Mold 2.3 Linker Allows Removing Superfluous ENDBR64 Instructions

[phoronix]

Phoronix: Mold 2.3 Linker Allows Removing Superfluous ENDBR64 Instructions

Mold 2.3 was released today by Rui Ueyama as the newest version of this high-speed linker alternative to LLVM LLD and GNU Gold...

https://www.phoronix.com/news/Mold-2.3-Released

[Sethox]

The day Mold can compile Wine is the day I will try out the Open source Windows (compiling it myself).

[dlq84]

The day Mold can compile Wine is the day I will try out the Open source Windows (compiling it myself).

You'll have to wait a while, given that it's a linker and not a compiler

[Quackdoc]

one unfortunate thing about mold is that it seems to break sccache for me when compile rust projects which is kinda sad

[archkde]

one unfortunate thing about mold is that it seems to break sccache for me when compile rust projects which is kinda sad

That seems quite unlikely. Sccache only caches the actual compilation, and mold is not involved in that as a linker.

[Setif]

The day Mold can compile Wine is the day I will try out the Open source Windows (compiling it myself).

https://github.com/rui314/mold/issues/190#issuecomment-1401485928

https://github.com/rui314/mold/issues/1069#issuecomment-1653436823

[Quackdoc]

That seems quite unlikely. Sccache only caches the actual compilation, and mold is not involved in that as a linker.

unlikely or not it's something I was able to replicate fairly easily

EDIT: forgot to mention im seemlingly not the only person with this issue https://github.com/mozilla/sccache/issues/1755

[archkde]

unlikely or not it's something I was able to replicate fairly easily

EDIT: forgot to mention im seemlingly not the only person with this issue https://github.com/mozilla/sccache/issues/1755

It seems that somehow cache misses occured when they should not. A good next step would be to figure out what inputs sccache considered changed.

[Quackdoc]

It seems that somehow cache misses occured when they should not. A good next step would be to figure out what inputs sccache considered changed.

yeah I plan on looking at it... one day, at least

[justinkb]

I don't think it's the job of the linker to be fucking with the opcodes produced by the compiler (other than massaging location opcodes), that seems totally out of scope