IRC channel logs

2019-12-15.log

back to list of logs

***Server sets mode: +nt
***JBB-xmpp is now known as jbb
***jbb is now known as JBB-xmpp
***JBB-xmpp is now known as jbb
***jbb is now known as JBB-xmpp
***JBB-xmpp is now known as jbb
***jbb is now known as JBB-xmpp
***JBB-xmpp is now known as jbb
***jbb is now known as JBB-xmpp
***JBB-xmpp is now known as jbb
***jbb is now known as JBB-xmpp
<gnu_srs>fsysopts /servers/socket/26 show that ipv4 addresses are used. That does not work with qemu. should I change qemu or /servers?
<kilobug>gnu_srs: could be in both from what I know, but qemu offer multiple networking options (different emulated cards, and different topology : bridge, nat, ...) perhaps it'll work with some options and not others ?
<gnu_srs>Correction: fsysopts /servers/socket/26 show that ipv6 addresses are used
<gnu_srs>Seems like both ipv4 and ipv6 is set up. ssh-ing _from_ the image works fine, but _to_ it fails. /etc/init.d/ssh start does not start the server.
<gnu_srs>Cannot start openssh-server! Any ideas?
<gnu_srs>Seem like libcrypt.so.1 was a symbolic link to libcrypt-2.25.so shadowing the real libcrypt.so.1.1.0
<gnu_srs>Now when installing exim4-daemon-light it complains on no certificates found, creating a selfsigned certificate?
<z3ntu>I'm geting a compilation error when using the latest glibc and having applied the tg-hurdsig-* patches from Debian (for getting SA_SIGINFO): https://pastebin.com/raw/PDD2pvdy
<z3ntu>I'm not sure why the #define SIGCONTEXT is in debug/segfault.c because it wil redefine the macro which is "struct sigcontext" for Hurd and "struct sigcontext *" for all other platforms. And because the new define has a castable types for everything non-Hurd it works but not for Hurd, I'm guessing
<z3ntu>youpi: haha you joined like half a minute after my last message, reposting them for you (as you probably know what to do)
<z3ntu>I'm geting a compilation error when using the latest glibc and having applied the tg-hurdsig-* patches from Debian (for getting SA_SIGINFO): https://pastebin.com/raw/PDD2pvdy
<z3ntu>I'm not sure why the #define SIGCONTEXT is in debug/segfault.c because it wil redefine the macro which is "struct sigcontext" for Hurd and "struct sigcontext *" for all other platforms. And because the new define has a castable types for everything non-Hurd it works but not for Hurd, I'm guessing
<youpi>Mmm, that's a touchy thing
<youpi>what happens is that the parameters of the signal handler depends whether SA_SIGINFO was set or not
<youpi>the default SIGCONTEXT content is for when SA_SIGINFO is not set
<youpi>segfault.c redefines it because it sets SA_SIGINFO
<AlmuHS>hi
<AlmuHS>I'm checking the " if (processor->state == PROCESSOR_IDLE) {" case in thread_setrun() function of sched_prim.c . But I have some questions
<AlmuHS>this lines seems to be duplicated: https://github.com/AlmuHS/GNUMach_SMP/blob/wip/kern/sched_prim.c#L1349-L1351
<AlmuHS>in " if (processor->state == PROCESSOR_IDLE) {" there are a simple_unlock() of pset and processor. But, after this code, these simple_unlock() are done again
<AlmuHS>mmm... I see a return after this block, but this return don't have any data
<AlmuHS>Is a "return;" without any variable
<AlmuHS>furthermore, this function is "void", so this return is wrong here
<youpi>it's not duplicated, it's done before the return
<youpi>not having data is not a problem
<youpi>that's what is expected for a function returning void
<youpi>there is nothing to return
<youpi>but the function still terminates there
<AlmuHS>It's a few chaotic this code
<AlmuHS>This can be candidate to refactor
<AlmuHS>(but I don't risk to refactor It)
<AlmuHS>I've just added some prints in " if ((processor = th->bound_processor) == PROCESSOR_NULL) {" case, but It seems don't enter here
<AlmuHS>youpi: http://paste.debian.net/1121174/
<AlmuHS>I go to reenable the another prints
***JBB-xmpp is now known as jbb
***jbb is now known as JBB-xmpp
***JBB-xmpp is now known as jbb
***jbb is now known as JBB-xmpp
<z3ntu>youpi: are you planning to make a patch to fix the glibc issue, or do I have to make a dirty patch myself? :)
<youpi>which glibc issue?
<youpi>ah, the siginfo one?
<youpi>I don't plan to, since I don't have the issue, it'll be hard for me to come up with a fix
<AlmuHS>youpi: I was adding more prints in thread_setrun(), and I've get more information http://paste.debian.net/1121192/