IRC channel logs


back to list of logs

<damo22>youpi: im not sure if you saw my bug-hurd email, do you think i should merge and push to debian's rumpkernel package?
<youpi>damo22: heh, I happened to go through it in my mailbox a minute ago, now processing
<youpi>(but got interrupted by something else)
<damo22>libpciaccess seems to be the root dep
<damo22>ah yes, thanks i need to fix the libstore patch
<damo22>but i want to know when i should push to rumpkernel, i dont want to lose my local tree, and it works now
<damo22>(with the pciaccess patches)
<damo22>can i push this to rumpkernel?
<damo22>i mean to this repo :
***Server sets mode: +nt
<damo22>ah nice thanks for your emails
<damo22>ive started this conversation upstream
<damo22>i think busspace_unmap is unimplemented in rump
<damo22>it could be trying to unmap correctly after the probe but hits that stub
<damo22>i would need to add the hypercall to rump
<damo22>and then implement the call to pciaccess
<damo22>but it means i need to probably change the src-linux tree as well
<damo22>i will try to add it in, and put a dummy stub into pci-userspace implementation for linux, it should be fine
<damo22>i fixed rump to not require the duplicate region mappings, it cleanly unmaps the regions instead
<damo22>pushed :)
***rekado_ is now known as rekado
<damo22>youpi: does is_master_device() work on custom master devices? I am emulating it via a node
<damo22>or maybe i should remove the check from ds_device_open
<youpi>see what it does, it just checks that it's in the proper bucket
<youpi>so I believe that it would work in your case as well
<youpi>since it worked for netdde too
<damo22>but doesnt netdde use the master device itself from mach?
<youpi>I believe here it's talking about the master node that applications get when opening the /dev/nettde node
<youpi>Pellescours: I fixed the linking issue, it's the +0xC0000000 which was incorrect
<damo22>weird my laptop rebooted while i was out
<damo22>i tested the split libmachdev/libmachdevrump/rumpdisk.static, it mounts
<janneke>finally, "hello" now builds natively on the #guix " core-updates" branch!
<janneke>native, running guix on Debian GNU/Hurd, that is
<youpi>yay janneke :D
<janneke>yes, vey happy with this
<youpi>ideally we'd have a CI of this, so we get to be notified when we break it
<youpi>instead of fixing the bits much later on :)
<janneke>i agree, rekado already offered to add hurd vms but we did not really have a useful target to build until now
***Emulatorman_ is now known as Emulatorman
***nckx is now known as SecondChoiceNick
***SecondChoiceNick is now known as nckx
***bq is now known as dp
***dp is now known as phillid