https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121757
--- Comment #12 from Richard Biener <rguenth at gcc dot gnu.org> --- (In reply to Richard Biener from comment #11) > Created attachment 62315 [details] > testcase > > This is the a.ltrans0.o object. Reproduce with > > ~/obj-arm-g/gcc> ./lto1 -quiet -dumpbase ./a.ltrans0.ltrans -mfloat-abi=hard > -mtls-dialect=gnu -mcpu=arm10e -mfloat-abi=hard -mtls-dialect=gnu -marm > -mlibarch=armv5te+fp -march=armv5te+fp -O2 -O2 -version -fno-openmp > -fno-openacc -fno-pie -fcf-protection=none -fltrans a.ltrans0.o -o > ./a.ltrans0.ltrans.s > > the cross was configured with --target=arm-suse-linux-gnueabi > --enable-valgrind-annotations --with-sysroot=/space/rguenther/armv7 > --with-float=hard > > but I guess any arm cross will do (that "sysroot" is simply sth like a > debboostrap tarball for armv7, in my case it's a JEOS image for openSUSE > Tumbleweed - but with the testcase you don't need one). I'll note the LTO IR was created with a compiler from trunk r16-3589-g86353186dc2e47 (unless there are LTO streaming changes the exact rev. shouldn't matter, but it won't work with GCC 15).
