For me when an update happens for firefox (not sure if it is a firefox update or just some libraries it uses, or both) it does not crash (that I notice) but simply stops working (web page refresh and other things don't work anymore). So in the firefox case something extreme enough is happening that I am surprised that it stays up and only breaks and does not crash. After that I have to make sure to completely exit firefox and restart it to get it to function again. Firefox has done this across multiple kernels and updates on multiple machines. So maybe something similar (update breakage) is happening for chromium also. On Sun, Jun 30, 2024 at 12:20 PM Chris Adams <linux@xxxxxxxxxxx> wrote: > > Once upon a time, Jeffrey Walton <noloader@xxxxxxxxx> said: > > On Sun, Jun 30, 2024 at 12:52 PM Chris Adams <linux@xxxxxxxxxxx> wrote: > > > After some recent updates, I seem to get an illegal instruction the > > > first time or two I start Chrome/Chromium after boot. If I close it and > > > reopen it a time or two, it'll work like normal from that point on, > > > until I reboot again. It's the weirdest thing, I can't think of > > > anything I've ever seen act like this before. > > > > > > Not sure if this is a problem with Chrome/Chromium, some library > > > dependency, or even the kernel. It's happening for me on both Fedora 39 > > > and 40 (both x86_64). dmesg shows: > > > > > > [ 127.676343] traps: chromium-browse[3675] trap int3 ip:55bcd5b053c6 sp:7fff6061e350 error:0 in chromium-browser[55bccfde1000+d8af000] > > > > It's hard to say whether a problem exists based on the information provided. > > > > In the case of crypto libraries, a common strategy at startup is to > > perform cpu feature testing by trying an instruction in an ISA (like > > SSE4.2, SecureKey or AVX2), and trapping a potential cpu fault. If the > > fault is raised, the ISA is not available. Then the library sets up > > its function calls based on availability. > > > > Video drivers probably perform similar probes at startup. > > > > So to investigate further, we probably need to see a disassembly > > around the offending code, the name of the offending function, and the > > callstack. > > I'm rusty on that kind of thing TBH, haven't needed to do that in many a > year, so I'd need some instructions. It's just weird to me that it goes > away on its own after 2-3 tries. > > I will say it doesn't appear to be video related at least; I'm hitting > this trying to build an RPM in mock that runs Chromium headless in %test > (so no video involved at all). > > -- > Chris Adams <linux@xxxxxxxxxxx> > -- > _______________________________________________ > users mailing list -- users@xxxxxxxxxxxxxxxxxxxxxxx > To unsubscribe send an email to users-leave@xxxxxxxxxxxxxxxxxxxxxxx > Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ > List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives: https://lists.fedoraproject.org/archives/list/users@xxxxxxxxxxxxxxxxxxxxxxx > Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue -- _______________________________________________ users mailing list -- users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to users-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@xxxxxxxxxxxxxxxxxxxxxxx Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
