No you dumbfcuk, high-level compilers don't work that way!

system error wrote:

No you dumbfcuk, high-level compilers don't work that way!

Dude, I *am* a contributor to GCC. I am fully aware how it works, stop talking shit you have *no idea* about. What the fuck is wrong with you?

Show me the code in GCC source code that handles the realignment then. I told you the exact function in GCC source code that does it, what else do you want? I don't give a shit what you say or how you imagine them to work. You have a nice imagination, but keep other people out of it.

I'd link the mirror of i386.c on github but I know you don't click on links like a true internet warrior man-child. So, other than telling you exactly what function does the realignment (and should swap the and with sub in RTL gen), I don't know what to tell you. Grow up.

Contributor (to open source software) means that you've submitted patches which were accepted. And that is a fact (and quite a few of them), so that makes me a contributor.

All I see from you is "Blah"

Internet anonymity can bring out the worst in people. User system error being a prime example. Would the self-proclaimed King of Poland dare to be as rude and condescending if Fuchs and he argued this face-to-face? Who brought system error up - where are his manners?

Knowledge and righteousness-of-argument aside, there is no excuse for a civil, well-brought up man to act as such. Absolutely disgusting.

Internet is a giant trash can, with a spoon of honey hidden deep inside. Sometimes you have weed through a lot of trash before you get to this honey, and many times this process is not worth it.

We currently live in the Golden Era of Internet Liberties - it should be not be surprising (in any society that wishes to enforce civility and prevent degeneracy) that real identities for telecommunications (along with logged chats and backdoored encryption) are mandated by the government. Yes, previous attempts at this have been made before (Google+ when it came out) and some are still in effect (Internet in China, Yarovaya Law in Russia). This can be discussed at length - whether it is good or bad, necessary or overreach by the government - I am simply stating facts and articulating trends.

On a smaller scale, I recommend that the moderators of this forum enforce civility in this particular forum thread.

Local variables? It's mentioned in the first post. I have a 256-bit AVX vector on the stack and an 4-byte int. 32+4 = 36, and to keep esp aligned (after the sub) to 32-byte the nearest is 64. Aligning esp after the subtraction, though, requires only a sub 36, because it gets aligned after anyway. If the sub 36 makes it misaligned, the bitwise and will align it. It guarantees there's at least 36 bytes of space, so [esp] will be aligned (and use the vector there).

You can't teach civility that easy to people who come from barbaric culture like Furs.

AFAIK

I'm not sure what part of "GCC generates sub esp, 64" do you not understand. Taking out the volatile int from the stack makes it "sub esp, 32" (only the vector). Must suck living in ignorance.

sub esp, 64  ; because it won't fit in 32-bytes and has to be aligned to 32 for some reason?    

sub esp, 36
and esp, -32    

No, the problem is why is and before sub. In fact, this question is broader than GCC: why is the "common idiom" to bitwise and before the subtraction?!?

Answer: No monkey. It doesn't always like that. In many cases, GCC also emits SUB before AND.

How do you fit a 32-byte vector plus a 4-byte int into 32 bytes again? What I literally said it won't fit in 32 bytes. Since it's 32+4 = 36 bytes minimum.

because it won't fit in 32-bytes

Show me the code that compiles to sub before and with GCC. "xD" (you can verify it on gcc.godbolt.org, so no, you can't lie yourself like a 6 year old out of this one)

Oh, also, so you admit that my method of sub before and is superior? Then in that case, you just shot yourself in the foot.

Quote from first post:

Furs wrote:

because it won't fit in 32-bytes