qemu user mode fails to run programs with large VM / built with address sanitizer (was: Re: [PATCH v4 4/4] meson: Warn when TCI is selected but TCG backend is available)

[Stefan Weil <sw@weilnetz.de>]

Am 27.01.21 um 22:47 schrieb Alex Bennée:

> Stefan Weil<sw@weilnetz.de>  writes:
>> I recently tried running tesseract with qemu-x86_64 because I had
>> expected that it might trigger some unimplemented TCG opcodes.
> qemu-x86-64 is a poor choice as a relatively under maintained front-end
> doesn't emulate a particularly new CPU or take advantage of the new TCG
> features. ARM64 is pretty good because the default cpu for linux-user is
> CPU max which a) enables all ISA features we have and b) exposes them
> fairly easily to guest detection routines which probe feature registers.
>
>> Instead
>> it showed a general problem for native TCG: qemu-x86_64 allocates too
>> much memory for tesseract and gets killed by the Linux kernel OOM
>> handler.
> Do you have a command line? That sounds like something that should be
> fixed.


The problem occurred with a locally built tesseract, but I now found 
that it is more general.

Any program which was compiled with address sanitizer uses huge virtual 
memory (TB) right at the start. QEMU user mode tries to allocate that 
memory until it is killed by the Linux kernel OOM handler.

A simple hello program compiled with "gcc -fsanitize=address hello.c" is 
sufficient to show the problem. Just run it with "qemu-x86_64 a.out".

I did not test but expect the same problem for other architectures, too, 
unless their VM is more limited.

Regards,

Stefan