<div><br></div><div><br><div class="gmail_quote"><div dir="ltr">On Wed, Jan 9, 2019 at 11:17 PM Robert Yang <<a href="mailto:liezhi.yang@windriver.com">liezhi.yang@windriver.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi,<br>
<br>
I reimplemented glibc's site config and did a lot of tests recently, finally,<br>
I found that it isn't as useful as I had thought, so I'm leaning to remove<br>
classes/siteconfig.bbclass.<br>
<br>
* Here are the test results of with/without glibc's siteconfig:<br>
$ bitbake recipe -cconfigure && bitbake recipe -ccleansstate && time bitbake <br>
recipe -cconfigure<br>
- gettext:<br>
without: 2m38.382s<br>
with: 2m33.099s<br>
<br>
- coreutils:<br>
without: 1m19.967s<br>
with: 1m13.574s<br>
<br>
- bison:<br>
without: 0m56.203s<br>
with: 0m50.134s<br>
<br>
These 3 recipes' do_configure are very time consuming, we can see that it saves<br>
about 6s for bison/coreutils/gettext's do_configure, but they need depend on<br>
glibc's site config recipe (<a href="http://glibc-site.bb" rel="noreferrer" target="_blank">glibc-site.bb</a>), while glibc-site requires about 30s<br>
to build, so if we count glibc-site's 30s, it seems not worth. And I also tested<br>
again and again on core-image-minimal and core-image-sato with/without glibc's<br>
siteconfig, there is nearly no difference, about 30s in 30m (1.6%)'s<br>
differences, this maybe because of host rather than siteconfig. So I think that<br>
we can remove classes/siteconfig.bbclass since it is not as useful as expected.<br>
<br>
I've also tested remove siteconfig.bbclass (zlib and ncurses use it), there is<br>
no difference when "time bitbake core-image-minimal".<br>
<br>
The new implementation for glibc's site config is:<br>
- Add a recipe called <a href="http://glibc-site_2.28.bb" rel="noreferrer" target="_blank">glibc-site_2.28.bb</a>, find out the top 100 used functions<br>
and headers, and add them to site_config/funcs and site_config/headers. Please<br>
see patches' commit message on how to find out the top 100.<br>
<br>
- Make target/nativesdk recipes depend on glibc-site when inherit autotools.bbclass<br>
<br>
Here are the 2 patches:<br>
git://<a href="http://git.openembedded.org/openembedded-core-contrib" rel="noreferrer" target="_blank">git.openembedded.org/openembedded-core-contrib</a> rbt/glibc-site<br>
<a href="http://cgit.openembedded.org/openembedded-core-contrib/log/?h=rbt/glibc-site" rel="noreferrer" target="_blank">http://cgit.openembedded.org/openembedded-core-contrib/log/?h=rbt/glibc-site</a><br>
<br>
<br>
Please feel free to give your comments, and I will remove it if no objections.<br>
</blockquote><div dir="auto"><br></div><div dir="auto">I think if we cache all values then It probably will be faster but partial caching will give results like you see what will be interring to see is world build time differences when doing build from scratch </div><div dir="auto">But it indeed sounds a over optimization</div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
-- <br>
Thanks<br>
<br>
Robert<br>
-- <br>
_______________________________________________<br>
Openembedded-core mailing list<br>
<a href="mailto:Openembedded-core@lists.openembedded.org" target="_blank">Openembedded-core@lists.openembedded.org</a><br>
<a href="http://lists.openembedded.org/mailman/listinfo/openembedded-core" rel="noreferrer" target="_blank">http://lists.openembedded.org/mailman/listinfo/openembedded-core</a><br>
</blockquote></div></div>