IRC channel logs

2025-03-12.log

back to list of logs

<gnucode>So I had an interesting thought for maybe how we could raise money to get the Hurd a better filesystem.
<gnucode>I was inspired by the OpenBSD guys that are developing got, the git alternative.
<gnucode>What if the Hurd project offered a dropbox like service? We will host your files for so much $ per month. We will use that revenue to either improve ext2 or build Hurdfs (to be implemented filesystem), or port bcachefs.
<user_oreloznog>gnucode: Hi, maybe that would be a good thing... If it happens, I'll be happy to make a financial contribution.
<gnucode>user_oreloznog: if you have any suggestions as far as hosting...let me know.
<gnucode>I could potentially host it myself...I bought a open hardware-ey router for such a purpose...I still need to set it up though.
<user_oreloznog>gnucode: I'm just an end user of the Free Software, with its technical limits, but yes, if I can help in this way, it will be a pleasure.
<camm>does mremap ever fail with ENOMEM based on the available runtime memory?
<camm>mmap apparently returns wildly over committed maps without error. What is the best way to probe runtime limits to mmaps that can actually be written, at least approximately? MAP_ANON.
<Pellescours>I’m getting page fault when the system is doing a vm_pageout
<Pellescours>at phys_attribute_clear
<Pellescours>Previously (1 month ago), it was working correctly. I’m running on master hurd & gnumach
<Pellescours>(I don’t have a swap configured on this VM)
<camm>how does overcommit work on hurd, e.g. with mmap?
<gnucode>camm: It probably doesn't work very well (but I really don't know).
<camm>Thanks so much. Under Linux, mmap will generally fail with ENOMEM on requests exceeding system memory, but this does not appear to be happening with Hurd.
<camm>Program I'm working on had used this feature to probe available memory.
<gnucode>camm: I think Hurd developers have a list of "easy and wanted contributions". I believe one of them is for setting per application memory.
<camm>Thanks again. What is the recommended best way to probe at runtime how much usable memory the system can deliver via mmap?
<gnucode>Hmm, from my expert opinion...you probably need a magic wand with the magical incantation, "HOW DOES THIS WORK!?"
<gnucode>I really have no idea. The Hurd's documentation is not the best. :(
<camm>thanks again so much!