summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2024-10-21user/rust: Disable print_vtable_sizes testA. Wilcox1-0/+4
This test requires floating point precision that is not available on i586 systems (pmmx), causing the test to fail because of rounding. Closes: #1259
2024-10-19user/clang: Use upstream fix for 32-bit BEA. Wilcox2-25/+156
This fixes the problem more generally, using a union so that 64-bit values can still be stored for offsets; this is required for C++20 module support. Ref: #1255 Fixes: fabdb20dca ("user/clang: Fixes for PPC32, and more generally")
2024-10-17user/llvm*: Generate less debug info on 32-bitZach van Rijn2-2/+16
Full debug-info for LLVM causes multiple failures on 32-bit builders, including with the linker, scanelf, and tar. On 32-bit only, use -ggdb1 to reduce the level of debuginfo output. Since debug information is reduced, we no longer need the `strip` command from #797. Closes: #1251, #1253
2024-10-17user/cbindgen: Fix build on PPC32Zach van Rijn2-2/+17
cbindgen vendors the libc crate, so this needs the same fix as the Rust 1.7x bootstrap compilers. Closes: #1252 Ref: #1248
2024-10-17user/rust: Reduce debug informationZach van Rijn1-1/+1
Similar to the bootstrap compilers, using debuglevel 2 causes 32-bit builders to fail with ENOMEM due to the size of generated debuginfo. Ref: #1249
2024-10-17bootstrap/rust-1.7x: Reduce debug informationZach van Rijn10-10/+10
This still provides debug information, but prevents runaway memory usage which was causing 32-bit builders to fail. Closes: #1249
2024-10-17user/zola: Disable LTO for ppc32 buildA. Wilcox1-1/+4
Closes: #1256
2024-10-17user/clang: Fixes for PPC32, and more generallyA. Wilcox3-11/+90
* Fix upstream issue on all 32-bit BE platforms where pointers are cast in an invalid way, causing crashes when building C++ code. * Improve ppc-interp.patch to ensure *all* executable repl tests will be skipped when run on ppc32. * Disable known-flaky test (found on ppc, but also Gentoo/arm). Closes: #1255
2024-10-17bootstrap/rust-1.[71-75]: Fix libc crate on PPCA. Wilcox10-3/+168
Closes: #1248
2024-10-17bootstrap/rust-1.67,68: Fix PPC crossbeam patchesA. Wilcox4-2/+66
2024-10-17bootstrap/rust-1.66: Fix crossbeam patches for ppcA. Wilcox2-1/+33
Closes: #1247
2024-10-17bootstrap/rust-1.65: Fix crossbeam patches for ppcA. Wilcox2-1/+33
Closes: #1246
2024-10-17bootstrap/rust-1.64: Fix crossbeam patches for ppcA. Wilcox2-1/+33
Closes: #1245
2024-10-17bootstrap/llvm16: Workarounds for PPC test issuesA. Wilcox2-0/+33
Closes: #1244
2024-10-17user/zola: Move back, update to 0.19.1A. Wilcox2-754/+1132
Closes: #1176
2024-10-17user/ripgrep: Update to 14.1.0A. Wilcox1-141/+113
2024-10-17user/cbindgen: Update to 0.26A. Wilcox1-75/+77
2024-10-17user/clazy: Update to 1.12A. Wilcox2-5/+5
Adds LLVM 18 and Clang 18 support.
2024-10-17user/clang: Update to 18.1.8A. Wilcox6-52/+31
2024-10-17user/llvm18: Tweaks and fixesA. Wilcox1-20/+17
* DWARFLinker and friends: most of the test errors are related to musl's thread stack size. Use chelf instead of disabling the tests. They pass on gwyn now, but there are still two flaky dsymutil tests. * Ensure the libLLVM.so.18 library is still present in /usr/lib/llvm18. * Install llvm-gtest, needed for Clang's test suite.
