Bugzilla – Bug 1181652
smlnj varies between builds
Last modified: 2021-03-02 21:28:47 UTC
While working on reproducible builds for openSUSE, I found that our smlnj package varies for each build. +++ new /usr/lib64/smlnj/lib/xml-lib.cm/.cm/amd64-unix/xml-lib.cm (hex) @@ -1,4 +1,4 @@ -00000000 ac 24 34 bc 54 ff 03 61 53 35 85 7e 76 45 dd c4 |.$4.T..aS5.~vE..| +00000000 cc f9 af ec d7 3a 5c c5 24 ac e7 60 af f4 dc d1 |.....:\.$..`....| @@ -30,7 +30,7 @@ 000001d0 b1 00 00 12 e6 6e b2 00 e4 a2 3c 50 65 72 76 61 |.....n....<Perva| 000001e0 73 69 76 65 3e 22 ff 86 3c ff 85 13 ff 86 3c 00 |sive>"..<.....<.| 000001f0 98 18 31 70 ff 81 30 37 47 63 6e c3 42 f0 a2 47 |..1p..07Gcn.B..G| -00000200 5a ab fb 4e cb f5 f8 8d 78 67 e8 4f 94 24 ed 22 |Z..N....xg.O.$."| +00000200 5a ab fb 4e cb f5 f9 84 7a 61 e8 4e 94 27 ed 22 |Z..N....za.N.'."| 00000210 00 f3 e3 a2 58 4d 4c 5f 53 43 48 45 4d 41 22 74 |....XML_SCHEMA"t| my tools tell me that such files are created by "/home/abuild/rpmbuild/BUILD/smlnj-110.99/bin/.run/run.amd64-linux", "@SMLcmdname=/home/abuild/rpmbuild/BUILD/smlnj-110.99/bin/sml", "@SMLload=/home/abuild/rpmbuild/BUILD/smlnj-110.99/bin/.heap/sml", "-m", "$smlnj/installer.cm" Also it is not one of the usual sources of non-determinism. Maybe from randomness or ASLR? https://salsa.debian.org/reproducible-builds/reproducible-notes/-/blob/0f182efded/packages.yml#L26854 lists - random_order_in_static_libraries
Confirmed. Saw the issue crop up in another file. This is a snippet from the recent build log after 1 manual rebuild: https://build.opensuse.org/package/live_build_log/devel:tools:compiler/smlnj/openSUSE_Factory/x86_64 ``` /usr/lib64/smlnj/bin/.heap/asdlgen.amd64-linux differs at offset '13' (data) --- old /usr/lib64/smlnj/bin/.heap/asdlgen.amd64-linux (hex) +++ new /usr/lib64/smlnj/bin/.heap/asdlgen.amd64-linux (hex) @@ -1,9 +1,9 @@ -00000000 33 22 11 00 04 20 08 09 02 00 00 00 d7 55 00 00 |3"... .......U..| -00000010 80 ea ac 3b d7 55 00 00 70 27 ac 3b d7 55 00 00 |...;.U..p'.;.U..| +00000000 33 22 11 00 04 20 08 09 02 00 00 00 56 55 00 00 |3"... ......VU..| +00000010 80 2a 16 97 56 55 00 00 70 67 15 97 56 55 00 00 |.*..VU..pg..VU..| 00000020 00 00 00 00 01 00 00 00 05 00 00 00 04 00 00 00 |................| 00000030 01 00 00 00 0a 00 00 00 02 00 00 00 00 00 08 00 |................| -00000040 00 00 00 00 01 00 00 00 00 00 00 00 08 00 e0 08 |................| -00000050 a0 7f 00 00 01 00 00 00 00 00 00 00 24 00 00 00 |............$...| +00000040 00 00 00 00 01 00 00 00 00 00 00 00 08 00 9c bc |................| +00000050 d4 7f 00 00 01 00 00 00 00 00 00 00 24 00 00 00 |............$...| 00000060 70 02 00 00 43 53 74 72 75 63 74 2e 44 69 76 49 |p...CStruct.DivI| 00000070 64 00 43 53 74 72 75 63 74 2e 4f 76 65 72 66 6c |d.CStruct.Overfl| 00000080 6f 77 49 64 00 43 53 74 72 75 63 74 2e 53 79 73 |owId.CStruct.Sys| ``` Unclear what the source of in-determinism is yet.