Marcus Holland-Moritz
5a25eb104c
chore: don't import duplicate benchmark results
2025-04-21 17:53:56 +02:00
Marcus Holland-Moritz
b155610ca3
build: try perf build for flac also on aarch64
2025-04-21 17:53:56 +02:00
Marcus Holland-Moritz
2d467246fd
chore(benchmark): add initial version of query.py
2025-04-21 17:53:56 +02:00
Marcus Holland-Moritz
d8bf45dae7
chore(benchmark): add history.sh while this isn't integrated yet
2025-04-21 17:53:56 +02:00
Marcus Holland-Moritz
3b70db9aad
chore: add benchmark script
2025-04-21 17:53:56 +02:00
Marcus Holland-Moritz
24f87bbeaa
chore: drop DWARFSEXTRACT_MINIMAL completely
...
Extracting by pattern is a mandatory feature for the `fuse-extract` use
case, so no point in being able to disable this.
2025-04-21 17:53:56 +02:00
Marcus Holland-Moritz
84157a1c9e
build: clean up workflows
2025-04-16 08:35:43 +02:00
Marcus Holland-Moritz
a32704e9e8
chore: update change log
2025-04-16 08:12:55 +02:00
Marcus Holland-Moritz
b6f25aea8c
build: fix flac perf regression on x86_64 by switching to opt_perf
2025-04-15 22:26:08 +02:00
Marcus Holland-Moritz
52a823fbe6
Revert "build: tweak jemalloc build"
...
This reverts commit 6322d5f9e6e9d002dc6a899bde4787b49faef7fa.
2025-04-15 20:28:52 +02:00
Marcus Holland-Moritz
71a6bc1c48
chore: update change log
2025-04-15 18:49:52 +02:00
Marcus Holland-Moritz
6322d5f9e6
build: tweak jemalloc build
2025-04-15 18:40:36 +02:00
Marcus Holland-Moritz
b03e0a6a65
chore: switch to jemalloc, but add mimalloc fuse-extract
2025-04-15 18:40:36 +02:00
Marcus Holland-Moritz
a22cbc3ed6
build: no need to build mimalloc version without minimal
2025-04-15 18:40:36 +02:00
Marcus Holland-Moritz
be348451c7
refactor: make block_cache_byte_buffer_factory internal
2025-04-15 18:40:36 +02:00
Marcus Holland-Moritz
b0bc9b20c4
refactor: simplify OS abstraction for mmap_block and fix move ops
2025-04-15 18:40:36 +02:00
Marcus Holland-Moritz
6cf07409dc
chore(block_cache): produce less noise in verbose logging mode
2025-04-15 13:50:43 +02:00
Marcus Holland-Moritz
4168649c2f
refactor(file_scanner): phmap is as least as fast as F14 (and smaller)
2025-04-14 09:07:56 +02:00
Marcus Holland-Moritz
6cb133c99b
refactor(metadata_v2): phmap is as least as fast as F14 (and smaller)
2025-04-14 08:51:08 +02:00
Marcus Holland-Moritz
43e42994cb
refactor(block_cache): phmap is as least as fast as F14 (and smaller)
2025-04-14 08:50:53 +02:00
Marcus Holland-Moritz
9cdfaffee6
build: update preferred versions of fmt, phmap, gtest
2025-04-14 00:19:48 +02:00
Marcus Holland-Moritz
132e020742
refactor(metadata_v2): use std::stable_sort instead of boost algo
2025-04-14 00:03:00 +02:00
Marcus Holland-Moritz
5cc6da0bda
refactor(filesystem_v2): really avoid dependency on history
2025-04-13 23:47:57 +02:00
Marcus Holland-Moritz
7b78074ce7
fix: *really* disable the performance monitor if requested
2025-04-13 22:27:19 +02:00
Marcus Holland-Moritz
e6ae08475a
feat: allow disabling advanced features in dwarfs-fuse-extract
2025-04-13 22:14:57 +02:00
Marcus Holland-Moritz
3f82610e75
build: simplify static linking
2025-04-13 16:29:52 +02:00
Marcus Holland-Moritz
1675b90d71
test: rename filter_test to tool_filter_test for uniqueness
...
This is essential when building a monolithic test as it would otherwise
collide with the `filter_test` from `dwarfs_test.cpp`.
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
bd961ccfb4
test: (unnecessary) refactor to use span instead of vector
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
8b6002a4e3
test(badfs_test): refactor so we don't require main function
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
3584158a58
build: no need to slow down build CI times by building twice
...
We're already testing that we can build from the source tarball in all
the non-git builds.
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
5f35470ad8
build: need to add -lz on aarch64 for libunwind
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
df63ae7737
build: update openssl, cpptrace
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
40cf58cf24
build: don't build glog unit tests, they cause LTO link errors on aarch64
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
3114401f62
build: also build libunwind
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
b4b930772f
build: simplify Alpine Dockerfile
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
f920d71d74
build: use zstd meson build
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
26ca12e94e
build: also build our own boost
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
ff55a62e77
build: always force using mold for static builds
2025-04-13 16:29:09 +02:00
Marcus Holland-Moritz
4ca5ae3ce8
build: re-enable mold and enable --icf=all / --icf=safe
2025-04-11 22:41:31 +02:00
Marcus Holland-Moritz
876dda3db5
build: apparently we don't have a direct dependency on gflags
2025-04-11 22:41:11 +02:00
Marcus Holland-Moritz
d42f06360f
build: remove unused SCRIPT argument
2025-04-11 21:03:28 +02:00
Marcus Holland-Moritz
af3b4f82bc
build: add lld
2025-04-11 21:03:28 +02:00
Marcus Holland-Moritz
84c6fb6e14
chore: add release-dwarfs.sh
2025-04-11 21:03:28 +02:00
Marcus Holland-Moritz
00f4d64036
docs(mkdwarfs): add a warning to the -B flag
2025-04-10 21:09:19 +02:00
Marcus Holland-Moritz
a30e23bccb
chore: update change log
2025-04-10 18:46:41 +02:00
Marcus Holland-Moritz
7de8cf9344
feat: use an optimized memcpy if possible
2025-04-10 18:45:15 +02:00
Marcus Holland-Moritz
35a1b95fb2
build: get back to the previous benchmark results
...
Turns out that optimizing everything for size, but *not* explicitly
setting `MinRelSize`, delivers the best results. Something to further
investigate in the future, but it's time consuming.
2025-04-10 18:45:15 +02:00
Marcus Holland-Moritz
f1c74b586c
build: optimizing xxhash for size makes it faster
2025-04-10 18:45:15 +02:00
Marcus Holland-Moritz
81639452b0
build: also build xz
2025-04-10 18:45:15 +02:00
Marcus Holland-Moritz
640239b78b
build: also build jemalloc
2025-04-10 18:45:15 +02:00
