I think, that I have the same issue. After some git bisect, I found out that commit ea32aaf1a72af102b855317b47a22e75ac2965a9 has introduced the problem. Attached is a patch that fixes the issue for me. Maybe you can try this out, too.
** Patch added: "0001-target-i386-Fix-conversion-infinity-from-float80-to-.patch" https://bugs.launchpad.net/qemu/+bug/1477683/+attachment/4433257/+files/0001-target-i386-Fix-conversion-infinity-from-float80-to-.patch -- You received this bug notification because you are a member of qemu- devel-ml, which is subscribed to QEMU. https://bugs.launchpad.net/bugs/1477683 Title: FPU in qemu-system-i386 works incorrectly Status in QEMU: New Bug description: FPU bug in qemu-system-i386 makes software which use floating point numbers work incorrectly. For instance, the one included in attachment prints out 0 instead of 2147483648. The same code works ok in qemu- system-x86_64. I have this issue in QEMU 2.3.0 on two different GNU/Linux guests (Parabola GNU/Linux-libre and libreCMC). To manage notifications about this bug go to: https://bugs.launchpad.net/qemu/+bug/1477683/+subscriptions
