IRC channel logs
2026-09-30.log
back to list of logs
<youpi>there's nothing special about it <GNU_Hurd_Rocks>ah apparently I couldn't do an "attach <pid>" in gdb because of ptrace_scope, fully understandable why it would be set the way it is though, linux stuff <youpi>the hurd console can be told what font to use <GNU_Hurd_Rocks>so perhaps I could use unifont then, will have to look into this later after dealing with some things <nbowler>Hi! I've been playing around with Gentoo GNU/Hurd and am now troubleshooting a boot hang on my PC... <nbowler>Last message printed normally is "Hurd server bootstrap: ext2fs[device:hd1s2] exec startup ... proc auth." <nbowler>With --verbose option we get to "Received startup essential message from: ... still waiting for: none!" <nbowler>I added more prints to startup.c, it seems in init_stdarrays the call to: __USEPORT (AUTH, auth_makeauth( ... )); never returns. <nbowler>At this point I'm a bit out of my depth on how to debug this further... any ideas about next steps? <nbowler>oh actually... I realized the exact same thing happens in qemu-system-i386 if I match the amount of memory... qemu-system-i386 -m 896M ... <damo22>nbowler: im not sure how much memory is strictly required but that seems small <nbowler>If I run some other size e.g., qemu-system-i386 -m 832M then I can boot successfully in emulation. <nbowler>I guess I can try popping out one of the dimms <nbowler>oh fantastic! grub command: cutmem 832M 1G and it boots! <etno>youpi: with the confirmation that gdb attach is known to work, I can dig further confidently, thanks. I hit this problem with debian/amd64 100% for more than a year. <etno>As specified in the man, waitpid() can only be called on a child. Linux seems to have a quirk when attaching to a process that allows waitpid(). But I can't figure out how this is supposed to work with Hurd. <etno>The glibc logic to ignore signals when traced makes sense. There is also the port used to grant a tracer process... but something is not right before all that ; ptrace(ATTACH) should do something more, and this is not happening over here. <etno>It reproduces with debian/i386. I might be doing something terribly stupid <etno>- gdb /usr/bin/sleep $(pidof sleep) <bigmike-sit91>hello! i installed the 64 bit Hurd kernel on a latitude e6420 but i noticed that ethernet speeds dropped down to 72Kb/s compared the linux partition, have any of you also noticed this especially if you use similar hw? thanks <Alicia>I think I've seen conversations about that. I don't remember if it's tied to hardware <bigmike-sit91>i wonder what the cause is. its like 3 full packets per second so maybe its like interrupts not arriving properly? idk <bigmike-sit>> I think I've seen conversations about that. I don't remember if it's tied to hardware <bigmike-sit>Seems to be, looking at the irc history other people on similar hw (same gen, same eth chip, like a t420) they also have the same issue with rumpnet and pfinet. interesting! 