IRC channel logs

<nikolar>so you did get it to work

<gnucode>sure. I am running netsurf right now.

<gnucode>I just #define PATH_MAX 250

<gnucode>of whatever number I chose.

<gnucode>or*

<gnucode>I'm going to try something in the Hurd console real quick.

<gnucode>be right back

<gnucode>well looks like I can update the open_issues/emacs page.

<gnucode>those 3 commands all worked in the console.

<gnucode>youpi: thanks for fixing sudo!

<youpi>I didn't

<damo22>youpi: i am having trouble with 9600 baud console over a real com port

<damo22>i *think* somehow rumpdisk uses 115200 baud so we get mixture of baud rates during boot

<damo22>should we add a parameter to gnumach cli to set the RCBAUD?

<damo22>but since its already hardcoded, can we just bump it to 115200?

<azert>damo22: how come rumpdisk uses the serial directly? Shouldn’t he pass through gnumach?

<youpi>azert: it does pass through gnumach

<azert>He mentioned that it mixed baudrates

<youpi>I know, but that' cannot be rumpdisk's fault

<youpi>rumpdisk just uses printf

<damo22>ok something is setting the baud rate to 115200 after start acpi:

<damo22>or start xxxx:

<damo22>but the startup baud rate is 9600

<damo22>#ifndef PORTSELECTOR

<damo22>#define ISPEED B9600

<damo22>#define IFLAGS (EVENP|ODDP|ECHO|CRMOD)

<damo22>#else

<damo22>#define ISPEED B4800

<damo22>#define IFLAGS (EVENP|ODDP)

<damo22>#endif

<damo22>is ISPEED always one value?

<damo22>i cant find it

<damo22>tty_set_status?

<damo22>kdsetstat

<damo22>does userspace configure the tty?

<damo22>console "@libexecdir@/getty 9600" mach-gnu-color on secure trusted console

<damo22>task loaded: exec

<damo22>start pci-arbiter:��ci��c�i�r�۠�i�t��ll(�� |

<damo22> ��.0000000ݠ�o�|ri��t�(c��996��99��99��999

<gnucode>morning friends!

<etno>ACTION waves

<gnucode>morning etno

<gnucode>sneek later tell solid_black For now, do you suppose it would be hard to package LadyBird for Debian ?

<sneek>Got it.

<etno>Thanks for the offer the other day, gnucode. I would enjoy a peering session one day. For now, my Inspiron still refuses to start any Hurd server, despite me trying to follow the advice I got here :-/

<gnucode>etno: You can always run it in qemu, but I personally cannot get that to work for me. the Hurd is soo much easier to use on this T43.

<dsmith>Easier to use natively instead of qemu, or easier to use than a different os?

<gnucode>dsmith: Well I personally have only been able to edit the Hurd wiki when running Debian.

<gnucode>I can't quite get ikiwiki set up properly on other distributions.

<gnucode>My goal is to edit the wiki using the Hurd. It's much easier to do that on this T43 than on real hardware.

<gnucode>Debian GNU/Linux is an easier way to edit the Hurd wiki.

<etno>I use a qemu VM to rebuild the parts for troubleshooting. Then trying to get that running on real hardware.

<gnucode>also, I've got emacs installed, but I don't have the emacs manual installed.

<gnucode>or the emacs-lisp manual.

<gnucode>anyone know how to install those?

<etno>Ohh, the instruction on which my Inspiron laptop panics is a XSAVEOPT (0f ae) in fp_save. My hardware probably doesn't have it.

<etno>Why did cpuid not say it ? 🤔

<etno>Both XSAVEOPT and XSAVEC are reported available, and both fail; so it may be something about GPU enablement flag ?

<etno>FPU*

<gnucode>hello friends!

<nikolar>hello gnucode

<gnucode>nikolar: what up dude? Guess what? the company I work for uses a proprietary software that is essential to our business.

<nikolar>isn't that the norm

<nikolar>also i guess i added a syscall to v7 unix on pdp11

<gnucode>I work at a hotel, and our property management software just died. Won't be fixed until Wednesday or Friday. I found an open source PMS. I'm learning how to use it right now. Pretty cool.

<gnucode>And it works!

<gnucode>that's pretty cool!

<gnucode>Did you write it in assembly just for fun?

<nikolar>hitch() {

<gnucode>Or did you go straight to binary?

<nikolar> u.u_r.r_val1 = 42;

<nikolar>}

<nikolar>y

<nikolar>full source

<gnucode>nice!

<azert>etno: files on xsaveopt with state on the stack? Only on real hardware ? Probably a problem with alignment

<azert>Should be 64 bytes boundary

<nikolar>gnucode: it's actually very hackable

<etno>azert: the trap says, "invalid opcode". Would it be the case for an alignment issue?

<azert>No, it would be GP

<azert>But I think the bug is there for real

<etno>Well, that is a real clue ! Thanks azert . I will have it print the save location pointer to check

<etno>Hmm, actually, I would better check the alignment, then print something; this is a pretty hot codepath

<etno>azert: the dest pointer should be in one of the registers already, isn't it?

<etno>EDX:EAX if I am not mistaken

<azert>etno: https://www.felixcloutier.com/x86/xsaveopt

<azert>The pointer argument is the value following the assembly instruction

<gnucode>nikolar: that's pretty awesome!

<azert>Just set EDX:EAX to -1 it will probably work

<nikolar>gnucode: sure is

<azert>Like both registers to -1

<azert>There might be other issues

<azert>Like I’m not sure if XCR0 is set by default

<etno>azert: at the beginning of the fp initialisation routine, a comment says that the FPU needs to be enabled to detect it's capabilities, and it sets bits in the control register.

<etno>What puzzles me completely here is that this is not an exotic CPU, it is an Intel pentium T4400. This code must have been used by quite some people, I guess.

<etno>Or, I am the lucky one, with an exotic Pentium :-D

<azert>I think the Hurd stopped working on real hardware at the time of the pentium 3, and then patches were introduced that might have introduced new bugs

<youpi>I did start hacking on the hurd on my pentium M

<youpi>and booted it several times on various iCore etc.

<azert>So we don’t know if it still works on old hardware. We just know it works on few hardwares

<azert>Ok, youpi knows better for sure

<youpi>well, I don't know what people do in various places :)

<youpi>but it's indeed possible that we haven't tested for hardware before pentium 3

<youpi>I mean, recently

<etno>git blame may give interesting insight then

<azert>Could be cool to do it, would the Hurd run on 128mb of ram?

<azert>Well, even if it does it’s not very useful

<youpi>it would, sure

<youpi>you just won't be able to run much software, probably not apt even

IRC channel logs

2024-01-15.log