2024-10-17user/rust: Add version 1.80.0A. Wilcox19-0/+3417
2024-10-17experimental/rust-beta: Add 1.81.0 beta patchsetA. Wilcox16-0/+1536
2024-10-17bootstrap/rusts: Don't check after allA. Wilcox18-18/+18
It wastes about 10-15 minutes per build. The tests will fail virtually everywhere until the late 70s (~76 for x86_64, ~78 for ppc64) anyway. I had originally thought it would be good to have a list of the failures present in logs in case we had other failures, but if we do we can always run them ourselves. No reason to waste hours of machine time for test results we throw away anyway.
2024-10-17bootstrap/rust-1.{61-79}: New packageA. Wilcox273-1/+24971
2024-10-17bootstrap/llvm16: New packageA. Wilcox11-0/+960
2024-10-17bootstrap/rust-1.62: New packageA. Wilcox18-0/+1589
2024-10-17bootstrap/rust-1.61: New packageA. Wilcox18-0/+1608
2024-10-17bootstrap/rust-1.60: Bootstrap rustfmtA. Wilcox1-0/+1
Needed to bootstrap newer Rust versions.
2024-10-17user/rust: rename to bootstrap/rust-1.60A. Wilcox19-0/+0
2024-10-10user/bearssl: actually use and install bearssl.pcLaurent Bercot1-1/+4
2024-10-10user/tipidee: fix sysconfdirLaurent Bercot1-0/+1
2024-10-10user/apaste: upgrade to 0.0.2.1Laurent Bercot3-29/+30
2024-10-10user/bcnm: upgrade to 0.0.1.8Laurent Bercot3-11/+31
Also add .pc files.
2024-10-10user/tipidee: upgrade to 0.0.5.1Laurent Bercot4-24/+31
Also add a .pc file.
2024-10-10user/netqmail: download from a place that is actually maintainedLaurent Bercot1-7/+18
Apparently nobody cares about maintaining netqmail.org or qmail.org anymore, has anyone ever heard of perennial URLs? *cue crickets*
2024-10-10user/smtpd-starttls-proxy: upgrade to 0.0.1.5Laurent Bercot1-10/+9
2024-10-10user/s6-networking: upgrade to 2.7.0.4Laurent Bercot2-8/+18
Also add a .pc file.
2024-10-10user/bearssl: upgrade to latest commit ↵Laurent Bercot2-3/+8
(3c040368f6791553610e362401db1efff4b4c5b8)
2024-10-10user/s6-dns: upgrade to 2.4.0.0Laurent Bercot2-8/+14
Also add a .pc file.
2024-10-10user/s6-linux-utils: upgrade to 2.6.2.1Laurent Bercot1-4/+4
2024-10-10user/s6-portable-utils: upgrade to 2.3.0.4Laurent Bercot1-4/+4
2024-10-10system/mdevd: upgrade to 0.1.6.5Laurent Bercot1-3/+3
2024-10-10system/s6-linux-init: upgrade to 1.1.2.1Laurent Bercot2-11/+25
Also add a .pc file.
2024-10-10system/s6-rc: upgrade to 0.5.5.0Laurent Bercot2-11/+18
Also add a .pc file.
2024-10-10system/s6: upgrade to 2.13.1.0Laurent Bercot2-10/+16
Also add a .pc file.
2024-10-10system/execline: upgrade to 2.9.6.1Laurent Bercot2-9/+15
Also add a .pc file.
2024-10-10system/utmps: upgrade to 0.1.2.3Laurent Bercot2-10/+16
Also add a .pc file
2024-10-10system/skalibs: upgrade to 2.14.3.0Laurent Bercot2-7/+14
Also add .pc file
2024-10-03user/tmux: bump { 3.3a --> 3.5 }.Dan Theisen1-2/+2
2024-08-25system/bc: bump { 6.5.0 --> 7.0.0 }.Zach van Rijn1-3/+3
Also disable parallelism during tests. The test suite is not broken, per se, but the output is not synced, so in the event of a failure the logs would be garbled.