| Message ID | 378315057.2531045.1535207584068@mail.yahoo.com |
|---|---|
| State | Superseded |
| Headers | show |
Please use git send-email to send your patches, or at least send each patch, created by git format-patch, as individual attachment. Your files seem to contain multiple patches one after another, which makes them very hard to follow. But nice work! Let's hope this does not cause any regressions.
On Sat, Aug 25, 2018 at 02:33:03PM +0000, avih wrote: > Hi, > > I noticed that configure can be a bit slow - few minutes on my macOS > system, 1:30m on linux/bash, 30s on linux/dash, and even if > pathological, still a too-long 10 minutes on Windows (MSYS2) (same > duration as a full build with make -j4 after configure). > > I added some timing information printouts for various parts in > configure, and eventually identified few culprits. > > The attachment "config-timing.patch" adds these printouts and sorted > summary. It's attached for reference and not intended to be merged. > It applies cleanly before or after the main patch. > > > The attached "main.patch" addresses three areas I identified as slow: > > 1. About 50-70% of configure runtime was being spent inside one > function: flatten_extralibs() and callees resolve() and unique(). > It manipulates strings and invoked nearly 20K (20000) subshells. > It was rewritten to avoid subshells, and ended up x50-x250 faster. > > 2. print_enabled_components() was invoking sed about 350 times on one > file. This is never instant but takes many seconds where fork is > slow (Windows). Invoke sed only once instead = x4-x10 speedup. > > 3. After the previous speedups, configure spent 20-60% of its runtime > at check_deps(). It's particularly slow with bash. After some local > optimizations - mainly avoid pushvar/popvar and abort early in one > notable case (empty deps), it's now x4-x25 faster. > > > Some resulting speedups (more details at "before-after.txt"): > > - macOS/bash: total: 98s -> 22s flatten_extralibs: 53s -> 0.7s > - Linux/bash: total: 87s -> 11s flatten_extralibs: 59s -> 0.9s > - Linux/dash: total: 27s -> 8s flatten_extralibs: 17s -> 0.2s > - FreeBSD/sh: total: 34s -> 9s flatten_extralibs: 23s -> 0.2s > - MSYS2/bash: total: 10m -> 2:30m flatten_extralibs: 400s -> 1.5s (!) > > > Notes: > > - unique() is modified with flatten_extralibs. It now outputs different > order: it was keeping the last instance of recurring items, now it > keeps the first. It affects libs order at ffbuild/config.{mak,sh} - > but I don't think it matters. If it does, "opt1-reorder-unique.patch" > restores the original order. Let me know if/why it matters and I'll > squash it and update the commit message accordingly if required. > > - After the check_deps() patch, pushvar() and popvar() are not used but > I was hesitant to remove them (nice to have). If you think it should > be removed, the patch "opt2-remove-pushvar.patch" removes them. > > - The patches assume POSIX shell and don't use anything "tricky". > It was tested with dash, bash, busybox-ash, freebsd-sh, ksh93u, mksh. > > - Thanks to tmm1 and atomnuker for their help with testing. > > > This is my first mail to this list, and I wasn't sure how to split the > patches. After asking a veteran, it was suggested to me that I should > put all of them in one email. Just let me know if you prefer something > else (dev mailing lists are not my forte...). with the main patch make distclean ; dash ./configure --enable-gpl && make -j12 testprogs fails: LD libavfilter/tests/filtfmts libavformat/libavformat.a(utils.o): In function `av_apply_bitstream_filters': ffmpeg/libavformat/utils.c:5577: undefined reference to `av_bitstream_filter_filter' libavformat/libavformat.a(codec2.o): In function `codec2_read_header_common': ffmpeg/libavformat/codec2.c:74: undefined reference to `avpriv_codec2_mode_bit_rate' ffmpeg/libavformat/codec2.c:75: undefined reference to `avpriv_codec2_mode_frame_size' ffmpeg/libavformat/codec2.c:76: undefined reference to `avpriv_codec2_mode_block_align' ffmpeg/libavformat/codec2.c:74: undefined reference to `avpriv_codec2_mode_bit_rate' ffmpeg/libavformat/codec2.c:75: undefined reference to `avpriv_codec2_mode_frame_size' ffmpeg/libavformat/codec2.c:76: undefined reference to `avpriv_codec2_mode_block_align' libavformat/libavformat.a(spdifdec.o): In function `spdif_get_offset_and_codec': ffmpeg/libavformat/spdifdec.c:63: undefined reference to `av_adts_header_parse' ffmpeg/libavformat/spdifdec.c:63: undefined reference to `av_adts_header_parse' libavformat/libavformat.a(spdifenc.o): In function `spdif_header_aac': ffmpeg/libavformat/spdifenc.c:357: undefined reference to `av_adts_header_parse' collect2: error: ld returned 1 exit status make: *** [libavfilter/tests/filtfmts] Error 1 [...]
