https://sourceware.org/bugzilla/show_bug.cgi?id=33349
--- Comment #2 from Rainer Orth <ro at CeBiTec dot Uni-Bielefeld.DE> --- > --- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> --- > Created attachment 16307 > --> https://sourceware.org/bugzilla/attachment.cgi?id=16307&action=edit > A patch > > Please try this. I've given it a try on amd64-pc-solaris2.11, i386-pc-solaris2.11, and amd64-pc-freebsd14.0: there are quite a number of failures everywhere: * Solaris/amd64: +FAIL: Build libplt-1a.so +FAIL: Build libplt-1b.so +FAIL: Build libplt-1a.so +FAIL: Build libplt-1a-x32.so +FAIL: Build libplt-1b.so +FAIL: Build libplt-1b-x32.so +FAIL: Build libplt-2a.so +FAIL: Build libplt-2a-x32.so +FAIL: Build libplt-2b.so +FAIL: Build libplt-2b-x32.so * Solaris/i386: +FAIL: Build libplt-1a.so +FAIL: Build libplt-1b.so * FreeBSD/amd64: FAIL: Build libplt-1a-x32.so FAIL: Build libplt-1b-x32.so FAIL: Build libplt-2a-x32.so FAIL: Build libplt-2b-x32.so Without having even started to analyze them, this looks to me that those tests have all/most of the issues with your GOT patch reported in PR ld/33350, which has a partial patch for those. I suspect much of this is applicable here, too. At least the x32 failures certainly are: the x32 exists nowhere but on Linux. -- You are receiving this mail because: You are on the CC list for the bug.
