OK to trunk, thanks for fixing this issue, I hit that issue before but I didn't figure out what happened...since that issue will disappear when I clean build :p
On Tue, Feb 1, 2022 at 5:52 AM Maciej W. Rozycki <ma...@embecosm.com> wrote: > > Make riscv-sr.o depend on target machine headers, removing spurious test > failures: > > FAIL: gcc.target/riscv/save-restore-3.c scan-assembler-not call[ > \t]*t0,__riscv_save_0 > FAIL: gcc.target/riscv/save-restore-3.c scan-assembler-not tail[ > \t]*__riscv_restore_0 > FAIL: gcc.target/riscv/save-restore-3.c scan-assembler tail[ \t]*foo > FAIL: gcc.target/riscv/save-restore-6.c scan-assembler-not call[ > \t]*t0,__riscv_save_0 > FAIL: gcc.target/riscv/save-restore-6.c scan-assembler-not tail[ > \t]*__riscv_restore_0 > FAIL: gcc.target/riscv/save-restore-6.c scan-assembler tail[ \t]*other_func > > if the definitions of UNSPECs are locally changed and GCC rebuilt from a > dirty tree. > > gcc/ > * config/riscv/t-riscv (riscv-sr.o): Add $(TM_H) dependency. > --- > Hi, > > Noticed while fiddling with `fmin'/`fmax' further. As not a code change > OK to apply despite stage 4? Might be worth backporting too. > > Maciej > --- > gcc/config/riscv/t-riscv | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > gcc-riscv-sr-dep.diff > Index: gcc/gcc/config/riscv/t-riscv > =================================================================== > --- gcc.orig/gcc/config/riscv/t-riscv > +++ gcc/gcc/config/riscv/t-riscv > @@ -6,7 +6,7 @@ riscv-builtins.o: $(srcdir)/config/riscv > $(srcdir)/config/riscv/riscv-builtins.cc > > riscv-sr.o: $(srcdir)/config/riscv/riscv-sr.cc $(CONFIG_H) \ > - $(SYSTEM_H) > + $(SYSTEM_H) $(TM_H) > $(COMPILER) -c $(ALL_COMPILERFLAGS) $(ALL_CPPFLAGS) $(INCLUDES) \ > $(srcdir)/config/riscv/riscv-sr.cc >
