Re: [Qemu-devel] [PATCH] target/i386: Return 'indefinite integer value' for invalid SSE fp->int conversions

From: Richard Henderson <rth@twiddle.net>
To: Peter Maydell <peter.maydell@linaro.org>, qemu-devel@nongnu.org
Cc: "Paolo Bonzini" <pbonzini@redhat.com>,
	"Alex Bennée" <alex.bennee@linaro.org>,
	"Eduardo Habkost" <ehabkost@redhat.com>
Subject: Re: [Qemu-devel] [PATCH] target/i386: Return 'indefinite integer value' for invalid SSE fp->int conversions
Date: Mon, 5 Aug 2019 14:13:14 -0700	[thread overview]
Message-ID: <d788f3ff-6814-117d-942d-7870b6201f30@twiddle.net> (raw)
In-Reply-To: <20190805180332.10185-1-peter.maydell@linaro.org>

On 8/5/19 11:03 AM, Peter Maydell wrote:
> The x86 architecture requires that all conversions from floating
> point to integer which raise the 'invalid' exception (infinities of
> both signs, NaN, and all values which don't fit in the destination
> integer) return what the x86 spec calls the "indefinite integer
> value", which is 0x8000_0000 for 32-bits or 0x8000_0000_0000_0000 for
> 64-bits.  The softfloat functions return the more usual behaviour of
> positive overflows returning the maximum value that fits in the
> destination integer format and negative overflows returning the
> minimum value that fits.
> 
> Wrap the softfloat functions in x86-specific versions which
> detect the 'invalid' condition and return the indefinite integer.
> 
> Note that we don't use these wrappers for the 3DNow! pf2id and pf2iw
> instructions, which do return the minimum value that fits in
> an int32 if the input float is a large negative number.
> 
> Fixes: https://bugs.launchpad.net/qemu/+bug/1815423
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> I've tested that this fixes the LP:1815423 test case. If anybody
> has an x86 VM image to hand that has node.js installed it would
> also be useful to test the operations in
> https://bugs.launchpad.net/qemu/+bug/1832281
> (I don't have such a VM.)
> 
> The other approach here would be to make the softfloat functions be
> flexible enough to allow this behaviour -- from my reading of IEEE754
> I think the exact returned result for 'invalid' inputs for float to
> int conversions is not specified.
> 
>  target/i386/ops_sse.h | 88 +++++++++++++++++++++++++++++--------------
>  1 file changed, 60 insertions(+), 28 deletions(-)

I guess this is exactly what we already do in fpu_helper.c.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~