[Mesa-stable] cherry-pick request ffeb77c7 for 10.2 and 10.3 branches

[Roland Scheidegger]

Am 23.09.2014 02:17, schrieb Emil Velikov:
> On 20/09/14 01:35, Roland Scheidegger wrote:
>> The following commit should be added to 10.2 and 10.3 branches, I forgot
>> about it when I commited it, it fixes
>> https://urldefense.proofpoint.com/v1/url?u=https://bugs.freedesktop.org/show_bug.cgi?id%3D83570&k=oIvRg1%2BdGAgOoM1BIlLLqw%3D%3D%0A&r=F4msKE2WxRzA%2BwN%2B25muztFm5TSPwE8HKJfWfR2NgfY%3D%0A&m=gjL3qmqrfPKgarYiong6T3EiIwTx3RUEOK7NslUilAg%3D%0A&s=aba6b82de763b9d3d2c6418d0244e4a4a9a0e5a30f8be32c0f995817ef103fcc (which I forgot to
>> mention in the commit message too...)
>>
>> commit ffeb77c7b0552a8624e46e65d6347240ac5ae84d
>> Author: rconde <rconde01 at outlook.com>
>> Date:   Wed Sep 17 18:30:23 2014 +0200
>>
>> gallivm,tgsi: fix idiv by zero crash
>>
>> While the result of signed integer division by zero is undefined by glsl
>> (and doesn't exist with d3d10), we must not crash, so need to make sure
>> we don't get sigfpe much like udiv already does.
>> Unlike udiv where we return 0xffffffff (as required by d3d10) there is
>> no requirement right now to return anything specific so we use zero.
>>
> Added the bugzilla entry, and pushed to both branches. The commit seems
> to have some gracious trailing white space but I've kept the cherry-pick
> as the original :)
> 
> 
> Thanks
> Emil
> 


Ah yes my bad didn't notice the whitespace errors when commiting. Though
the ones in the comment are actually already from copy/paste from other
code...
With some luck the code will work correctly even with whitespace errors
however ;-).

Thanks,
Roland