Re: [Qemu-discuss] ARM: timedelta # of days is too large: 1000000000

[Peter Maydell]

On 2 July 2018 at 13:27, Nerijus Baliunas <address@hidden> wrote:
> Hello,
>
> when running virt-2.9 machine with /usr/bin/qemu-system-arm on Fedora 28,
> I get this on F28 arm VM:
>
> # dnf update
> Traceback (most recent call last):
>   File "/usr/bin/dnf", line 57, in <module>
>     from dnf.cli import main
>   File "/usr/lib/python3.6/site-packages/dnf/__init__.py", line 23, in 
> <module>
>     import dnf.exceptions
>   File "/usr/lib/python3.6/site-packages/dnf/exceptions.py", line 21, in 
> <module>
>     from dnf.i18n import ucd
>   File "/usr/lib/python3.6/site-packages/dnf/i18n.py", line 22, in <module>
>     from dnf.pycomp import unicode
>   File "/usr/lib/python3.6/site-packages/dnf/pycomp.py", line 24, in <module>
>     import email.mime.text
>   File "/usr/lib/python3.6/email/mime/text.py", line 10, in <module>
>     from email.mime.nonmultipart import MIMENonMultipart
>   File "/usr/lib/python3.6/email/mime/nonmultipart.py", line 10, in <module>
>     from email.mime.base import MIMEBase
>   File "/usr/lib/python3.6/email/mime/base.py", line 9, in <module>
>     import email.policy
>   File "/usr/lib/python3.6/email/policy.py", line 6, in <module>
>     from email._policybase import Policy, Compat32, compat32, 
> _extend_docstrings
>   File "/usr/lib/python3.6/email/_policybase.py", line 9, in <module>
>     from email.utils import _has_surrogates
>   File "/usr/lib/python3.6/email/utils.py", line 30, in <module>
>     import datetime
>   File "/usr/lib/python3.6/datetime.py", line 655, in <module>
>     microseconds=999999)
>   File "/usr/lib/python3.6/datetime.py", line 447, in __new__
>     raise OverflowError("timedelta # of days is too large: %d" % d)
> OverflowError: timedelta # of days is too large: 1000000000
>
> I found in https://github.com/CopterExpress/clever/issues/21 that it is
> "error between Python and C type conversion on use qemu-static interpreter. 
> The number 999999999 is rounded to 1000000000."

Your use case is not using qemu-static, it's using the system emulation
mode. It does look like it's a similar thing you're hitting though.

Could you please try this with a newer version of QEMU?
(ideally, try with current head of git; we fixed some floating
point emulation issues recently.)

Presumably this code works on real arm hardware ?

thanks
-- PMM