On Sun, 21 Apr 2024 at 06:42, Izumi Tsutsui <tsut...@ceres.dti.ne.jp> wrote: > > > cat work.mbr.truncated work.efi imgroot.fs work.gpt > work.img > > /dumps/sysbuild/amd64/tools/bin/nbgpt work.img biosboot -i 2 > > -c > > /dumps/sysbuild/amd64/obj/home/sysbuild/src/distrib/amd64/liveimage/emuimage/work/usr/mdec/gptmbr.bin > > nbgpt: work.img: No primary GPT header; run create or recover > > If you are trying an update build (with -u) and your previous build > was using source tree files before UEFI changes against liveimage, > https://mail-index.netbsd.org/source-changes/2024/04/13/msg150802.html > a stale work.mbr (${WORKMBR}) image doesn't contain GPT (but MBR) > so ${TOOL_GPT} on the update build may fail.
That explains it - I do use update build, but on occasion, when something fails, I tend to remove the destdir and|or obj. In this case I started afresh a little earlier than the abovementioned commit. BTW my next question was to be 'Why live image is only MBR, as the install image is GPT?', but I see I am late for it... > > The similar issue (against the secondary GPT headers) occured > if IMAGE size is changed: > https://gnats.netbsd.org/56132 > > Maybe we should note it (remove work.mbr on updating builds) > in doc/UPDATING? > > --- > Izumi Tsutsui Thanks, Chavdar -- ----
