IRC channel logs

2022-11-08.log

back to list of logs

<damo22>i have reached a really difficult part of trying to add smp support. My rebased/cleaned up version of the code no longer works, but i cant go back to the old one because it doesnt have an essential part of the changes.
<damo22>so im kind of stuck for the moment
<damo22>youpi: build$ ../configure --enable-kdb --enable-ncpus=1 --disable-apic --disable-pae
<damo22>is broken on master :(
<damo22>it seems to hang in linux_init
<damo22>waiting for jiffies to tick over
<youpi>did you try to revert 3202d5bb4ae9200eb05858e3bdfae13d6d1ecc18 or 21982d148b72095fb37f9b54df2292ddcd9928b3 ?
<youpi>(I had assumed that you had tested them)
<damo22>i thought i did test it, maybe not in complete isolation
<damo22>reverting 21982d14 has no effect
<damo22>i'll find the culprit
<damo22>even this one is still broken so its earlier than this * cf66707d (HEAD) kdb: Fix TODO for multiprocessor
<damo22>i dont understand this
<damo22>let me do a clean build of a tag
<damo22>clean build of v1.8+git20220827 works
<damo22>* 44f67d31 (HEAD) std_types: Drop char, short and int type definitions
<damo22>this one does not work
<damo22>one of these commits causes some kind of timer interrupt problem:
<damo22>* 3b5f8a07 x86_64: ignore warning
<damo22>* e2fcf261 interrupt: Fix saving irq/ipl when linux drivers are disabled
<damo22>* f6a7b166 kd_mouse: Fix warning
<damo22>* e64b35e9 x86_64: fix warning
<damo22>* 3c499e27 x86_64: fix build
<damo22>* 66775cd2 hack vm memory object proxy creation for vm arrays
<damo22>* 4f82ae66 copyinmsg: Set msgh_size inside copyinmsg rather than the caller
<damo22>* 30c93547 kmsg: factorize uint32_t into an alignment type
<damo22>* 286ae33f kmsg: fix msg body alignment
<ThinkT510>only one or could it be a combination?
<damo22>* e2fcf261 interrupt: Fix saving irq/ipl when linux drivers are disabled
<damo22>it boots here on master with this commit reverted using native gcc 11.3.0 otherwise gets stuck with no timer running
<youpi>master boots fine here :/
<damo22>with that commit reverted, i seem to be able to boot my refactored branch without smp enabled
<damo22>its indeed some strange behaviour
<gnu_srs1>Hi. Lately the console in qemu does not echo the commandline when in recovery mode.
<gnu_srs1>E.g. e2fsck -y /dev/hd0s1 has to be entered blindly :(
<youpi>you can run "stty echo" to get echo back
<youpi>it's sysvinit-core that is responsible for running the emergency shell, so you can have a look at downgrading there for instance
<gnu_srs1>youpi: Thanks. stty echo works. I'll try to find which version of sysvinit-core made the change.
<gnu_srs1>And: Why not submit the patch [PATCH]: Fix static-pie on Hurd target to Debian gcc maintainers so that golang-go can be installed?
<gnu_srs1>Since gcc upstream is not very responsive. Alternately build a 12.2.0-9+hurd.1
<youpi>golang-go is already installed
<youpi>buildds are building packages
<gnu_srs1>I could install gcc-12.2.0-9 just now. Is the patch for static-pie no longer needed?
<gnu_srs1>Or is it upstreamed?
<youpi>it is
<youpi>just not for golang-go
<gnu_srs1>I can submit a patch to Debian for static-pie if you want. There are still two patches in Debian that are not attended upstream, though reported.
<iska>does rust-gcc work?
<jab>iska: I don't think that the Hurd supports Rust yet...
<iska>jab I think it should support rust-gcc
<iska>because it supports gcc fully, of course..
<jab>iska: but I believe that rust has to be ported to the OS too. :( I do not believe that has happened yet.
<iska>jab I think it's the libc stuff.. I may try porting it.
<jab>I imagine that would be a welcome contribution.
<iska>I really want icecat to run there
<jab>iska: you can probably get an old version to work
<iska>jab a 4x version of iceweasel was shown in one of the screenshots.
<youpi>gnu_srs: btw, thanks for the golang-go work, that unlocked quite a few packages :)
<youpi>(now locked again due to the icu transition, but that'll sort out quickly)
<artuyild> /msg NickServ VERIFY REGISTER artuyild XYRvLUtQ5jRO8Nvb
<damo22>fail
<zamfofex>Thankfully this looks like a generated password, so that???s good at least. ????