hannes changed the topic of #mirage to: MirageOS are OCaml unikernels - https://mirage.io - this channel is logged at http://irclog.whitequark.org/mirage/ - MirageOS 3.7.1 is released - happy hacking!
Haudegen has quit [Ping timeout: 250 seconds]
_whitelogger has joined #mirage
_whitelogger has joined #mirage
busu has joined #mirage
boke has quit [Ping timeout: 260 seconds]
_whitelogger has joined #mirage
_whitelogger has joined #mirage
Haudegen has joined #mirage
icristescu[m] has joined #mirage
vds[m] has joined #mirage
talex5 has joined #mirage
<dinosaure> hannes: it's possible to have something close to the old X509.Authenticator.null?
<hannes> dinosaure: well, it won't be in the API. you can write your own: "let null ~host:_ _ = Ok None" (see the conduit changes)
<dinosaure> ok, thx :)
talex5 has quit [Remote host closed the connection]
talex5 has joined #mirage
<talex5> This time error on PVH is quite strange. I edited mirage-skeleton's hello main.ml to call unix_gettimeofday directly and print the result. It always gets zero.
<talex5> If I return an int (instead of a float) then it's OK.
<talex5> Also, if I raise an exception then it comes out as `Failure ("")`.
<talex5> And if I use `printk` in the mirage-xen C stub, then nothing happens. But the `printk` in the mini-os function it calls does work.
<talex5> Ah, looks like the C flags that mini-os puts in its .pc file need updating.
cheers has quit [Ping timeout: 272 seconds]
Guest31018 has joined #mirage
<talex5> Also, division doesn't seem to work.
<mato[m]> talex5: If you're seeing FPU-related errors, it may be that Mini-OS is not init'ing the FPU and SSE units.
<mato[m]> talex5: Check if it's doing the equivalent of https://github.com/Solo5/solo5/blob/master/bindings/virtio/boot.S#L170 somewhere.
<mato[m]> talex5: If not, then you want to do that, can be done in the low-level mirage-xen startup (no need to patch Mini-OS) before doing anything substantial. Note that w/o special CFLAGS C compilers will happily use SSE instructions so technically you want to do this before jumping into "normal" C code.
<hannes> in arch/x86/setup.c there's arch_init which AFAICT is called (and which calls out to sse_init() in the same file)
<hannes> talex5: it may be very well that the CFLAGS in the pc files need to be updated to enable printk
<hannes> arch_init is called by x86_64.S
<hannes> and there's a pending patch from talex on the old mini-os to add floating point printing to mini-os, which i did not merge (too many conflicts), i.e. the printf with float (%f etc.) will not work (not sure whether this affects the time issue above)
<tg> moin
def` has joined #mirage
<talex5> The %f issue makes debugging a little harder, but I got around it by calling truncate first.
<hannes> talex5: feel free to include that patch, and push to the minios branch directly :)
<talex5> -fno-builtin seems to be the important flag.
<samoht_[m]> hannes: for nocrypto/mirage-nocrypto linking errors, should we try to detect this at `mirage configure` time? The failure mode seems a bit bad
Haudegen has quit [Quit: Bin weg.]
<hannes> samoht_[m]: as i wrote, to me the best solution would be to conflict in mirage-entropy* to mirage-crypto-entropy
<hannes> samoht_[m]: i don't think embedding magic in the mirage tool is the right way to go
<samoht_[m]> well the packages are co-installable, the issue appears if you try to link them together. So `conflict` will not work here.
<samoht_[m]> (and it will cause a pain to make the transition, as you will have to upgrade all the packages at once)
<samoht_[m]> so what we want is some kind of conflicts at the mirage device layer. Let me try something...
Armael has joined #mirage
<Armael> yo
<yomimono> hello!
<mato> hi mindy!
<yomimono> hi mato!
<h01ger> yo yomimono!
<yomimono> ho h01ger!
<yomimono> if you see a bunch of nonsense from me, assume cat interference and infer my apologies :D
<h01ger> :))
* h01ger might get a cat visitor these days. unclear. collateral win from the crisis if it happens :)
<samoht_[m]> Hey Mindy!
<Armael> yomimono: with all the lovely cat pictures on mastodon, I wanted to ask
<Armael> how does the "foster cat" thing work?
