[Bug 489926] New: build-compare: difference in .note.gnu.build-id and .gnu_debuglink
https://bugzilla.novell.com/show_bug.cgi?id=489926 Summary: build-compare: difference in .note.gnu.build-id and .gnu_debuglink Classification: openSUSE Product: openSUSE 11.2 Version: Factory Platform: Other OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Basesystem AssignedTo: coolo@novell.com ReportedBy: vuntz@novell.com QAContact: qa@suse.de Found By: --- Here's a build-compare output from vinagre in GNOME:Factory compare /.build.oldpackages/vinagre-2.26.0-22.2.src.rpm /usr/src/packages/SRPMS/vinagre-2.26.0-22.3 src.rpm compare /.build.oldpackages/vinagre-2.26.0-22.2.x86_64.rpm /usr/src/packages/RPMS/x86_64/vinagre-2. 26.0-22.3.x86_64.rpm /usr/bin/vinagre differs in ELF sections --- /tmp/tmp.LPRRwLCWKg 2009-03-21 18:40:00.000000000 +0000 +++ /tmp/tmp.YvCffRVmqh 2009-03-21 18:40:00.000000000 +0000 @@ -12,8 +12,8 @@ 400284 00000000 01000a02 ........ Contents of section .note.gnu.build-id: 40028c 04000000 14000000 03000000 474e5500 ............GNU. - 40029c a7587db9 36f3be3f a739d816 1adc9f78 .X}.6..?.9.....x - 4002ac 09829141 ...A + 40029c 1a92b19b 91f18bdc 9a6ce72b 8fa4d0a7 .........l.+.... + 4002ac d98bd033 ...3 Contents of section .hash: 4002b0 09020000 31020000 00000000 00000000 ....1........... 4002c0 00000000 19010000 00000000 1b010000 ................ @@ -11309,4 +11309,4 @@ 0000 4f735057 4700 OsPWG. Contents of section .gnu_debuglink: 0000 76696e61 6772652e 64656275 67000000 vinagre.debug... - 0010 6a768cc6 jv.. + 0010 ba6084b8 .`.. As far as I can tell, it's expected that .note.gnu.build-id is different. See: http://en.opensuse.org/Build_ID http://fedoraproject.org/wiki/Releases/FeatureBuildId#Unique_build_ID http://sources.redhat.com/gdb/current/onlinedocs/gdb_17.html#SEC166 So this section should probably be ignored. As for the .gnu_debuglink part, what's different is the checksum. My understanding is that the different build id could cause the different checksum. Not sure if it's enough to ignore it, though... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=489926 User vuntz@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=489926#c1 --- Comment #1 from Vincent Untz <vuntz@novell.com> 2009-03-27 16:10:27 MST --- Also, related to this: compare /.build.oldpackages/glib2-debuginfo-2.20.0-40.1.x86_64.rpm /usr/src/packages/RPMS/x86_64/gl ib2-debuginfo-2.20.0-40.2.x86_64.rpm --- /tmp/tmp.PmCgxzBEYI 2009-03-18 01:26:15.000000000 +0000 +++ /tmp/tmp.SWmeKuakbo 2009-03-18 01:26:15.000000000 +0000 @@ -1,12 +1,12 @@ -debuginfo(build-id) 32776 0615556079075c895873dedca8e65ce1981304b1 -debuginfo(build-id) 32776 100d4cebaa659dbc05e4d1ed67dd67fedc42f1e7 -debuginfo(build-id) 32776 26ad12a2005eddab19ca3e466f3bbbfe7a544855 -debuginfo(build-id) 32776 4819a679e95388fdb52e8960b13709eb0a684b5a -debuginfo(build-id) 32776 4cca066b4b333df0b3521de4eb8fe81e433cf44d -debuginfo(build-id) 32776 52fe39f1d97135ed16a12e03158920f2acd8e917 -debuginfo(build-id) 32776 550555f00146971c05f893328c27912e6f8d921f -debuginfo(build-id) 32776 7649a517dd3add94954b423ae32d6acef7c515ab -debuginfo(build-id) 32776 96e30267e6a6ff80eabb96814625e9191531577a +debuginfo(build-id) 32776 149630344285b301f6ebbb68a77c292dfe5c11f8 +debuginfo(build-id) 32776 3df4abf084cc1ff6e888f62ab762e449febc3b8b +debuginfo(build-id) 32776 40ee01a14dd77392cf109bbc115f26a6fb2aac04 +debuginfo(build-id) 32776 5b01e0660bace1a179f08db078226ec60634f201 +debuginfo(build-id) 32776 6e81aa4aff25a39bfa312c83a4ccccbaf38aaa1a +debuginfo(build-id) 32776 b9697f25e98cf467eb5510e54e7816f6b00fbd1a +debuginfo(build-id) 32776 d17e0746e7a5634b818ac8cdbfc4a312c40aeb6d +debuginfo(build-id) 32776 dc703abcf434883ac415d6e5fc84f6dfaecbc190 +debuginfo(build-id) 32776 f567cc11b1a44cdb4b01173cfab9ec4c86a23a92 glib2-debuginfo 8 2.20.0-@RELEASE@ rpmlib(PayloadFilesHavePrefix) 16777290 4.0-1 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=489926 User coolo@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=489926#c2 Stephan Kulow <coolo@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #2 from Stephan Kulow <coolo@novell.com> 2009-03-30 02:35:38 MDT --- not sure if INVALID is right here, but this is not a bug. The build id is different for different debuginfos. But the debuginfos will be the same if they are build with the same compiler and the same includes and the same sources. What's different in glibc are the sources. glibc generates several files with build dates in there. What makes this hard to understand for the one looking at the logs is that you only see the first diff - and that's very often the buildid. For vinagre I would think it's the actual binary that's different due to different includes or libraries. So if you'd look past the different build ids, you would see different assembler - otherwise the build id wouldn't be different as it's a hash over the assembler (ok, compiled assembler). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=489926 User vuntz@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=489926#c3 --- Comment #3 from Vincent Untz <vuntz@novell.com> 2009-03-30 04:05:54 MDT --- (In reply to comment #2)
What's different in glibc are the sources. glibc generates several files with build dates in there. What makes this hard to understand for the one looking at the logs is that you only see the first diff - and that's very often the buildid.
s/glibc/glib/ ;-) But you're probably right here.
For vinagre I would think it's the actual binary that's different due to different includes or libraries. So if you'd look past the different build ids, you would see different assembler - otherwise the build id wouldn't be different as it's a hash over the assembler (ok, compiled assembler).
Hrm. But the assembler output is before the elf sections output. So if build-compare doesn't find a difference in the assembler but find one in the elf sections only for the build id, then I have the feeling it's doing something wrong. (and this is happening for really a lot of packages, fwiw) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=489926 User coolo@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=489926#c4 --- Comment #4 from Stephan Kulow <coolo@novell.com> 2009-03-30 04:17:41 MDT --- build-compare doesn't find a difference between internal and external build system's build of factory's vinagre. The buildid can be different too if only strings affecting the debug info will differ. It might be that this is the case if an underlying library changes macros or something like that. I can't tell without having the 2 vinagre rpms ;( But still: only ignoring the build id will not work, there will be other changes that caused the checksum difference. And while the build id might be possible to ignore, those other changes will very likely not. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠novell.com