Marco Bodrato <bodr...@mail.dm.unipi.it> writes: Just looking at the report, it looks like an incorrect definition of HUGE_VAL by the compiler/headers, or an incorrect handling of the "infinity" value by the arithmetic operations on it. So yes, it can be a qemu bug on floats.
Now the emulated systems on servile/qemuusr (the 3950X system's qemu user level emulator system) have all pre-compiled qemu under /usr/bin/qemu/. It seems that this bug was introduced with qemu 3.1 and fixed in qemu 4.2. (The default qemu used is 4.1.1.) /usr/bin/qemu/qemu-20181126/qemu-ppc mpz_cmp_d wrong (from check_infinity) Aborted /usr/bin/qemu/qemu-3.0.1/qemu-ppc OK /usr/bin/qemu/qemu-3.1.0/qemu-ppc mpz_cmp_d wrong (from check_infinity) Aborted /usr/bin/qemu/qemu-3.1.1/qemu-ppc mpz_cmp_d wrong (from check_infinity) Aborted /usr/bin/qemu/qemu-4.0.0/qemu-ppc mpz_cmp_d wrong (from check_infinity) Aborted /usr/bin/qemu/qemu-4.1.1/qemu-ppc mpz_cmp_d wrong (from check_infinity) Aborted /usr/bin/qemu/qemu-4.2.0/qemu-ppc OK -- Torbjörn Please encrypt, key id 0xC8601622 _______________________________________________ gmp-devel mailing list gmp-devel@gmplib.org https://gmplib.org/mailman/listinfo/gmp-devel
