IRC channel logs

<emilytrau[m]>ACTION uploaded an image: (3664KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/WUybOtgwmYsfvAaUeeYwsmua/ima_56f6fbb.jpeg >

<emilytrau[m]>my risc-v dev board is here! so much testing to do 😋

<muurkha>awesome! a VisionFive 2!

<Irvise_>Ö

<nektro>i think its cool how matrix submits attachments at /me statements on irc, really creative

<[exa]>emilytrau[m]: c'mon how can you do this to us, picture of a closed box :D

<emilytrau[m]>courtesy of the risc-v developer board program. keep an eye on it 👀 https://riscv.org/risc-v-developer-boards/details/

<emilytrau[m]><[exa]> "emilytrau: c'mon how can you..." <- so many goodies in this box! comes with power supply and acrylic case!

<janneke>emilytrau[m]: oh, nice

<janneke>i'll be interested how fast it is, compared to say, running QEMU on my laptop

<emilytrau[m]>if you're curious happy to share ssh access between friends :))

<janneke>emilytrau[m]: that's very kind, i may come back on that :))

<emilytrau[m]>stikonas if u wanna do some testing hit me up! ☝️

<stikonas[m]>It's fine. mescc can be developed with x86 mes interpreter

<[exa]>emilytrau[m]: I was actually considering to buy one as a mailing desktop, any comments on how feasible that is very welcome

<[exa]>(actually buy two and use the second one for kernel gore)

<Irvise_>Afaik, the board is faster than a RPi 3B but slower than RPi 4. it is much slower in tasks that require HW acceleration.

<pder>rickmasters: I was curious to try live-bootstrap in VirtualBox and ran into a strange issue before Fiwix is booted.

<pder>I started by running rootfs.py -b -p to create sysa.img

<pder>Then I ran vboxmanage convertfromraw sysa.img sysa.vdi

<pder>I booted a VM with this sysa.vdi hard drive attached and everything proceeded as normal through building mes, tcc, and make

<pder>The last lines I see before the VM reboots are as follows: cp gzip.c.new gzip.c

<pder>make

<pder>make: *** No targets. Stop.

<pder>Then the Subprocess error ABORTING HARD several times followed by Build Finished and ERROR: no hda writes!

<rickmasters>pder: hmm, strange. I can try to reproduce that but there may be a delay as I'm in the middle of something.

<rickmasters>pder: I know you're around a lot so hopefully I'll be able to ask you for more details if necessary.

<rickmasters>pder: just checking, have you tried with qemu?

<pder>qemu mode works perfectly for me. I'll try playing with a few settings in VirtualBox. Does builder-hex0 require PAE by any chance?

<stikonas>janneke: I've just pushed a couple of fixes and now global-array and local-static-array both work, though anything more complicated such as scaffold/tiny-mes crashes

<stikonas>one was my typo, one issue was in original patch

<stikonas>(32 vs 64-bit issue)

<rickmasters>pder: no doesn't require PAE but it does make some assumptions about mapped memory:

<rickmasters>pder: https://github.com/ironmeld/builder-hex0/blob/main/builder-hex0-x86-stage2.hex2#L68

<rickmasters>pder: how much memory are you providing?

<pder>4096MB

<rickmasters>pder: ok

<pder>I am seeing a different sha256 checksum of the sysa.vdi image compared to when it was originally created. Does builder-hex0 write to the disk?

<rickmasters>pder: no

<pder>ok, could just be virtualbox modifying the file

<rickmasters>pder: another possibility is small disk geometry that causes cylinder to exceed 1024

<rickmasters>pder: if there is a monitor that allows examining memory then check 0x7e52

<rickmasters>pder: that stores max_head reported by BIOS: https://github.com/ironmeld/builder-hex0/blob/main/builder-hex0-x86-stage2.hex0#L213

<rickmasters>pder: the current disk CHS read location might be relevant:

<rickmasters> https://github.com/ironmeld/builder-hex0/blob/main/builder-hex0-x86-stage2.hex0#L74

<rickmasters>pder: I have to step away for a while; will check back later

<rickmasters>pder: one last thing, the address of last file by be relevant:

<rickmasters> https://github.com/ironmeld/builder-hex0/blob/main/builder-hex0-x86-stage2.hex0#L815

<rickmasters>s/by/might

<pder>rickmasters: thanks for your help. might be a bit later for me to try your suggestions

<janneke>stikonas: great work!

<stikonas>actually, if I build scaffold apps against full .a libraries from meslibc they also crash (though whey I selected some small set of c files and cat'ed them they passed)

<rickmasters>pder: I reproduced the problem with VirtualBox and will let you know when I've make any progress.

<stikonas>it might be interesting if somebody tries it on real hardware too...

<stikonas>but one has to be careful to not accidentally format your data if that's your main system

IRC channel logs

2023-05-24.log