<trungnt2910[m]>
<waddlesplash> "likely it won't and this is..." <- Now it gives me a page fault.
<Begasus>
eeps trungnt2910[m] "HaikuOS" ;)
<trungnt2910[m]>
Begasus: Why not?
<Begasus>
it's "Haiku" ;)
<Begasus>
Some are pretty strict on the name use :)
B2IA has quit [Ping timeout: 480 seconds]
illwieckz has quit [Remote host closed the connection]
<jessicah>
haha, yeah, there's no OS in the name
<jessicah>
the website is haiku-os, because we couldn't get haiku
<Begasus>
heading out, dogschool
<Begasus>
bbl
<jessicah>
trungnt2910[m]: it seems like a better option might be having an explicit flag for not readable/writable, so the fix protection flags thing can avoid setting flags that shouldn't be set instead of inferring it
<jessicah>
or... that we find cases where 0 is passed when it should really have a read flag, and change those
illwieckz has joined #haiku
jmairboeck has joined #haiku
illwieckz has quit [Ping timeout: 480 seconds]
<trungnt2910[m]>
<jessicah> "or... that we find cases where 0..." <- You mean cases that are scattered around the OS kernel?
<trungnt2910[m]>
s/around/across/
<trungnt2910[m]>
<jessicah> "trungnt2910: it seems like a..." <- So calls to `mmap` and `mprotect` from userland would set those flags (0 protection for user, explicitly not readable/writable for kernel) and `fix_protection` would catch these flags and stop making them accessible to the kernel, right?
<trungnt2910[m]>
And this should also prevent messing up with kernel allocated memory like the ones used by drivers during boot?
<jessicah>
possibly, it is an idea, I don't know how feasible that would be
illwieckz has joined #haiku
illwieckz has quit [Remote host closed the connection]
illwieckz has joined #haiku
<trungnt2910[m]>
In Haiku's kernel `_kern_`something is a syscall while `_user_`something is the call's corresponding implementation in the kernel, right?
<jessicah>
right
AlaskanEmily has quit [Remote host closed the connection]
<trungnt2910[m]>
Might this patch have any consequences I don't know of (I'm just removing the kernel read/write flags for calls from userland).
MajorBiscuit has joined #haiku
vdamewood_ has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<jessicah>
trungnt2910[m]: oh nice, are there still calls to the non-user function?
<trungnt2910[m]>
jessicah: Yes, everything that is not a syscall handler.
<trungnt2910[m]>
(Confirmed in my IDE)
<jessicah>
cool
<jessicah>
I'm sure at the very least, waddlesplash will like to review
doomquak1keen has joined #haiku
doomquakekeen has quit [Ping timeout: 480 seconds]
MajorBiscuit has quit [Ping timeout: 480 seconds]
binky has joined #haiku
MajorBiscuit has joined #haiku
<humdinger>
hello binky. how are you doing?
<binky>
ho, ho, humdinger. I'm trying out the Konversation IRC client.
<humdinger>
ah. I bet you need to take a screenshot to upload to HaikuDepotServer...
<binky>
Exactly right. You can read my...
<binky>
...mind.
<binky>
done. bye bye
* binky
waves
* humdinger
waves back
binky has quit []
<humdinger>
wait up, binky
humdinger has quit [Quit: Vision[]: Oi with the poodles already!!]
MajorBiscuit has quit [Ping timeout: 480 seconds]
<jessicah>
trungnt2910[m]: does the rpath patch embed the actual path, or does it embed the variable "$ORIGIN"?
<jessicah>
also, LD_LIBRARY_PATH is simply LIBRARY_PATH on Haiku
<jessicah>
so anywhere that uses LD_LIBRARY_PATH needs patching for Haiku
<trungnt2910[m]>
<jessicah> "trungnt2910: does the rpath..." <- It just embeds the variable $ORIGIN. However, it seems that Haiku does support that variable (And I actually tested that, the test printed a different error after the patch).
<trungnt2910[m]>
(That error is, again, related to exception catching)
<trungnt2910[m]>
<jessicah> "so anywhere that uses LD_LIBRARY..." <- I'll keep that in mind, will fix when it actually surfaces.
<jessicah>
huh, so it does
<andreasdr[m]>
Good moorning.
mmu_man has joined #haiku
tqh has quit [Quit: Leaving]
pch has quit []
pch has joined #haiku
<Begasus>
augiedoggie, awake?
Gaspare has joined #haiku
vdamewood has joined #haiku
vdamewood_ has joined #haiku
vdamewood has quit [Ping timeout: 480 seconds]
Gaspare has quit [Quit: Gaspare]
Gaspare has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
<augiedoggie>
Begasus: barely
<Begasus>
;)
<Begasus>
maybe you could merge the PR for gcompris?
<Not-d5e4>
[haikuports/haikuports] augiedoggie 38040a8 - gcompris: remove testing flag from secondary arch
pch has quit [Remote host closed the connection]
pch has joined #haiku
<trungnt2910[m]>
<waddlesplash> "anyway I'll investigate more..." <- Glad to hear that. But if the impact is too far and wide and takes time to fix might you consider this patch as a _temporary_ fix for userland applications?
<trungnt2910[m]>
> <@_oftc_waddlesplash:matrix.org> anyway I'll investigate more this coming week. thanks for the work so far
<trungnt2910[m]>
* Glad to hear that. But if the impact of the change is too far and wide and takes time to adjust might you consider this patch as a _temporary_ fix for userland applications?
<waddlesplash>
maybe but probably not because I suspect this makes things inconsistent between userland and kernel APIs
<waddlesplash>
the consistency matters
<waddlesplash>
i.e. create_area should behave identically whether called in userland or in the kernel, and while it probably doesn't already, we shouldn't make it worse.
<waddlesplash>
anyway I don't expect the real change to take much longer and in fact you can likely start on it yourself if you want
<waddlesplash>
just add a panic() to the original case where everything is 0 and then boot, and see what it trips on and fix it
<waddlesplash>
e.g. the SCSI stack it appears
Gaspare has joined #haiku
pch is now known as kinkinkijkin
x10z has joined #haiku
<trungnt2910[m]>
<waddlesplash> "just add a panic() to the..." <- Ok I'll try now.
<waddlesplash>
we should likely check to see when this code was written and if any of it ever ran on BeOS
<waddlesplash>
otherwise this is very probably the better solution by far
B2IA has joined #haiku
austin987 has quit [Remote host closed the connection]
austin987 has joined #haiku
Vitto has quit [Remote host closed the connection]
x10z has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
Vitto has joined #haiku
bbjimmy has joined #haiku
extrowerk has quit [Quit: Bye!]
extrowerk has joined #haiku
aeryndunham has joined #haiku
<PulkoMandy>
B_CONTIGUOUS makes sure that the area is a single chunk of physical memory, meaning it can be used easily for DMA transfers
<trungnt2910[m]>
<waddlesplash> "we should likely check to see..." <- So if that code was written a long time ago, it'd be safe to modify it to like this?
<waddlesplash>
no, it's safe either way
<waddlesplash>
the point is to check whether we are changing behavior from BeOS here, or whether this is a "Haiku-ism"
<waddlesplash>
anyway, it is the weekend, so I'll still be in and out, and I can focus more on this tomorrow
<trungnt2910[m]>
Most of the lines I checked are like 10 years ago.
<trungnt2910[m]>
* checked are last changed like 10, * like 10 or more years ago.
<trungnt2910[m]>
* Most of the lines I checked were last changed like 10 or more years ago.
Vitto has quit [Remote host closed the connection]
Vitto has joined #haiku
<trungnt2910[m]>
<waddlesplash> "otherwise this is very probably..." <- If that's the case, I've uploaded my changes to the review server.
<trungnt2910[m]>
Hope you can see it tomorrow.
elias has joined #haiku
<elias>
does haiku support dual monitor setups?
<rennj>
matrox g400,450 perhaps, beos had driver for...
<wannabee[m]>
humdinger: A port exists or Haiku uses it is 2 different thing.
linearcannon has joined #haiku
<humdinger>
indeed, I misread... sorry
jezek2_ has joined #haiku
jezek2 has quit [Ping timeout: 480 seconds]
<andreasdr[m]>
Wanted to know If Haiku OS uses internally Harfbuzz to do the text layouting. Like if it is used for what I usually see on Screen. Tracker and Deskbar :D
<humdinger>
I don't really know, but when trying to uninstall harfbuzz, it'd uninstall most of the system's packages. So I guess it is used...
<humdinger>
possibly by freetype(?)
<andreasdr[m]>
Ah cool! Thank you!
<humdinger>
though probably not all features, given that Haiku doesn't really support right-to-left languages AFAIK.
<andreasdr[m]>
Oh. I would have thought that exactly HarfBuzz implements those things.
<humdinger>
I suppose it would still need to be wired in in all places...
<andreasdr[m]>
Hmm. Very good question. From an API point of view as developer I would be surprised a bit.
<humdinger>
take a bucket of salt with that. I may be completely wrong...
<humdinger>
wait for the devs to pipe up :)
Atomozero has quit [Ping timeout: 480 seconds]
<Begasus>
g'night peeps
Begasus has quit [Quit: Leaving]
<andreasdr[m]>
Good night Begasus
tuaris has joined #haiku
humdinger has quit [Quit: Vision[]: Oi with the poodles already!!]
aeryndunham has quit [Ping timeout: 480 seconds]
DKnoto has quit [Ping timeout: 480 seconds]
DKnoto has joined #haiku
ClaudioM has joined #haiku
m199 has joined #haiku
diver1 has joined #haiku
Diver is now known as Guest1201
diver1 is now known as Diver
Guest1201 has quit [Ping timeout: 480 seconds]
macdude22 has joined #haiku
jason123santa has left #haiku [Disconnected: Replaced by new connection]
jason123santa has joined #haiku
jmairboeck has quit [Quit: Konversation terminated!]