Posts by wada@plma.plus.st
(DIR) Post #AMnWiANGdP9wrtMu00 by wada@plma.plus.st
2022-08-22T20:54:28.612116Z
2 likes, 0 repeats
#introductionI'm wada, I'm sorta new to fediverse, I don't really have any opinions on things and I have trouble interacting with peopleThis place looked chill, so I'll just keep to myself and hopefully I won't cause any trouble :menherahide:
(DIR) Post #AMthTGvoEo8lIzNwMi by wada@plma.plus.st
2022-08-25T19:30:51.574017Z
4 likes, 5 repeats
(DIR) Post #AMthVe5p5wUlicCVsm by wada@plma.plus.st
2022-08-25T20:24:22.499224Z
1 likes, 1 repeats
@itzzenxx Dear itZzenXX,Thank you for your cat-based correspondence.Appreciated, wada
(DIR) Post #AMv9exoO8rMfM9EmEy by wada@plma.plus.st
2022-08-26T04:50:55.677203Z
0 likes, 0 repeats
@GNUxeava so the last statement in the gcd function gcd(divisor, remainder); has to be explicitly set to return gcd(divisor, remainder);What’s happening is that the recursive call to gcd should still return the value from the call, which will insert the instructions popq %rbp and retq.Since there’s no explicit return, clang decided to insert the ub2 instruction, which is technically true (not explicitly returning a value is undefined).Turns out gcc is just more lax about it.
(DIR) Post #AMwCUJzlokkRzb8cc4 by wada@plma.plus.st
2022-08-26T16:35:30.714856Z
1 likes, 0 repeats
@GNUxeava Looking over the assembly for the gcc version, it’s actually a bit more crazy than I initially thought.So the thing about recursive calls is that you need to still return the results even when you call it from within the function: this is due to how stack allocation works.Each new call will push a new stack frame onto the stack, and when when you reach the terminal condition, the stack unwinds and all the frames are popped off in reverse order.If there’s no explicit return, the specific frame that doesn’t have a return doesn’t return the value when it’s popped off the stack, which means that any subsequently popped frames wouldn’t have the value to return when they pop: which is why clang decided to write in the ub2 instead.The gcc version handles things weirdly, however, and I can’t tell if that’s just due to the luck of optimization or if gcc is actually really aware of what’s going on. It writes the divisor value to rax when it reaches return divisor, and as the stack unwinds there are no other values written to rax (because all calls are right before leave) so technically it does return the proper result. The clang version emits roughly the same assembly if you add return to the gcd call, but for some reason creates two different branching paths when you don’t so that it hits ub2
(DIR) Post #ANg6LVbvzaaxYJK9uy by wada@plma.plus.st
2022-09-18T04:48:38.698240Z
2 likes, 0 repeats
@neko @jojo @nekonigger30 @TheMadPirate @p @RustyCrab
(DIR) Post #ANgDt4JUgnbKPqeGMC by wada@plma.plus.st
2022-09-18T06:13:08.731367Z
2 likes, 0 repeats
@RustyCrab @jojo @nekonigger30 @TheMadPirate @p @neko @Bro-Drillard Memory management is actually not that hard if you design you system well.
(DIR) Post #AOFHLV7ZjR0TJV4fqK by wada@plma.plus.st
2022-10-05T04:06:26.484467Z
3 likes, 0 repeats
@inference @SoyMagnus @itzzenxx @Coyote I’m not buying it.A proper security model would mean properly sandboxing applications (which you can do in X11), rather than expecting the protocol to do it for you and limit what you can do overall as a result.Also, Wayland is nightmare for accessibility.
(DIR) Post #AOFIs5TeCEwrI4V6NE by wada@plma.plus.st
2022-10-05T04:23:32.241641Z
2 likes, 0 repeats
@inference @SoyMagnus @itzzenxx @Coyote The protocol shouldn't have to require you to come up with an entire security module system in order to make accessibility functional.
(DIR) Post #AOFKfNCtANP0QDj9DE by wada@plma.plus.st
2022-10-05T04:43:38.927786Z
2 likes, 0 repeats
@SoyMagnus @itzzenxx @inference @Coyote I’m actually mostly indifferent towards Wayland or X11 (aside from issues with accessibility), but I get especially bothered when people say things like ‘don’t use __ on a secure or private system’ because %99 of the time they are used on a secure/private system, just that their insecurity is mitigated by a whole confluence of factors (isolation, sandboxing, OPSEC, ‘burnability’, etc.).If you want to take security more seriously, just learn to be more secure???
(DIR) Post #AOoaMlpfZOQSviAjSq by wada@plma.plus.st
2022-10-22T04:54:02.789833Z
2 likes, 0 repeats
@bot @thebitchisback @sun_eater @nenetl >the cells don't identify as male because the cells are femaleyou know for someone who keeps harping on people being 'biologically' of some sort of gender you really missed the fucking point on this one
(DIR) Post #AOoaYFaNHvFWOSUZTU by wada@plma.plus.st
2022-10-22T04:56:07.350686Z
2 likes, 0 repeats
@bot @thebitchisback @sun_eater @nenetl >the cells are female>no Y chromosomehow fucking stupid are you?
(DIR) Post #AOoauv9NQt4ffsfray by wada@plma.plus.st
2022-10-22T05:00:13.042825Z
2 likes, 0 repeats
@bot @thebitchisback @sun_eater @nenetl i never once identified as trans or specified any gender, cis or notyou actually have no idea what i am