A race condition exists in SMD where incoming messages might be processed before we've finished executing the rpmsg device's probe function. The driver's callback function will in this case be unable to handle the incoming message and might return an error. Using the announce_create ops we can invoke the handler for incoming messages once again after probe returns, solving this issue. With SMD and QRTR this shows a high failure rate, but there are (at least theoretical) similar issues in glink and virtio-rpmsg, so this needs to be further investigated. Bjorn Andersson (3): rpmsg: smd: Fix container_of macros rpmsg: Only invoke announce_create for rpdev with endpoints rpmsg: smd: Use announce_create to process any receive work drivers/rpmsg/qcom_smd.c | 22 ++++++++++++++++++++-- drivers/rpmsg/rpmsg_core.c | 2 +- 2 files changed, 21 insertions(+), 3 deletions(-) -- 2.16.2 -- To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
