Date: Sun, 18 Dec 2016 22:28:26 +0000 From: bugzilla-noreply@freebsd.org To: freebsd-bugs@FreeBSD.org Subject: [Bug 215395] ctfconvert from -r310152: gcc 6.2.0 based powerpc64-gcc output sometimes gets new dwarf error/abort from ctfconvert: die <SOME NUMBER>: failed to get ref Message-ID: <bug-215395-8@https.bugs.freebsd.org/bugzilla/>
next in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D215395 Bug ID: 215395 Summary: ctfconvert from -r310152: gcc 6.2.0 based powerpc64-gcc output sometimes gets new dwarf error/abort from ctfconvert: die <SOME NUMBER>: failed to get ref Product: Base System Version: CURRENT Hardware: powerpc OS: Any Status: New Severity: Affects Only Me Priority: --- Component: bin Assignee: freebsd-bugs@FreeBSD.org Reporter: markmi@dsl-only.net [My archive of typescripts for buildkernel and the like goes back into 2016-July currently.] Prior to the switch to powerpc64-gcc that is based on 6.2.0 none of my buildkernel logs have messages like: ERROR: ctfconvert: die <SOME NUMBER>: failed to get ref: No entry found [dwarf_attrval_unsigned(177)] As far as I can tell from my history this is new, not historical. I'm guessing ctfconvert and its libdwarf have more to do now instead of devel/powerpc64-gcc being wrong. In part this is because of the following side note. [Bugzilla 215350 is about another new ctfconvert issue as of 6.2.0 powerpc64-gcc output --but that leads to SIGSEGV in ctfconvert. Once the SIGSEGV is fixed what lead to the failing code path's use might fit in this this report, although the message might be different.] But I could be wrong. I do not know just how bad of a thing it is for some files to have ctfconve= rt call its terminate instead of finishing normally. Note there are historically lots of messages with text including one of: rc =3D -1 No entry found [dwarf_next_cu_header_c(61)] rc =3D -1 No entry found [dwarf_next_cu_header_c(68)] file does not contain dwarf type data (try compiling with -g) All of those also say "ERROR: ctfconvert:" in the front. It would be easier= to notice the important messages if the above 3 types of messages did not also= say ERROR. Then a grep for ERROR would have fewer false positives to go through. Example messasges, including the ones being reported here: (The agp_if.o messages are tied to the ctfconvert SIGSEGV case.) # grep "ERROR" /root/sys_typescripts/typescript_make_powerpc64vtsc_nodebug_incl_clang_xtoo= lchain_kernel-amd64-host-2016-12-16:14:43:08 | more ERROR: ctfconvert: die 10273: failed to get ref: No entry found [dwarf_attrval_unsigned(177)] ERROR: ctfmerge: No ctf sections found to merge .ERROR_TARGET=3D'agp_if.o' .ERROR_META_FILE=3D'/usr/obj/powerpc64vtsc_xtoolchain_kernel/powerpc.powerp= c64/usr/src/sys/GENERIC64vtsc-NODBG/modules/usr/src/sys/modules/agp/agp_if.= o.meta' .ERROR_TARGET=3D'agp_if.o' .ERROR_META_FILE=3D'/usr/obj/powerpc64vtsc_xtoolchain_kernel/powerpc.powerp= c64/usr/src/sys/GENERIC64vtsc-NODBG/modules/usr/src/sys/modules/agp/agp_if.= o.meta' .ERROR_TARGET=3D'all_subdir_agp' . . . ERROR: ctfconvert: die 26932: failed to get ref: No entry found [dwarf_attrval_unsigned(177)] . . . ERROR: ctfconvert: die 46063: failed to get ref: No entry found [dwarf_attrval_unsigned(177)] . . . ERROR: ctfconvert: die 52618: failed to get ref: No entry found [dwarf_attrval_unsigned(177)] . . . Those new dwarf_attrval_unsigned(177) notes are from processing: sys/modules/accf_http/accf_http.o sys/modules/acl_posix1e/subr_acl_posix1e.o sys/modules/ae/if_ae.o sys/modules/age/if_age.o (The build stops because of the agp_if.o related SIGSEGV so there could be more.) --=20 You are receiving this mail because: You are the assignee for the bug.=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-215395-8>