<Armael> I must say that having nice cats visit your home does sound great
<yomimono> armael: there's a local shelter that gives us cats to hold on to for a while until they get adopted
<yomimono> (also, salut samoht!)
<yomimono> sometimes they give us cats that aren't well socialized, and they stay with us for a while to learn that humans are okay
<Armael> oh, very nice
<Armael> that sounds like a cool plan
<yomimono> yeah, it's really great :) if you're going out of town or whatever you can just say "hey, I can't take any cats for a bit"
<dinosaure> hi yomimono :)
<Armael> I love cats but until now I've refrained from getting one, since I didn't like the idea of it having to live in a small apartment
<yomimono> hi dinosaure! :)
<yomimono> the apartment thing is super real, but that's another nice part of fostering - even a small apartment is nicer than a cage :(
<Armael> (I grew up in the countryside where the cat had a lot of space where to run / hunt rabbits etc)
<Armael> yep
<Armael> then I'll try to see if there's a local shelter where I live
<yomimono> :DDDDD
<yomimono> post photos if you find one!!
<Armael> I will definitely do!
<yomimono> h01ger: you are also on the hook for cat photos ;)
<h01ger> yes, but i try to avoid regular social media visits, even for cat pics
<yomimono> you should probably just photobomb the marrakesh list!
<yomimono> everyone is interested in cat photos, I'm pretty sure
<h01ger> ah, sending, not receiving
<samoht_[m]> definitely :-)
<h01ger> that i guess i can do :)
<talex5> Hi yomimono!
<yomimono> hi talex5!
<yomimono> you're doing some valiant deep-diving on this time stuff
<talex5> Yeah, mshinwell is debugging over on ocamllabs slack right now...
<reynir> Armael: my friend mentioned that she wanted to volunteer at a local shelter. I can ask her if they do fostering if you like :)
<Armael> ohh
<Armael> yes, that'd be awesome!
Guest31018 has quit [Quit: Guest31018]
cheers has joined #mirage
<reynir> Are you still travelling, or did you manage to get back?
<Armael> I'm at the airport right now
<Armael> in paris
<Armael> I have a flight to billund tonight
<Armael> can't wait to get home tbh…
<hannes> Armael: is gabriel back in paris?
<Armael> yes
<hannes> good, i understood you both are stuck in gibraltar
<Armael> we both came back to paris on Friday
<Armael> yeah, the three of us including Steffi
<Armael> we got stuck Friday morning, but we managed to grab direct flights to Paris/Berlin later that day
<Armael> and now I'm fleeing from the complete lockdown that might or might not be happening tomorrow in France
<hannes> ok, good
<hannes> marmarek: around? would you happen to know whether the virtual cpu in pv and pvh mode needs to be setup differently (i.e. fpu / sse)?
<marmarek> hannes: idk
<hannes> samoht_[m]: the linking error occurs if you link mirage-entropy and mirage-crypto-entropy into the same binary. this is not nice. but: the only thing emitting a dependency to either of these packages is mirage itself (<3.7.5 mirage-entropy, >= 3.7.5 mirage-crypto-entropy). the only unikernel i'm aware of which needs mirage{-crypto}-entropy is prng (since it accesses its API to print the entropy sources -->
<hannes> i'm really not sure whether we need code in mirage to check for this...
<haesbaert> moin moin
Haudegen has joined #mirage
engil has quit [*.net *.split]
hark has quit [*.net *.split]
smondet has quit [*.net *.split]
derpeter2 has quit [*.net *.split]
smondet has joined #mirage
hark has joined #mirage
derpeter2 has joined #mirage
engil has joined #mirage
engil has quit [Changing host]
engil has joined #mirage
mahmudov has quit [Ping timeout: 260 seconds]
talex5 has quit [Quit: Saliendo]
jnavila has joined #mirage
jnavila has quit [Read error: No route to host]
jnavila has joined #mirage
mahmudov has joined #mirage
adhux0x0f0x3f has joined #mirage
jnavila has quit [Ping timeout: 256 seconds]
adhux0x0f0x3f-- has joined #mirage
adhux0x0f0x3f has quit [Ping timeout: 240 seconds]