Hello Kevin, Tuesday, July 3, 2007, 3:11:03 AM, you wrote:
> On Mon, Jul 02, 2007 at 11:56:47PM +0300, Paul Sokolovsky wrote: >> Here's updated per feedback and improved linload implementation able >> to handle 20Mb+ initrds. This implementation was used to produce a >> LiveRamdisk for hx4700, >> http://pfalcon-oe.blogspot.com/2007/06/first-angstrom-liveramdisk.html >> For 25 downloads, there 0 issue reports, so that must be a good sign >> ;-). >> >> So, I'd like to decide now if it's ok to replace old implementation, >> or better provide alternative, and commit it. > Thanks Paul. > I have a couple of comments: > 1 - As before, if you implement the concat method, I'd prefer your > changes be merged into haret.cpp instead of linboot.cpp. (If the > signature is present run the linked in script, otherwise boot > haret normally.) Well, this adds more codepaths for startup... But Ok. > 2 - The wince prototypes should go into the appropriate header (looks > like pwindbas.h). I'll see what I can do, though if you mean cegcc, let that be on TODO for now ;-). > 3 - I'm leery of the SetSystemMemoryDivision call - how about adding a > new haret command (eg, SETMEMDIV). That way the linked in script > can call it if it chooses to. It also allows regular haret users > to play with the setting. Well, that call is what it took me to get boot of ~23Mb file on hx4700 without it complaining about lack of memory. That calls is completely adhoc for this purpose, and/or requires calculation of arguments which is hard to do in script, so that hardly help. > BTW, what is the md5 stuff? That's hook for future, but we'll need to check integrity of stuff user loads. Other extension I have in mind is to pack several kernels and select them based on machine ID. (Yes, that's rather adhoc, but would be a good stub until we'll tackle at modular kernel). > -Kevin -- Best regards, Paul mailto:[EMAIL PROTECTED] _______________________________________________ Haret mailing list [email protected] https://handhelds.org/mailman/listinfo/haret
