flat assembler
Message board for the users of flat assembler.
Index
> Non-x86 architectures > [fasmg] RISC-V support? Worth to port fasmg to RISC-V? |
Author |
|
sylware 29 Jul 2021, 16:25
It seems RISC-V is gaining serious momentum.
RISC-V cross-assembly includes for fasmg? Is fasmg, in its current state, worth a RISC-V (64bits) port? |
|||
29 Jul 2021, 16:25 |
|
sylware 29 Jul 2021, 22:34
you read well: a risc-v assembly implementation of fasmg macro language. It means the macro language would be rigorously defined in order to allow implementation in other assembly languages.
|
|||
29 Jul 2021, 22:34 |
|
revolution 29 Jul 2021, 23:04
Are you starting a project to rewrite fasmg?
|
|||
29 Jul 2021, 23:04 |
|
sylware 31 Jul 2021, 11:52
That would not be a rewrite, more likely a "port" of fasmg macro language.
But, as I said, the macro language "specs" would have to be extremely rigorously defined and relatively stable (it is already assembly friendly by definition, this is why fasmg is "special"). I am personally working on other projects already and don't have brain and energy time for this unfortunately (like when I wanted to add x86_64 ELF64 TLS support to fasmg... and realize how nasty TLS is... and actually dropping it). This thread is more to bring attention to this matter than anything else. But once we can buy risc-v mini computers at affordable prices on amazon (which works with noscript/basic (x)html browsers), things may "explode". |
|||
31 Jul 2021, 11:52 |
|
bitRAKE 31 Jul 2021, 16:17
RISCV support is already in UEFI. I haven't really looked at the ISA, tbh. But I remember hearing that it was very modular and the spec is defined in a computer readable form. What would be awesome is a tool that translates the spec into fasmg macros - to capture future extensions or specializations.
|
|||
31 Jul 2021, 16:17 |
|
sylware 01 Aug 2021, 13:54
until this tool is written itself in assembly, it should be fine but... there is not enough variance in time of risc-v specs to justify such a generator. Actually, most, if not all, similar generators (even in other projects) end up kludging SDKs more than anything else.
|
|||
01 Aug 2021, 13:54 |
|
bitRAKE 02 Aug 2021, 03:43
Yeah, I can see your point. I was just hoping RISCV might be different - I know they were debugging simulations before there was hardware and I imagined such a system might really be exact. Might be totally wrong though - maybe it was a patchwork of Band-Aids and bubblegum. Maybe that research isn't public? Such a system might be able to get 80-90% of the way and then require auditing.
|
|||
02 Aug 2021, 03:43 |
|
< Last Thread | Next Thread > |
Forum Rules:
|
Copyright © 1999-2025, Tomasz Grysztar. Also on GitHub, YouTube.
Website powered by rwasa.