x86: 0.611
kernel: 0.467
graphic: 0.461
device: 0.433
boot: 0.407
architecture: 0.325
performance: 0.291
TCG: 0.204
semantic: 0.198
debug: 0.142
ppc: 0.130
PID: 0.117
socket: 0.109
user-level: 0.109
register: 0.103
permissions: 0.078
mistranslation: 0.078
network: 0.071
vnc: 0.068
risc-v: 0.061
virtual: 0.048
files: 0.037
assembly: 0.036
VMM: 0.034
i386: 0.030
arm: 0.026
hypervisor: 0.026
peripherals: 0.026
KVM: 0.020

"accel/tcg: demacro cputlb" break qemu-system-x86_64 on 32-bit x86 host

As described in https://lists.gnu.org/archive/html/qemu-devel//2019-05/msg07362.html I run into TCG regression in qemu-git.

Unfortunately, fix from bug https://bugs.launchpad.net/qemu/+bug/1830872 seems to be nonn-effective for my case.

For reproduction (on 32-bit x86 host, in my case Slackware with gcc 5.5.0):

./configure --target-list=x86_64-softmmu --disable-werror --enable-debug-tcg

make (-j5 in my case)

try to boot any 64-bit kernel:

x86_64-softmmu/qemu-system-x86_64 -kernel /boot/bzImage-4.12.0-x64 -accel tcg

result is - qemu appear to hang right after "Booting the kernel" line. Decompression (xz) was ok.

Tested with qemu-git commit  e2a58ff493a2e00db3e963c1839c5374500110f2 

32-bit OS can be booted fine, and -enable-kvm also allow 64 bit kernel/os to boot.

bug fixed in current git (commit 474f3938d79ab36b9231c9ad3b5a9314c2aeacde). Thanks, Alex!