On Tue, Oct 14, 2014 at 5:18 PM, Jan Engelhardt
On Tuesday 2014-10-14 20:11, Yamaban wrote:
The way I understand it, the installer is built on QT5, which depends on sse2, which is not supported by all i686 CPUs.
The installer is also built on ruby - another source of unconditional SSE2.
The OpenQA cluster definitely needs an AXP in the testsuite. (My offer of donating a fully equipped box still stands, but I might have overheard OQA uses a form of virtualization [which is principially tweakable]...)
Could the OpenQA 32bit machines modified to emulate a non-SSE2 cpu? That would get the trouble visible and fast.
If there is no hardware exception for an illegal instruction, how do you know it's illegal? You would have to analyze/tweak the binary code beforehand, meaning your half-emulation layer turns into a full-blown one, and that appears to be likely a slow business.
That is why it seems easiest to just throw a non-SSE silicon in the loop which will always raise the exception.
The only faster thing is looking at the build logs for low-hanging fruit, I named several packages where this seems to be the case.
Perhaps rpmlint could be taught about the issue? -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org