Thanks.
I'll post the 3 parts of main.patch unmodified as individualemails like Timo requested, and then I'll look at the failures.
On Saturday, August 25, 2018 9:00 PM, Michael Niedermayer <michael@niedermayer.cc> wrote:
On Sat, Aug 25, 2018 at 02:33:03PM +0000, avih wrote:
> Hi,
>
> I noticed that configure can be a bit slow - few minutes on my macOS
> system, 1:30m on linux/bash, 30s on linux/dash, and even if
> pathological, still a too-long 10 minutes on Windows (MSYS2) (same
> duration as a full build with make -j4 after configure).
>
> I added some timing information printouts for various parts in
> configure, and eventually identified few culprits.
>
> The attachment "config-timing.patch" adds these printouts and sorted
> summary. It's attached for reference and not intended to be merged.
> It applies cleanly before or after the main patch.
>
>
> The attached "main.patch" addresses three areas I identified as slow:
>
> 1. About 50-70% of configure runtime was being spent inside one
> function: flatten_extralibs() and callees resolve() and unique().
> It manipulates strings and invoked nearly 20K (20000) subshells.
> It was rewritten to avoid subshells, and ended up x50-x250 faster.
>
> 2. print_enabled_components() was invoking sed about 350 times on one
> file. This is never instant but takes many seconds where fork is
> slow (Windows). Invoke sed only once instead = x4-x10 speedup.
>
> 3. After the previous speedups, configure spent 20-60% of its runtime
> at check_deps(). It's particularly slow with bash. After some local
> optimizations - mainly avoid pushvar/popvar and abort early in one
> notable case (empty deps), it's now x4-x25 faster.
>
>
> Some resulting speedups (more details at "before-after.txt"):
>
> - macOS/bash: total: 98s -> 22s flatten_extralibs: 53s -> 0.7s
> - Linux/bash: total: 87s -> 11s flatten_extralibs: 59s -> 0.9s
> - Linux/dash: total: 27s -> 8s flatten_extralibs: 17s -> 0.2s
> - FreeBSD/sh: total: 34s -> 9s flatten_extralibs: 23s -> 0.2s
> - MSYS2/bash: total: 10m -> 2:30m flatten_extralibs: 400s -> 1.5s (!)
>
>
> Notes:
>
> - unique() is modified with flatten_extralibs. It now outputs different
> order: it was keeping the last instance of recurring items, now it
> keeps the first. It affects libs order at ffbuild/config.{mak,sh} -
> but I don't think it matters. If it does, "opt1-reorder-unique.patch"
> restores the original order. Let me know if/why it matters and I'll
> squash it and update the commit message accordingly if required.
>
> - After the check_deps() patch, pushvar() and popvar() are not used but
> I was hesitant to remove them (nice to have). If you think it should
> be removed, the patch "opt2-remove-pushvar.patch" removes them.
>
> - The patches assume POSIX shell and don't use anything "tricky".
> It was tested with dash, bash, busybox-ash, freebsd-sh, ksh93u, mksh.
>
> - Thanks to tmm1 and atomnuker for their help with testing.
