https://bugs.kde.org/show_bug.cgi?id=468575
--- Comment #62 from Paul Floyd <pjfl...@wanadoo.fr> --- (In reply to Mark Wielaard from comment #60) > (In reply to Mark Wielaard from comment #55) > > nestedfns fails with: > > > > +Process terminating with default action of signal 11 (SIGSEGV) > > + Bad permissions for mapped region at address 0x........ > > + at 0x........: ??? > > + by 0x........: call_func (nestedfns.c:14) > > + by 0x........: test1 (nestedfns.c:23) > > + by 0x........: main (nestedfns.c:37) > > + > > > > Not sure this simply means (noexecstack) nested functions simply don't work > > on riscv? > > This isn't riscv specific. It also fails on other arches. > Seems to be caused by: > > commit c5552fe28b8d89ed5a92cc736fa00d7e336a3f2c > Author: Paul Floyd <pjfl...@wanadoo.fr> > Date: Fri Dec 6 21:24:50 2024 +0100 > > regtest: fix a warning building none/tests/nestedfns on some platforms > > diff --git a/none/tests/Makefile.am b/none/tests/Makefile.am > index 53a6e1f6bc95..fa0128d3a8e4 100644 > --- a/none/tests/Makefile.am > +++ b/none/tests/Makefile.am > @@ -335,6 +335,7 @@ endif > > if HAVE_NESTED_FUNCTIONS > check_PROGRAMS += nestedfns > + nestedfns_LDFLAGS = -Wl,-z,noexecstack > endif > > # This doesn't appear to be compilable on Darwin. > > So this sets noexecstack for the nestedfns exec. > But as the comment at the top of nestedfns says: > > /* This is a test program from Lee Kindness which used to fail on V > because gcc implements the nested function mumbo jumbo using self > modifying code on the stack, at least on x86 and amd64. It now > works transparently because by default V now generates > self-checking translations for translations taken from stack-like > segments. > */ > > And indeed even on amd64-linux this program now crashes even when not run > under valgrind: > > Program received signal SIGSEGV, Segmentation fault. > 0x00007fffffffd9d0 in ?? () > (gdb) where > #0 0x00007fffffffd9d0 in ?? () > #1 0x0000000000401138 in call_func (sel=0x7fffffffd9d0) at nestedfns.c:14 > #2 0x0000000000401196 in test1 () at nestedfns.c:23 > #3 0x000000000040120b in main (argc=1, argv=0x7fffffffdb48) at > nestedfns.c:37 > > And that is indeed the sel function stub on the stack... > > Paul, what was the warning? And can we just accept that warning? > Reverting this seems to make the test work again as expected. On arm raspberrypios /usr/bin/ld: warning: nestedfns-nestedfns.o: requires executable stack (because the .note.GNU-stack section is executable) but the exe runs OK No problem on FreeBSD because this is a non-standard GCC extension that clang doesn't accept. I've pushed a fix that turns off the warning rather than turning off the exec stack. -- You are receiving this mail because: You are watching all bug changes.
