https://sourceware.org/bugzilla/show_bug.cgi?id=29592
--- Comment #8 from matoro <matoro_bugzilla_glibc at matoro dot tk> --- (In reply to Nick Clifton from comment #7) > (In reply to matoro from comment #6) > Hi Matoro, > > > > Then I compared the scripts with the linker invocations that I got out of > > pgrep and saw that they were different. The linker is getting invoked > > differently depending on whether the execstack warning is enabled or not. > > OK - so does that mean that the bug is inside whatever is constructing the > linker > command line ? > > You said that Gentoo is adding --enable-warn-execstack. How does it do this > ? Could whatever method it is using also be the source of the > mis-constructed linker command line ? > > > > With --enable-warn-execstack: > > I might be missing something here, but I do not actually see > --enable-warn-execstack on the command line that you listed. > > > -melf64btsmip -pie -o go-sqlite3.test -s > > It looks to me like the -pie option is also added here... > > Cheers > Nick --enable-warn-execstack is the flag you yourself added to default-enable this feature at binutils build-time in ba951afb99912da01a6e8434126b8fac7aa75107. To test this, I am toggling this flag on/off and rebuilding binutils. When it's toggled on, it always enables the warning, and -pie gets added to link invocations from ld. When toggled off, it defers to the target-specific rules, which you already used to disable the warning on mips, and -pie does NOT get added to link invocations. As for what actually invokes the linker, I believe it is the usual gcc -> collect2 -> ld chain. I will check and see if these command lines are different also. -- You are receiving this mail because: You are on the CC list for the bug.