>
>
> This is my first mail to this list, and I wasn't sure how to split the
> patches. After asking a veteran, it was suggested to me that I should
> put all of them in one email. Just let me know if you prefer something
> else (dev mailing lists are not my forte...).
with the main patch
make distclean ; dash ./configure --enable-gpl && make -j12 testprogs
fails:
LD libavfilter/tests/filtfmts
libavformat/libavformat.a(utils.o): In function `av_apply_bitstream_filters':
ffmpeg/libavformat/utils.c:5577: undefined reference to `av_bitstream_filter_filter'
libavformat/libavformat.a(codec2.o): In function `codec2_read_header_common':
ffmpeg/libavformat/codec2.c:74: undefined reference to `avpriv_codec2_mode_bit_rate'
ffmpeg/libavformat/codec2.c:75: undefined reference to `avpriv_codec2_mode_frame_size'
ffmpeg/libavformat/codec2.c:76: undefined reference to `avpriv_codec2_mode_block_align'
ffmpeg/libavformat/codec2.c:74: undefined reference to `avpriv_codec2_mode_bit_rate'
ffmpeg/libavformat/codec2.c:75: undefined reference to `avpriv_codec2_mode_frame_size'
ffmpeg/libavformat/codec2.c:76: undefined reference to `avpriv_codec2_mode_block_align'
libavformat/libavformat.a(spdifdec.o): In function `spdif_get_offset_and_codec':
ffmpeg/libavformat/spdifdec.c:63: undefined reference to `av_adts_header_parse'
ffmpeg/libavformat/spdifdec.c:63: undefined reference to `av_adts_header_parse'
libavformat/libavformat.a(spdifenc.o): In function `spdif_header_aac':
ffmpeg/libavformat/spdifenc.c:357: undefined reference to `av_adts_header_parse'
collect2: error: ld returned 1 exit status
make: *** [libavfilter/tests/filtfmts] Error 1
[...]
On 08/25/18 11:11 AM, avih wrote: > After the previous speedups, configure spent 20-60% of its runtime > at check_deps(). It's particularly slow with bash. After some local > optimizations - mainly avoid pushvar/popvar and abort early in one > notable case (empty deps), it's now x4-x25 faster. Works great on OS/2, from 700 seconds to 144 seconds Dave
On 26-8-2018 22:27, Dave Yeo <daveryeo@telus.net> wrote: > On 08/25/18 11:11 AM, avih wrote: >> After the previous speedups, configure spent 20-60% of its runtime >> at check_deps(). It's particularly slow with bash. After some local >> optimizations - mainly avoid pushvar/popvar and abort early in one >> notable case (empty deps), it's now x4-x25 faster. > > Works great on OS/2, from 700 seconds to 144 seconds > Dave I've gone from 857s to 268s on my old WinXP pc. That's really impressive! -- Reino
From 77f897c8ed4eec9119d758037b0311629f549a5b Mon Sep 17 00:00:00 2001 From: "Avi Halachmi (:avih)" <avihpit@yahoo.com> Date: Wed, 1 Aug 2018 09:10:12 +0300 Subject: [PATCH] configure: remove unused pushvar()/popvar() --- configure | 19 ------------------- 1 file changed, 19 deletions(-) diff --git a/configure b/configure index e9cb7703..d6c1d032 100755 --- a/configure +++ b/configure @@ -619,25 +619,6 @@ get_sanitized(){ eval echo \$$(sanitize_var_name "$1") } -pushvar(){ - for pvar in $*; do - eval level=\${${pvar}_level:=0} - eval ${pvar}_${level}="\$$pvar" - eval ${pvar}_level=$(($level+1)) - done -} - -popvar(){ - for pvar in $*; do - eval level=\${${pvar}_level:-0} - test $level = 0 && continue - eval level=$(($level-1)) - eval $pvar="\${${pvar}_${level}}" - eval ${pvar}_level=$level - eval unset ${pvar}_${level} - done -} - request(){ for var in $*; do eval ${var}_requested=yes -- 2.17.1
