Re: HAVE_GCC=10 notes

Hi,

On 2021/04/16 8:23, John Klos wrote:
> I've updated to -current with HAVE_GCC=10, and I've come across a few issues. I'm not entirely sure how much is related to GCC 10, but I can reload a non-GCC 10 -current if it'd help.
> 
> First, over the course of about a dozen reboots, the system locked up twice at this point:
> 
> [   1.0000000] tsciic0 at tsc0
> [   1.0000000] iic0 at tsciic0: I2C bus
> [   1.0000000] entropy: WARNING: extracting entropy too early
> ...
> 
> What normally comes after that is:
> 
> [     1.000000] timecounter: Timecounter "PCC" frequency 999953920 Hz quality 1000
> [     1.000003] timecounter: Timecounter "clockinterrupt" frequency 1024 Hz quality 0
> [     3.591834] scsibus0: waiting 2 seconds for devices to settle...
> 
> Both times I had to use the RMC to do a hardware reset.

I've never seen this problem for my DS10. Maybe because mine is
single processor and yours is MP?

> Next, running a pkg_rolling-replace fails in textproc/gnome-doc-utils consistently here:
> 
> checking for msgfmt... /usr/pkgsrc/textproc/gnome-doc-utils/work/.tools/bin/msgfmt
> checking for gmsgfmt... /usr/pkgsrc/textproc/gnome-doc-utils/work/.tools/bin/msgfmt
> [1]   Segmentation fault (core dumped) ${XGETTEXT} --version |
>       Done(1)                 grep "(GNU " 2>/dev/null
> configure: error: GNU gettext tools not found; required for intltool
> *** Error code 1
> 
> Stop.
> make[1]: stopped in /usr/pkgsrc/textproc/gnome-doc-utils
> *** Error code 1

This is not a regression for GCC10. devel/gettext-tools seems to be
miscompiled by GCC9 and GCC10. (IIRC, there is no such a problem in
GCC8 era.) As a workaround, building devel/gettext-tools with -O0
fixes problems for me.

> Finally, when trying to run atf tests, the testing stops here:
> 
> tc-start: 1618465564.997950, bytes_transfer_eof_piod_write_i
> tc-end: 1618465565.22800, bytes_transfer_eof_piod_write_i, passed
> tc-start: 1618465565.23292, bytes_transfer_eof_pt_read_d
...(snip)...

This is also not a regression for GCC10. This problem was reported as
port-alpha/55652:

	http://gnats.netbsd.org/55652

Many tests in tests/lib/libc/sys cause similar lock-up. I just skip
this directory when I test alpha.

Thanks,
rin