| Summary: | perl reproducibility issue in perl-dev (host config leaking to target?) | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] OE-Core | Reporter: | Richard Purdie <richard.purdie> | ||||
| Component: | core | Assignee: | Richard Purdie <richard.purdie> | ||||
| Status: | RESOLVED FIXED | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | High | CC: | jpuhlman, meta.mr.watcher, meta.watcher, randy.macleod, ross.burton | ||||
| Version: | 3.0 | ||||||
| Target Milestone: | 3.1 M3 | ||||||
| Hardware: | x86 | ||||||
| OS: | Multiple | ||||||
| Whiteboard: | |||||||
| OS type for building Yocto: | --- | Type of Regression: | --- | ||||
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |||||
| Attachments: |
|
||||||
|
Description
Richard Purdie
2020-02-03 12:15:33 UTC
Filed with perl-cross at https://github.com/arsv/perl-cross/issues/87 Further testing suggests HAS_C99_VARIADIC_MACROS is a dash vs bash issue. Whether that is the only issue, not sure but its one problem. Created attachment 4630 [details]
Output of configure with +x
It appears at a glance that the problem is the test is using the host compiler to determine if the variadic macros are available.
It is a try and run test.
set try
if eval $compile && $run ./try 2>&1 >/dev/null; then
case "`$run ./try`" in
"123 456 789")
echo "You have C99 variadic macros." >&4
d_c99_variadic_macros="$define"
;;
*)
echo "You don't have functional C99 variadic macros." >&4
d_c99_variadic_macros="$undef"
;;
esac
else
echo "I couldn't compile and run the test program, so I assume that you don't have functional C99 variadic macros." >&4
d_c99_variadic_macros="$undef"
fi
$rm_try
The output from the configure run with all the output indicates that it is using the host gcc to make the determination.
nm, was looking at the wrong configure code. HAS_C99_VARIADIC_MACROS is a dash-vs-bash issue and can be fixed by removing a \n. The rest are in a host-specific file that shouldn't be shipped, Alex has a sent a patch. Also seeing HOSTCFLAGS·=·-DPERL_CORE·-DUSE_CROSS_COMPILE·-D_LARGEFILE_SOURCE·-D_FILE_OFFSET_BITS=64 vs HOSTCFLAGS·=·-DPERL_CORE·-DUSE_CROSS_COMPILE·-D_LARGEFILE_SOURCE·-D_FILE_OFFSET_BITS=64·-fwrapv·-fno-strict-aliasing in ptest/Makefile.config which is another piece of host contamination :( (In reply to comment #5) > nm, was looking at the wrong configure code. It gets confusing as the configure code runs twice, once for native for building miniperl, then again for the target... |