<font face="Default Sans Serif,Verdana,Arial,Helvetica,sans-serif" size="2"><br><br><font color="#990099">-----Milton Miller II/Austin/IBM wrote: -----</font><br><br>>I have completed initial testing of my updates to initfs to allow <br>>running from RAM. Comments on the option names and functions are <br>>welcome. The branch is available at <br>><br>><a target="_blank" href="https://github.com/mdmillerii/openbmc/tree/wip-from-ram">https://github.com/mdmillerii/openbmc/tree/wip-from-ram</a><br>><br><br>The lastest update to Andrew's github comments is totally<br>untested but available at:<br><a target="_blank" href="https://github.com/mdmillerii/openbmc/tree/wip-from-ram-20160303">https://github.com/mdmillerii/openbmc/tree/wip-from-ram-20160303</a><br><br>I did not reorder the series because I believe the series is still <br>bisectable, as the options added at shutdown go from ignored to <br>parsed, and the options added for init go from default to explicit <br>to optimial.<br><br>The update script currently ignores arguments starting with the first <br>that does not include a - to allow the image names to be passed in <br>the future.<br><br>Following is a 1 line summary of how patches were changed.<br>The instructions remain as yesterday's email.<br><br>milton<br><br><br>shutdown: Complain if update images exist without update script<br><br>Expanded when / why the update may be missing and purpose of emiting message<br><br>update: Add command line parsing to clean or preserve saved files<br><br>It is currently called from init for jffs2 conversion, and will be called<br>from init and in the future at runtime with different constraints.<br><br>I clarified that the need for alternate behaviors are in the future init<br>patches in this series but that it already returns to shutdown (which was<br>in the last series).<br><br><br><br>update: Add options to skip the save and restore phases<br><br>Change to dosave and dorestore. Originally I only had no options<br>then decided to go to positive and negative options for future<br>proofing. As noted the logic was wrong in a prior revision of<br>the patch set.<br><br><br>update: Do not check if the whitelist is empty<br><br>removed double negative in title, expanded on future support.<br><br><br> update: Read whitelist from the /run/initramfs directory<br>expanded on why: both consistency and allowing future environments<br><br><br>update: Save whitelist files from ram or mounted filesystem<br><br>*** parent wrong<br><br>Significantly reworded commit message for what an why, including background.<br><br>init: Add a hook to download files<br><br>Factored out get_fw_env_var and commented with known issues<br><br><br></font><BR>