On Tue, 20 Nov 2012, Wolfgang Denk wrote:
Dear Lee Jones,
In message 1353422034-28107-5-git-send-email-lee.jones@linaro.org you wrote:
Here we add boottime tags to the start of the main loop and just before the opportunity to break into the u-boot shell. This will provide a more verbose bootgraph when viewed within debugfs.
Assuming we would take this route - then why do we need to add new boottime_tag() entries - is there anything wrong with the existing show_boot_progress() code? Did you consider using this instead? If so, why did you not use it?
No, I didn't know it existed. Basically I've taken responsibility to upstream someone else's driver. It's more of a kernel thing, but it requires boottime information from u-boot too, as the intention is to cover full system booting, rather than the kernel-only tracing mechanisms which already exist.
I've just taken a look at show_boot_process() though. It doesn't appear to be suitable for what we require. Each tag needs to be individually identifiable, and I'm not sure how you would achieve this if we were to call back into a single function which would do the tagging.