diff options
author | Alex Richardson <arichardson@FreeBSD.org> | 2020-10-18 18:35:23 +0000 |
---|---|---|
committer | Alex Richardson <arichardson@FreeBSD.org> | 2020-10-18 18:35:23 +0000 |
commit | 3ac62888fc211e92785751c9bb6e73868101f65c (patch) | |
tree | 3765298765360cd5847a1925b888abda4bc00564 /Makefile.inc1 | |
parent | d061adc48d54ebb91b1709b16c59e8ceca895be2 (diff) | |
download | src-3ac62888fc211e92785751c9bb6e73868101f65c.tar.gz src-3ac62888fc211e92785751c9bb6e73868101f65c.zip |
Significantly speed up mkimg_test
It turns out that the majority of the test time for the mkimg tests isn't
mkimg itself but rather the use of jot and hexdump which can be quite slow
on emulated platforms such as QEMU.
On QEMU-RISC-V this reduces the time for `kyua test mkimg_test` from 655
seconds to 200. And for CheriBSD on QEMU-CHERI this saves 4-5 hours (25%
of the time for the entire testsuite!) since jot ends up triggering slow
functions inside the QEMU emulation a lot.
Reviewed By: lwhsu
Differential Revision: https://reviews.freebsd.org/D26796
Notes
Notes:
svn path=/head/; revision=366815
Diffstat (limited to 'Makefile.inc1')
-rw-r--r-- | Makefile.inc1 | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/Makefile.inc1 b/Makefile.inc1 index fbeb0d23d12e..1f5a9641a8db 100644 --- a/Makefile.inc1 +++ b/Makefile.inc1 @@ -2342,6 +2342,8 @@ _other_bootstrap_tools+=tools/build/cross-build/fake_chflags .endif # mkfifo is used by sys/conf/newvers.sh _basic_bootstrap_tools+=usr.bin/mkfifo +# jot is needed for the mkimg tests +_basic_bootstrap_tools+=usr.bin/jot .if ${MK_BOOT} != "no" # md5 is used by boot/beri (and possibly others) |