.com \
    --cc=david@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-s390x@nongnu.org \
    --cc=richard.henderson@linaro.org \
    --cc=thuth@redhat.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;
as well as URLs for NNTP newsgroup(s).