On Fri, Jun 23, 2017 at 6:50 AM, Alexey Brodkin <alexey.brod...@synopsys.com> wrote: > Hi Khem, > > On Fri, 2017-06-23 at 06:46 -0700, Khem Raj wrote: >> On Fri, Jun 23, 2017 at 6:40 AM, Richard Purdie >> <richard.pur...@linuxfoundation.org> wrote: >> > >> > Hi, >> > >> > On Fri, 2017-06-23 at 15:23 +0200, Mylene Josserand wrote: >> > > >> > > We are currently adding to Yocto project the Synopsys ARC CPU >> > > architecture, which is only available with uClibc library. >> > > >> > > We currently have a layer which adds the support of this >> > > architecture >> > > and uClibc support as it is, now, removed from OE-core. >> > > >> > > Many configurations/fixes are needed in OpenEmbedded-core's recipes >> > > to >> > > build successfully with this C library (library's dependencies, >> > > autotools configuration, etc). >> > > >> > > We wanted to send patches to fix them but it seems that all uClibc's >> > > references are now removed from OE-core, even in recipes: >> > > >> > > https://urldefense.proofpoint.com/v2/url?u=https-3A__patchwork.openembedded.org_patch_140943_&d=DwIBaQ&c=DPL6_X_6JkXFx7AXWqB0tg&r=lqdeeSSEes0GFD >> > > Dl656eViXO7breS55ytWkhpk5R81I&m=vXM6W0k4FOex1pzJnHfSDtrgw08G-uIiKgZJinDWFtI&s=wd0w1NqlVaNUjpgkDUqxfK2izS-a-XotxpeCqt3Mcro&e= >> > > https://urldefense.proofpoint.com/v2/url?u=https-3A__patches.openembedded.org_patch_140906_&d=DwIBaQ&c=DPL6_X_6JkXFx7AXWqB0tg&r=lqdeeSSEes0GFDDl >> > > 656eViXO7breS55ytWkhpk5R81I&m=vXM6W0k4FOex1pzJnHfSDtrgw08G-uIiKgZJinDWFtI&s=4Y1JNumo1mAFnPDzX2CaFZw9gfkSF2TJa_vIGO8A9yA&e= >> > > >> > > What is the status of it? Remove entirely uClibc from OE-core? >> > > Could you give us advice on how to handle uClibc? Should we create >> > > a >> > > "meta-uclibc" layer which adds support of it + bbappend for all the >> > > OE-core's recipes that currently fail because of uClibc's >> > > misconfiguration? >> > > >> > > I guess that there is no plan to add uClibc's configuration in OE- >> > > core's >> > > recipes (even without uClibc itself), right? >> > >> > We dropped uclibc from OE-Core around a year ago in favour of musl >> > since it seems to work for all the use cases we've been aware of from >> > our users and nobody was willing to maintain it. >> > >> > Since then the uclibc overrides and other pieces in core have bitrotted >> > to the point it was clear nobody was actively using or maintaining >> > them. I therefore proposed some patches removing them, those were >> > agreed by several developers and no objections were seen. We therefore >> > cleaned up the remaining pieces. >> > >> > I believe that we have enhanced the extension mechanisms within the >> > core such that a meta-uclibc layer would be possible now, even for the >> > site files and insane.bbclass mappings and so on. It would need to add >> > some of the extensions removed from the core and then it should be >> > buildable again. >> > >> > If you can't make anything work from the layer let us know and we can >> > try and figure out a way to solve that. I don't think the uclibc pieces >> > would be coming back into OE-Core any time soon though as whilst it >> > does have some niche markets left (e.g. arc), I can't see it being >> > widely used by the majority of OE users/arches. >> > >> > (As an aside I haven't looked at it at all but does musl support arc?) >> > >> >> I dont see arc support in upstream musl, It could good if it was there >> since that could unleash whole OE on arc architecture. We can do >> successful world builds on musl much like glibc now a days. > > Indeed there's no musl port for ARC [yet]. > But we're almost done with glibc port and it looks like we'll need to switch > to glibc for OE/Yocto which I believe is doable. > > I guess glibc is still the first class-citizen in OE, right? >
yes that is right. > -Alexey -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