Hi, I noticed that configure can be a bit slow - few minutes on my macOS system, 1:30m on linux/bash, 30s on linux/dash, and even if pathological, still a too-long 10 minutes on Windows (MSYS2) (same duration as a full build with make -j4 after configure). I added some timing information printouts for various parts in configure, and eventually identified few culprits. The attachment "config-timing.patch" adds these printouts and sorted summary. It's attached for reference and not intended to be merged. It applies cleanly before or after the main patch. The attached "main.patch" addresses three areas I identified as slow: 1. About 50-70% of configure runtime was being spent inside one function: flatten_extralibs() and callees resolve() and unique(). It manipulates strings and invoked nearly 20K (20000) subshells. It was rewritten to avoid subshells, and ended up x50-x250 faster. 2. print_enabled_components() was invoking sed about 350 times on one file. This is never instant but takes many seconds where fork is slow (Windows). Invoke sed only once instead = x4-x10 speedup. 3. After the previous speedups, configure spent 20-60% of its runtime at check_deps(). It's particularly slow with bash. After some local optimizations - mainly avoid pushvar/popvar and abort early in one notable case (empty deps), it's now x4-x25 faster. Some resulting speedups (more details at "before-after.txt"): - macOS/bash: total: 98s -> 22s flatten_extralibs: 53s -> 0.7s - Linux/bash: total: 87s -> 11s flatten_extralibs: 59s -> 0.9s - Linux/dash: total: 27s -> 8s flatten_extralibs: 17s -> 0.2s - FreeBSD/sh: total: 34s -> 9s flatten_extralibs: 23s -> 0.2s - MSYS2/bash: total: 10m -> 2:30m flatten_extralibs: 400s -> 1.5s (!) Notes: - unique() is modified with flatten_extralibs. It now outputs different order: it was keeping the last instance of recurring items, now it keeps the first. It affects libs order at ffbuild/config.{mak,sh} - but I don't think it matters. If it does, "opt1-reorder-unique.patch" restores the original order. Let me know if/why it matters and I'll squash it and update the commit message accordingly if required. - After the check_deps() patch, pushvar() and popvar() are not used but I was hesitant to remove them (nice to have). If you think it should be removed, the patch "opt2-remove-pushvar.patch" removes them. - The patches assume POSIX shell and don't use anything "tricky". It was tested with dash, bash, busybox-ash, freebsd-sh, ksh93u, mksh. - Thanks to tmm1 and atomnuker for their help with testing. This is my first mail to this list, and I wasn't sure how to split the patches. After asking a veteran, it was suggested to me that I should put all of them in one email. Just let me know if you prefer something else (dev mailing lists are not my forte...). For convinience, the patches are also available at: https://gist.github.com/avih/f51008225d4a20a0981daed1faca4bc2 Regards, Avi configure - speedup - before/after details Systems: - Linux: Ubuntu 16.04 VM: bash 4.3/4.4, dash 4.4, busybox-ash 1.29, ksh93u - macOS: iMac 5k 2017 i7: bash 3.2 - macOS: MBA late 2010: bash 3.2, dash 0.5.10 - FreeBSD: GhostBSD 11.1: sh - MSYS2: latest (gcc 8): bash 4.4, dash 0.5.10 - Linux: debian unstable: dash, bash --------------------------------------------- Linux - Ubuntu 16.04 - VM (i7 3630qm) /bin/sh (dash): 28s -> 8.3s = x3.3 bash 4.3 : 103s -> 12.5s = x8 bash 4.4 : 87s -> 11.3s = x8 busybox ash : 37s -> 9s = x4 ksh93 : 23s -> 9s = x2.5 --------------------------------------------- bash 4.3 BEFORE: 102787 ms, 1 ms/measure, Linux/bash (4.3.48) ----- 72526 ms 71 % flatten_extralibs_wrapper() 18329 ms 18 % check_deps() 8295 ms 8 % Compilation tests 2062 ms 2 % print_enabled_components() 992 ms 1 % Writing config 343 ms 0 % Prepare deps 190 ms 0 % Options review 33 ms 0 % Init 17 ms 0 % Finalize bash 4.3 AFTER: 12529 ms, 1 ms/measure, Linux/bash (4.3.48) ----- 8358 ms 67 % Compilation tests 1408 ms 11 % check_deps() 920 ms 7 % flatten_extralibs_wrapper() 735 ms 6 % Writing config 520 ms 4 % print_enabled_components() 346 ms 3 % Prepare deps 190 ms 2 % Options review 35 ms 0 % Init 17 ms 0 % Finalize bash 4.4 BEFORE: 86601 ms, 1 ms/measure, Linux/bash (4.4.19) ----- 58889 ms 68 % flatten_extralibs_wrapper() 16288 ms 19 % check_deps() 8127 ms 9 % Compilation tests 2013 ms 2 % print_enabled_components() 793 ms 1 % Writing config 253 ms 0 % Prepare deps 188 ms 0 % Options review 34 ms 0 % Init 16 ms 0 % Finalize bash 4.4 AFTER: 11373 ms, 1 ms/measure, Linux/bash (4.4.19) ----- 8170 ms 72 % Compilation tests 985 ms 9 % check_deps() 758 ms 7 % flatten_extralibs_wrapper() 555 ms 5 % Writing config 425 ms 4 % print_enabled_components() 250 ms 2 % Prepare deps 182 ms 2 % Options review 32 ms 0 % Init 16 ms 0 % Finalize dash 0.5.8 BEFORE: 28308 ms, 1 ms/measure, Linux/dash ----- 18306 ms 65 % flatten_extralibs_wrapper() 7450 ms 26 % Compilation tests 1260 ms 4 % check_deps() 895 ms 3 % print_enabled_components() 227 ms 1 % Writing config 88 ms 0 % Options review 59 ms 0 % Prepare deps 14 ms 0 % Init 9 ms 0 % Finalize dash 0.5.8 AFTER: 8312 ms, 1 ms/measure, Linux/dash ----- 7390 ms 89 % Compilation tests 298 ms 4 % check_deps() 218 ms 3 % flatten_extralibs_wrapper() 136 ms 2 % Writing config 100 ms 1 % print_enabled_components() 88 ms 1 % Options review 59 ms 1 % Prepare deps 14 ms 0 % Init 9 ms 0 % Finalize busybox-ash 1.29 BEFORE: 36580 ms, 1 ms/measure, Linux/busybox (ash) ----- 25559 ms 70 % flatten_extralibs_wrapper() 7580 ms 21 % Compilation tests 1974 ms 5 % check_deps() 962 ms 3 % print_enabled_components() 277 ms 1 % Writing config 107 ms 0 % Options review 92 ms 0 % Prepare deps 20 ms 0 % Init 9 ms 0 % Finalize busybox-ash 1.29 AFTER: 8888 ms, 1 ms/measure, Linux/busybox (ash) ----- 7603 ms 86 % Compilation tests 452 ms 5 % check_deps() 311 ms 3 % flatten_extralibs_wrapper() 189 ms 2 % Writing config 113 ms 1 % print_enabled_components() 101 ms 1 % Options review 91 ms 1 % Prepare deps 18 ms 0 % Init 10 ms 0 % Finalize ksh93u BEFORE: 22483 ms, 0 ms/measure, Linux/ksh93 ----- 11985 ms 53 % flatten_extralibs_wrapper() 7353 ms 33 % Compilation tests 1690 ms 8 % check_deps() 914 ms 4 % print_enabled_components() 330 ms 1 % Writing config 108 ms 0 % Options review 79 ms 0 % Prepare deps 14 ms 0 % Init 10 ms 0 % Finalize ksh93u AFTER: 8898 ms, 1 ms/measure, Linux/ksh93 ----- 7511 ms 84 % Compilation tests 353 ms 4 % check_deps() 333 ms 4 % flatten_extralibs_wrapper() 328 ms 4 % Writing config 162 ms 2 % print_enabled_components() 107 ms 1 % Options review 80 ms 1 % Prepare deps 15 ms 0 % Init 9 