Monday, 2023-03-20

*** toshywoshy <toshywoshy!> has quit IRC00:43
*** toshywoshy <toshywoshy!> has joined #libre-soc00:44
*** openpowerbot_ <openpowerbot_!> has quit IRC00:44
*** tplaten <tplaten!~tplaten@> has quit IRC00:56
*** openpowerbot_ <openpowerbot_!> has joined #libre-soc01:17
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC01:35
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc01:55
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC02:00
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc02:27
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC02:34
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc02:37
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC03:08
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc03:08
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC04:18
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc04:46
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC04:50
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc05:12
*** pangelo[m] <pangelo[m]!~pangeloma@2001:470:69fc:105::3ec5> has joined #libre-soc07:38
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC08:06
markoslkcl, ok, I got it to load the binary file, but strangely both files start with 0x10 as first byte and this does not correspond to an actual asm instruction so the simulator fails08:52
markosah, I also need to run the ld before the objcopy08:54
markosyup, that was it09:05
markos"Cryptographic tests passed" on x8609:20
lkclremember to git push09:27
* lkcl trying to find the bugreport09:28
markoswill fix all the other programs and then push everything (av1/libvpx)09:31
markosok, that's a big difference, 3m53s on my Xeon vs 12m25s on my Power9 :(10:14
markosyeah, only libvpx is possible to enable for x86, av1/dav1d has other ppc asm code in there in inline asm and that would require significant rework10:34
lkclrats, needs the version of binutils with -mlibresoc, but the wrappers compile fine10:49
lkcl(i have binutils compiled/installed in a different path, will test later)10:49
lkclmmm and... the functions are likely to be different... you couldn't just take the x86-assembler-optimised version and expect its low-level functions to be the same10:50
markoswell, yes ofc, our binutils is required for all this to work10:56
markoswhat functions are you referring to? for dav1d?10:57
markosthe ones that can work with minimal effort, are libvpx and chacha20 which is already done10:58
markosworking on libvpx now there was some change in the gtest api which I'm working on fixing now -the x86 system is debian 12 but the power9 one is buster10:59
markosah no, that's incorrect11:09
markosthe schroot is also buster11:09
markosright, needs the buster-backports libgtest-dev package installed11:14
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc12:02
lkclmmmm be very *very* careful with backports. they may need pinning12:36
lkclalso they are a "moving target" which utterly breaks (makes a mockery of) the concept "reproducible builds"12:36
markoswell, the buster libgtest-dev is old and didn't work with libvpx12:40
markosso I had to upgrade12:40
lkclmeh. source would in some ways be better13:17
lkclit only depends on googletest, which in turn depends only on python (any version)13:18
markosI honestly didn't want to mess with bringing in the whole of googletest in our git13:19
lkclit doesn't need modification so wouldn't need to be hosted on git.libre-soc.org13:20
lkclalthough a case could be made "yes it should be"13:20
lkclsigh ok yes it needs adding to the devscripts13:20
* lkcl ermermerm... hunting...13:20
lkclthe only reason that doesn't break is because of this:13:22
lkcl 132 # oooo annoying, pin preferences for buster-backports13:22
lkcl 133 # make it low pririty for reproduceability13:22
lkcl 134 cat << EOF > /opt/chroot/$chrootdir/etc/apt/preferences.d/99buster-backports13:22
lkcl 135 Package: *13:22
lkcl 136 Pin: release a=buster-backports13:22
lkcl 137 Pin-Priority: 10013:22
lkclotherwise all hell breaks loose13:22
* lkcl working on ls00713:25
lkcljust added a "wording" section to it.  every RFC needs an english language description of the pseudocode13:26
markosOT, should all references to libreriscv be changed?13:59
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC15:21
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc15:21
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC16:42
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc16:59
lkclmarkos, yes17:08
lkclthere should be very few left17:08
lkclexcept for the git repository, that has to stay as-is for now17:08
*** tplaten <tplaten!~tplaten@> has joined #libre-soc17:25
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc17:59
*** josuah <josuah!~irc@> has quit IRC18:58
*** irc <irc!~irc@> has joined #libre-soc18:58
*** irc is now known as josuah18:59
*** josuah <josuah!~irc@> has quit IRC19:03
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC19:12
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc19:12
*** tplaten <tplaten!~tplaten@> has quit IRC19:13
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC19:22
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC19:23
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc19:24
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc19:24
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC19:57
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc20:37
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC20:50
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc20:51
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC22:19
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC22:42
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc22:42
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc22:45

Generated by 2.17.1 by Marius Gedminas - find it at!