On Mon, Jun 08, 2020 at 06:11:53PM +0300, Αγαθοκλής wrote: > On Sun, Jun 07, at 10:51 Bruce wrote: > > On 6/7/20 10:35 AM, Ken Moffat wrote: > > > On Sun, Jun 07, 2020 at 03:43:30PM +0100, Ken Moffat wrote: > > > > Finally replying to this: > > Hi Ken, > > I was meant to ask you about this, to verify that the issue was really > -march=native, > because in that case it would be a compiler bug. Those days -march=native is > almost > assumed, as it is almost guarranteed that the results will be the same in > that same > machine, so it looked a bit strange to affect the instruction set. >
Hi Ag, as I've now seen, it was not -march=native. But using that does mean restrictions on moving the binaries to a new machine (for those of us, maybe only me, who have given up on distros). e.g. I managed to move from AMD Fam15 (Kaveri, it was so slow from lack of CPU cache that it had become not useful) to Zen+ using all components except motherboard, CPU, RAM - but I had to rebuild quite a few LFS packages to get them all working, and I don't recommend the process. > > > > > > So, I've been looking in the wrong place, it's embedded perl in > > > rxvt-unicode: google found > > > http://lists.schmorp.de/pipermail/rxvt-unicode/2013q3/001839.html > > > > > > and earlier in that thread was the link to the still-open issue > > > at perl: https://github.com/Perl/perl5/issues/12349 > > > > The issue has been around since 2001 ??!! > > > > Perhaps it's time for you to use another terminal emulator. > > I use urxvt since my gentoo days at 2004, and honestly i do not remember a > single > crash all those years, despite that i use the daemon/client functionality, > which has > a bit of risk, since if an instance/client crashes, all the other clients > will crash > too. Besides that, it is really run low in resources and was one of the first > terminals > that supported UTF-8. Granted I do not bind perl though these days. > My build of urxvt uses the defaults (whatever they are, I have not looked for a while). I'm just coming to the end of an LFS trunk build on another system (this weekend's builds were cross-chap5) where the host (gcc-9) used -O3 -march=native for glibc : using xfce4-terminal I didn't get sigfpe in diffs from the bash tests, and all of the check tests passed. What misled me initially when I looked at my logs re the check tests was that the machines where I'd run the tests and they all passed were low-powered (SandyBridge i3 with very slow memory, athlon200ge). What I did not spot was that those tests were also on my server and test-server, so not building from within Xorg. > > -- Bruce > > Regards, > ag ĸen -- +++ OUT OF CHEESE ERROR. REDO FROM START +++ -- http://lists.linuxfromscratch.org/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page Do not top post on this list. A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail? http://en.wikipedia.org/wiki/Posting_style