ms 0 % Finalize --------------------------------------------- tmm1: macOS (iMac 5k 2017 i7) /bin/sh (bash 3.2): 98s -> 22s = x4.5 --------------------------------------------- bash BEFORE: 98134 ms, 30 ms/measure, Darwin/sh ----- 52947 ms 54 % flatten_extralibs_wrapper() 23608 ms 24 % check_deps() 17392 ms 18 % Compilation tests 2603 ms 3 % print_enabled_components() 994 ms 1 % Writing config 281 ms 0 % Prepare deps 251 ms 0 % Options review 39 ms 0 % Init 19 ms 0 % Finalize bash AFTER: 21981 ms, 29 ms/measure, Darwin/sh ----- 17783 ms 81 % Compilation tests 1396 ms 6 % check_deps() 791 ms 4 % Writing config 733 ms 3 % print_enabled_components() 682 ms 3 % flatten_extralibs_wrapper() 285 ms 1 % Prepare deps 248 ms 1 % Options review 40 ms 0 % Init 23 ms 0 % Finalize --------------------------------------------- macOS (MBA late 2010) /bin/sh (bash 3.2): 272s -> 61s = x4.5 dash : 114s -> 51s = x2.3 --------------------------------------------- bash 3.2 BEFORE: 272643 ms, 67 ms/measure, Darwin/sh (bash 3.2.57) ----- 140802 ms 52 % flatten_extralibs_wrapper() 71122 ms 26 % check_deps() 48587 ms 18 % Compilation tests 7445 ms 3 % print_enabled_components() 2898 ms 1 % Writing config 797 ms 0 % Prepare deps 791 ms 0 % Options review 133 ms 0 % Init 68 ms 0 % Finalize bash 3.2 AFTER: 60802 ms, 68 ms/measure, Darwin/sh (bash 3.2.57) ----- 48835 ms 80 % Compilation tests 3663 ms 6 % check_deps() 2212 ms 4 % Writing config 2209 ms 4 % print_enabled_components() 2084 ms 3 % flatten_extralibs_wrapper() 807 ms 1 % Options review 790 ms 1 % Prepare deps 147 ms 0 % Init 55 ms 0 % Finalize dash 0.5.10 before and after: Note, I got some reports that configure sometimes fails in dash, possibly under specific configurtions. This wasn't investigated further. dash 0.5.10 BEFORE: 114689 ms, 66 ms/measure, Darwin/dash ----- 56977 ms 50 % flatten_extralibs_wrapper() 46503 ms 41 % Compilation tests 5466 ms 5 % check_deps() 3732 ms 3 % print_enabled_components() 1216 ms 1 % Writing config 393 ms 0 % Options review 256 ms 0 % Prepare deps 102 ms 0 % Init 44 ms 0 % Finalize dash 0.5.10 AFTER: 51010 ms, 67 ms/measure, Darwin/dash ----- 46599 ms 91 % Compilation tests 1201 ms 2 % check_deps() 999 ms 2 % flatten_extralibs_wrapper() 835 ms 2 % Writing config 577 ms 1 % print_enabled_components() 405 ms 1 % Options review 254 ms 0 % Prepare deps 93 ms 0 % Init 47 ms 0 % Finalize --------------------------------------------- FreeBSD - GhostBSD 11.1 live DVD - VM (i7 4500u) /bin/sh: 34.5s -> 9s = x4 --------------------------------------------- sh BEFORE: 34410 ms, 15 ms/measure, FreeBSD/sh ----- 23270 ms 68 % flatten_extralibs_wrapper() 7606 ms 22 % Compilation tests 1875 ms 5 % print_enabled_components() 789 ms 2 % check_deps() 506 ms 1 % Writing config 271 ms 1 % Options review 53 ms 0 % Prepare deps 23 ms 0 % Init 17 ms 0 % Finalize sh AFTER: 9045 ms, 15 ms/measure, FreeBSD/sh ----- 7691 ms 85 % Compilation tests 381 ms 4 % Writing config 270 ms 3 % check_deps() 266 ms 3 % Options review 186 ms 2 % flatten_extralibs_wrapper() 149 ms 2 % print_enabled_components() 59 ms 1 % Prepare deps 23 ms 0 % Init 20 ms 0 % Finalize --------------------------------------------- MSYS2 - Windows 8.1 (i7 4500u) /bin/sh (bash 4.4.19): 600s -> 148s = x4 dash : 438s -> 139s = x3 --------------------------------------------- bash