Per the error log, the glGetFloatv return value are bound to [-32768, 32768]

 <Text>glGetFloatv(GL_VIEWPORT, 0x00007fffc202fca0);</Text>
 <Text>// data = { -1729, 32768, 7307, 4013 }</Text>
 <Text>// ERROR: expected -1729, 33539, 7307, 4013</Text>
 <Text>glGetError();</Text>
 <Text>// GL_NO_ERROR returned</Text>
 <Text>glViewport(41066, -9108, 11099, 30024);</Text>
 <Text>glGetFloatv(GL_VIEWPORT, 0x00007fffc202fca0);</Text>
 <Text>// data = { 32768, -9108, 11099, 30024 }</Text>
 <Text>// ERROR: expected 41066, -9108, 11099, 30024</Text>
 <Text>glGetError();</Text>

Anuj believes this is a bug in the tests, and submitted a patch to fix that:
https://gitlab.khronos.org/opengl/cts/merge_requests/399

Not sure what the status is.

(In reply to Kenneth Graunke from comment #2)
> Anuj believes this is a bug in the tests, and submitted a patch to fix that:
> https://gitlab.khronos.org/opengl/cts/merge_requests/399
> 
> Not sure what the status is.

Thanks, Graunke. It's very helpful. I am curious why the viewport coordinates were not clamped on HSW, I use the same test app and driver binary.

I got the answer in code, the OES_viewport_array is only enabled when gen >= 8

set_viewport_no_notify

   /* The GL_ARB_viewport_array spec says:
    *
    *     "The location of the viewport's bottom-left corner, given by (x,y),
    *     are clamped to be within the implementation-dependent viewport
    *     bounds range.  The viewport bounds range [min, max] tuple may be
    *     determined by calling GetFloatv with the symbolic constant
    *     VIEWPORT_BOUNDS_RANGE (see section 6.1)."
    */
   if (ctx->Extensions.ARB_viewport_array ||
       (ctx->Extensions.OES_viewport_array &&
        _mesa_is_gles31(ctx))) {
      x = CLAMP(x,
                ctx->Const.ViewportBounds.Min, ctx->Const.ViewportBounds.Max);
      y = CLAMP(y,
                ctx->Const.ViewportBounds.Min, ctx->Const.ViewportBounds.Max);
   }

Both Pyry and Daniel have reviewed my respin of Anuj's change, so I expect it'll be landing in the CTS soon.  Then I'll need to patch dEQP.