Re: Failure of the xxx test on s390x

[Dan Horák <dan@...>]

Hi Matej,

On Tue, 14 Nov 2023 15:22:25 +0100
Matěj Cepl <mcepl@cepl.eu> wrote:

> When trying to run the Lua test suite for 5.4.6 on IBM zSystem
> (with either SUSE SLE-15 or openSUSE/Tumbleweed), I get this test
> failing (line 355 of testes/calls.lua):
> 
>   do   -- another bug (since 5.2)
>     -- corrupted binary dump: list of upvalue names is larger than number
>     -- of upvalues, overflowing the array of upvalues.
>     local code =
>      "\x1b\x4c\x75\x61\x54\x00\x19\x93\x0d\x0a\x1a\x0a\x04\x08\x08\x78\x56\z
>       \x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x28\x77\x40\x00\x86\x40\z
>       \x74\x65\x6d\x70\x81\x81\x01\x00\x02\x82\x48\x00\x02\x00\xc7\x00\x01\z
>       \x00\x80\x80\x80\x82\x00\x00\x80\x81\x82\x78\x80\x82\x81\x86\x40\x74\z
>       \x65\x6d\x70"
>   
>     assert(load(code))   -- segfaults in previous versions
>   end
> 
> with "binary string: bad binary format (integer format
> mismatch)". The complete log of the test run is on
> https://mcepl.fedorapeople.org/tmp/_log-lua54-s390x.txt .
> 
> Any ideas how to debug more, how to provide more detailed logs
> (debugging on s390x is a bit bother)?
> 
> Thank you in advance for any reply,

please my email from August 1st [1] and its follow-ups :-)

[1]
https://listmaster.pepperfish.net/hyperkitty/list/lua-l@lists.lua.org/message/DCLRVJ74XRDKG75623VB6L2XVS6KCAUH/


		Dan