BEFORE: 599395 ms, 37 ms/measure, MINGW64_NT-6.3/sh ----- 396352 ms 66 % flatten_extralibs_wrapper() 133812 ms 22 % Compilation tests 36044 ms 6 % check_deps() 21561 ms 4 % print_enabled_components() 6902 ms 1 % Writing config 3364 ms 1 % Options review 742 ms 0 % Prepare deps 351 ms 0 % Init 267 ms 0 % Finalize bash AFTER: 147484 ms, 36 ms/measure, MINGW64_NT-6.3/sh (bash 4.4.19) ----- 133968 ms 91 % Compilation tests 3502 ms 2 % Options review 3247 ms 2 % Writing config 2307 ms 2 % print_enabled_components() 1553 ms 1 % flatten_extralibs_wrapper() 1504 ms 1 % check_deps() 728 ms 0 % Prepare deps 379 ms 0 % Init 296 ms 0 % Finalize FFmpeg's configure doesn't work in MSYS2's dash out of the box. mingw pkg-config outputs \r\n line endings, and while MSYS2's bash is patched to handle them, MSYS2's dash is not, and requires to strip them, for instance using this pkg-config-wrapper file: ``` #!/bin/sh out=$("$MINGW_PREFIX"/bin/pkg-config "$@"; printf %s "-$?") e=${out##*-} out=${out%-*} printf %s "$out" | tr -d '\r' exit $e ``` and then invoke configure --pkg-config=path/to/pkg-config-wrapper ... dash BEFORE: 497919 ms, 30 ms/measure, MINGW64_NT-6.3/dash ----- 337320 ms 68 % flatten_extralibs_wrapper() 129702 ms 26 % Compilation tests 18802 ms 4 % print_enabled_components() 6150 ms 1 % Writing config 3269 ms 1 % Options review 1707 ms 0 % check_deps() 400 ms 0 % Prepare deps 331 ms 0 % Init 238 ms 0 % Finalize dash AFTER: 138592 ms, 29 ms/measure, MINGW64_NT-6.3/dash ----- 128895 ms 93 % Compilation tests 3067 ms 2 % Options review 2570 ms 2 % Writing config 1508 ms 1 % print_enabled_components() 1013 ms 1 % flatten_extralibs_wrapper() 425 ms 0 % Prepare deps 422 ms 0 % check_deps() 374 ms 0 % Finalize 318 ms 0 % Init ------------------------------------------ atomnuker: Debian Unstable, Thinkpad X1 Carbon 4th Gen, Intel(R) Core(TM) i7-6600U CPU @ 2.60GHz dash: 16s -> 6.3s = x2.6 bash: 50s -> 8.3s = x6 ------------------------------------------ dash BEFORE: Command line: dash ./configure --disable-autodetect ----- 16132 ms, 2 ms/measure, Linux/dash ----- 8402 ms 52 % flatten_extralibs_wrapper() 5796 ms 36 % Compilation tests 1024 ms 6 % check_deps() 626 ms 4 % print_enabled_components() 129 ms 1 % Writing config 72 ms 0 % Options review 48 ms 0 % Prepare deps 30 ms 0 % Init 5 ms 0 % Finalize dash AFTER: Command line: dash ./configure --disable-autodetect ----- 6366 ms, 0 ms/measure, Linux/dash ----- 5661 ms 89 % Compilation tests 244 ms 4 % check_deps() 205 ms 3 % flatten_extralibs_wrapper() 90 ms 1 % Writing config 52 ms 1 % print_enabled_components() 49 ms 1 % Prepare deps 48 ms 1 % Options review 9 ms 0 % Init 8 ms 0 % Finalize bash BEFORE: Command line: bash ./configure --disable-autodetect ----- 50069 ms, 3 ms/measure, Linux/bash ----- 31212 ms 62 % flatten_extralibs_wrapper() 9665 ms 19 % check_deps() 6309 ms 13 % Compilation tests 1955 ms 4 % print_enabled_components() 591 ms 1 % Writing config 168 ms 0 % Prepare deps 99 ms 0 % Options review 61 ms 0 % Init 9 ms 0 % Finalize bash AFTER: ----- 8279 ms, 3 ms/measure, Linux/bash ----- 6255 ms 76 % Compilation tests 623 ms 8 % check_deps() 432 ms 5 % flatten_extralibs_wrapper() 397 ms 5 % Writing config 208 ms 3 % print_enabled_components() 193 ms 2 % Prepare deps 102 ms 1 % Options review 61 ms 1 % Init 8 ms 0 % Finalize