debug subtract error

debug subtract error

R=ahaas@chromium.org,bradnelson@chromium.org
BUG=

[aseemgarg, 2016-04-11 21:12:45]

Hey Andreas

Seems like the subtract 0 error is not fully fixed. Please see this test case
that generates the error.

Thanks

[bradn, 2016-04-12 17:19:38]

bradnelson@google.com changed reviewers:
+ bradnelson@google.com

[bradn, 2016-04-12 17:19:38]

I take it the wasm level test doesn't repro this anymore?
What's the difference in wasm bytes?

[ahaas, 2016-04-12 18:13:34]

On 2016/04/12 at 17:19:38, bradnelson wrote:
> I take it the wasm level test doesn't repro this anymore?
> What's the difference in wasm bytes?

Hi,

The problem is that when you pass an int32 to wasm it is first converted to a
float64 and then back to an int64. Through these conversions the -1 gets sign
extended and thereby causes the problem. Do you need the bug fix soon or can it
wait until Thursday?  I attend a workshop today and tomorrow so I don't really
have time for it.

[ahaas, 2016-04-12 18:22:03]

On 2016/04/12 at 18:13:34, ahaas wrote:
> On 2016/04/12 at 17:19:38, bradnelson wrote:
> > I take it the wasm level test doesn't repro this anymore?
> > What's the difference in wasm bytes?
> 
> Hi,
> 
> The problem is that when you pass an int32 to wasm it is first converted to a
float64 and then back to an int64. Through these conversions the -1 gets sign
extended and thereby causes the problem. Do you need the bug fix soon or can it
wait until Thursday?  I attend a workshop today and tomorrow so I don't really
have time for it.

This CL (https://codereview.chromium.org/1877133004) would fix your problem.

[ahaas, 2016-04-14 07:29:32]

On 2016/04/12 at 18:22:03, ahaas wrote:
> On 2016/04/12 at 18:13:34, ahaas wrote:
> > On 2016/04/12 at 17:19:38, bradnelson wrote:
> > > I take it the wasm level test doesn't repro this anymore?
> > > What's the difference in wasm bytes?
> > 
> > Hi,
> > 
> > The problem is that when you pass an int32 to wasm it is first converted to
a float64 and then back to an int64. Through these conversions the -1 gets sign
extended and thereby causes the problem. Do you need the bug fix soon or can it
wait until Thursday?  I attend a workshop today and tomorrow so I don't really
have time for it.
> 
> This CL (https://codereview.chromium.org/1877133004) would fix your problem.

Problem fixed in https://codereview.chromium.org/1877133004

[ahaas, 2016-04-21 06:57:40]

On 2016/04/14 at 07:29:32, ahaas wrote:
> On 2016/04/12 at 18:22:03, ahaas wrote:
> > On 2016/04/12 at 18:13:34, ahaas wrote:
> > > On 2016/04/12 at 17:19:38, bradnelson wrote:
> > > > I take it the wasm level test doesn't repro this anymore?
> > > > What's the difference in wasm bytes?
> > > 
> > > Hi,
> > > 
> > > The problem is that when you pass an int32 to wasm it is first converted
to a float64 and then back to an int64. Through these conversions the -1 gets
sign extended and thereby causes the problem. Do you need the bug fix soon or
can it wait until Thursday?  I attend a workshop today and tomorrow so I don't
really have time for it.
> > 
> > This CL (https://codereview.chromium.org/1877133004) would fix your problem.
> 
> Problem fixed in https://codereview.chromium.org/1877133004

Can we close this CL?

[aseemgarg, 2016-04-21 18:24:06]

Description was changed from

==========
debug subtract error

R=ahaas@chromium.org,bradnelson@chromium.org
BUG=
==========

to

==========
debug subtract error

R=ahaas@chromium.org,bradnelson@chromium.org
BUG=
==========

[aseemgarg, 2016-04-21 18:24:32]

On 2016/04/21 06:57:40, ahaas wrote:
> On 2016/04/14 at 07:29:32, ahaas wrote:
> > On 2016/04/12 at 18:22:03, ahaas wrote:
> > > On 2016/04/12 at 18:13:34, ahaas wrote:
> > > > On 2016/04/12 at 17:19:38, bradnelson wrote:
> > > > > I take it the wasm level test doesn't repro this anymore?
> > > > > What's the difference in wasm bytes?
> > > > 
> > > > Hi,
> > > > 
> > > > The problem is that when you pass an int32 to wasm it is first converted
> to a float64 and then back to an int64. Through these conversions the -1 gets
> sign extended and thereby causes the problem. Do you need the bug fix soon or
> can it wait until Thursday?  I attend a workshop today and tomorrow so I don't
> really have time for it.
> > > 
> > > This CL (https://codereview.chromium.org/1877133004) would fix your
problem.
> > 
> > Problem fixed in https://codereview.chromium.org/1877133004
> 
> Can we close this CL?

closed