[Bug 89508] Bad int(floatBitsToInt(vec4))

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=89508

Iago Toral <itoral at igalia.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #114257|0                           |1
        is obsolete|                            |

--- Comment #7 from Iago Toral <itoral at igalia.com> ---
Created attachment 114260
  --> https://bugs.freedesktop.org/attachment.cgi?id=114260&action=edit
Simplest piglit shader_test showcasing the problem

This is the simplest version of the shader test I could get. By comparing the
generated vec4 code for the case that fails:

addr0 = int(floatBitsToInt(temps[5].xxxx))

and the case that passes:

addr0 = int(floatBitsToInt(temps[10].x));

it seems that the problem is that we are losing this code in the failed
version:

mov(8)    g126<1>UD     g0<4,4,1>UD        { align16 WE_all 1Q };
mov(1)    g127<1>UD     0x0000000aUD       { align1 WE_all compacted };
mov(1)    g127.4<1>UD   0x0000000bUD       { align1 WE_all compacted };
send(8)   g27<1>F       g126<4,4,1>F
    data ( DC OWORD dual block read, 255, 0) mlen 2 rlen 1 { align16 1Q };

so we are losing a variable load into a register that for some reason.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20150312/33a4ff31/attachment.html>