Closed jjakob closed 1 year ago
Tried it on PureOS byzantium, could not reproduce the error. There must be some difference in toolchain configuration, environment or versions.
root@librem:~# /usr/bin/mipsel-linux-gnu-gcc --version
mipsel-linux-gnu-gcc (Debian 10.2.1-6) 10.2.1 20210110
It looks like newer versions of gcc are detecting code that looks odd. Certainly that code does look odd. What it is doing is quite sensible, but it isn't try to determine something about the memory hardware, which is not the sort of thing the compiler would normally expect to see. It is looking to see if memory appear to "repeat" after some power-of-2, which would indicate what the usable memory size is. We'll probably need to eventually come up with a work-around to stop the compiler from complaining, but if you can just used a different compiler version, that is easiest for now.
any news on that? Trying to make a yocto layer for gnubee and I would really like not to fallback to older version of gcc.
It looks like there is mt7621.c in 5.18 kernel, is there a way to backport it?
The offending code was rewritten in Commit: cc19db8b312a ("MIPS: ralink: mt7621: do memory detection on KSEG1") in Linux 5.17. That patch applies cleanly on 5.15 and compiles. I haven't tested if it runs. I've updated my v5.15 branch to be based on 5.15.22 and added this patch. So you could try building from there, or just add the patch yourself and try that.
This issue seems to be fixed, so I'm closing.
Hi, I'm trying to compile gbpc2-5.15 on Gentoo, I followed the steps:
I'm getting the following error:
Info on environment: