On hardened systems racket dont want build, because it use intermediate racket for building itself: racketcgc it placed in racket/.libs/racketcgc also here is lt-racketcgc (in racket/.libs/lt-racketcgc) I'm thinking that pax marks with "mr" flags can be useful for this build
Created attachment 263717 [details] emerge fail log for building racket
When i'm try do such thing in the ebuild, it give error because .libs directory also contain lt-racketcgc executable: src_compile() { #355285: Split the make targets to invoke pax-ctl emake -C racket cgc || die "emake racketcgc failed" pax-mark mxer racket/.libs/racketcgc emake || die "emake default target(3m) failed" if use cgc || use sgc ; then time emake cgc || die "emake cgc-install target failed" fi } so, for fix this error we need run pax-marking after creating lt-racketcgc but before running it.
Here PaX log example for this: Feb 24 23:05:16 localhost kernel: [576744.075427] PAX: execution attempt in: /dev/zero, 9edee000-9ee2e000 00000000 Feb 24 23:05:16 localhost kernel: [576744.075434] PAX: terminating task: /var/tmp/portage/dev-scheme/racket-5.1/work/racket-5.1/src/racket/.libs/lt-racketcgc(lt-racketcgc):21088, uid/euid: 0/0, PC: 9ee17507, SP: b636ba9c Feb 24 23:05:16 localhost kernel: [576744.075443] PAX: bytes at PC: 55 89 e5 53 56 57 89 e8 8b 00 5f 5e 5b 5d c3 50 e8 d4 8d 45 Feb 24 23:05:16 localhost kernel: [576744.075457] PAX: bytes at SP-4: b636bb28 9f28146f 00000008 00000000 00000000 e711ab49 00000000 00000000 9f16296b 9edf62a0 9f175e48 00000008 b636bb38 e711ab49 9f3ee9a0 b636c498 9f24f12b 9f3b830c b636ba88 9f3ee9a0 b636c498 Feb 24 23:18:29 localhost kernel: [577536.594352] PAX: execution attempt in: /dev/zero, a1a95000-a1ad5000 00000000 Feb 24 23:18:29 localhost kernel: [577536.594360] PAX: terminating task: /var/tmp/portage/dev-scheme/racket-5.1/work/racket-5.1/src/racket/.libs/lt-racketcgc(lt-racketcgc):3850, uid/euid: 0/0, PC: a1abe507, SP: b35c872c Feb 24 23:18:29 localhost kernel: [577536.594368] PAX: bytes at PC: 55 89 e5 53 56 57 89 e8 8b 00 5f 5e 5b 5d c3 50 e8 d4 8d 45 Feb 24 23:18:29 localhost kernel: [577536.594383] PAX: bytes at SP-4: b35c87b8 a1f2846f 00000008 00000000 00000000 8695d1d8 00000000 00000000 a1e0996b a1a9d2a0 a1e1ce48 00000008 b35c87c8 8695d1d8 a20959a0 b35c9128 a1ef612b a205f30c b35c8718 a20959a0 b35c9128
*** Bug 355285 has been marked as a duplicate of this bug. ***
Please mark this bug, so that it is obvious that it is in an Overlay. I believe that the tag 'INOVERLAY' is used.
(In reply to comment #5) > Please mark this bug, so that it is obvious that it is in an Overlay. I believe > that the tag 'INOVERLAY' is used. Actually, the InOverlay keyword is more used to flag a bug that is fixed in overlay but not yet in the main tree. As the racket ebuild is only available in lisp overlay, your first remark still holds, so I changed the topic accordingly.
racket-5.1.1 is in main tree and it probably suffers from the same issue. Please test 999 from the overlay though. Anton, mailing upstream directly is alright, but try to remember that they do not use gentoo and are not familiar with ebuilds or their internals, so try to describe problems in generic terms.
(In reply to comment #7) > racket-5.1.1 is in main tree and it probably suffers from the same issue. > Please test 999 from the overlay though. > > Anton, mailing upstream directly is alright, but try to remember that they do > not use gentoo and are not familiar with ebuilds or their internals, so try to > describe problems in generic terms. Thx! Racket-999 works ok!
That is good news! I'll leave this bug open until the next version is in main tree.
Newer versions are in tree so I'm closing this bug. Please reopen if problem reappears.
