I regularly run VMs at 256MB RAM and relinking completes without issue
(it will swap as needed). So I don't think you are running out of RAM.
A frequent cause of this LLVM error is reported to be low disk space.
Regards
Lloyd
Stuart Henderson wrote:
> On 2026/05/20 12:34, Terry Cocksworth wrote:
> > I run a few low-spec VMs in VMD which as of 7.9 are seeing occasional
> > crashes
> > during kernel relinking. They have quite little RAM, only 768 MiB, but it
> > has
> > worked well for many years, 7.8 included. Relinking has never even touched
> > swap memory, and with 7.9 it still doesn't when relinking succeeds.
> >
> > I slightly upped the VMs' memory allotment to 1024 MiB to see if a memory
> > shortage is the cause after all, but the crashes still happen.
> >
> > The VMs are 7.9/amd64 and here's the log output from reorder_kernel:
> >
> > === /usr/share/relink/kernel/GENERIC/relink.log ===
> >
> > (SHA256) /bsd: OK
> > LD="ld" sh makegap.sh 0xcccccccc gapdummy.o
> > ld -T ld.script -X --warn-common -nopie --no-mmap-output-file -o newbsd
> > ${SYSTEM_HEAD} vers.o ${OBJS}
> > LLVM ERROR: IO failure on output stream: Bad address
> > *** Error 1 in /usr/share/relink/kernel/GENERIC (Makefile:2477 'newbsd':
> > @echo ld -T ld.script -X --warn-common -nopie --no-mmap-output-file...)
> >
> >
>
> IIUC that is how an out-of-memory would often appear.
>
> Note that multiple runs of a program will result in slightly different
> memory use, run under '/usr/bin/time -l' and look at max rss. Not by
> _loads_ but if you are up against the line it could be enough to
> work/fail.
>
> Also bear in mind that the run done at boot is in the background after
> various daemons have started, so you can have different workloads
> (and thus free memory) on the machine in different situations.
>
>
