On Wed, Jun 30, 2021 at 03:21:46AM +0100, Bryan O'Donoghue wrote: > On 29/06/2021 22:57, Bryan O'Donoghue wrote: > > On 29/06/2021 21:30, Bjorn Andersson wrote: > > > I liked this, and it worked when I tested it, but iirc it suffered from > > > the problem that the core's probe may or may not have finished > > > successfully at the time that of_platform_populate() returns. > > > > > > But fixing this problem would save us quite a bit of headache. > > > > OK. > > > > I will take a look at resurrecting the old patches either fixing the > > probe order - or perhaps using something like Wesley's role-switch to > > have dwc3 core optionally trigger dwc3-qcom > > > > Binding tcpm into &usb_1_dwc3 instead of &usb_1 > > > > --- > > bod > > So here's a potential way forward. Not technically breaking my "no patches > at 3am rule" > > Basically we can fix the probe order problem if we have dwc3 drd call into > dwc3-qcom. > > In order to make that not be a problem for all non qcom platforms - use a > function with weak binding in drd - which a wrapper - in this case the qcom > wrapper can over ride.. I'm afraid I'm not too familiar with weak symbols. Would that still work if dwc3 & dwc3-qcom are built as modules? Also is it supported with Clang/LLVM? Jack -- The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
