On Wed, Feb 1, 2012 at 11:14 AM, Scott Wood scottwood@freescale.com wrote:
On 02/01/2012 04:24 AM, Wolfgang Denk wrote:
Dear Tom Rini,
In message 1328047438-26294-1-git-send-email-trini@ti.com you wrote:
From: Matt Porter mporter@ti.com
Adds support for loading U-Boot from UART using YMODEM protocol. If YMODEM support is enabled in SPL and the romcode indicates that SPL loaded via UART then SPL will wait for start of a YMODEM transfer via the console port.
Does it really make sense to add all kind of bloat to the SPL?
What originally has been designed to be minimal code now start to pull in all kinds of other crap,
What's the alternative -- another name and separate makefile infrastructure for non-minimal initial loaders?
It's important that SPL be *able* to produce very small images when needed, but its purpose has grown beyond that.
I talked with Wolfgang a little on IRC on this, and we didn't reach a conclusion nor come up with an alternative, but I guess the problem is we need to figure something out, now. We've hit a roadblock on "allow UART loading" and regret over keeping MMC+VFAT working. How bad are things going to get when patches pop up for USB (AM33xx ROM allows for RNDIS USB, iirc) ?
Maybe we need to separate out the systems which are very resource constrained from the ones that aren't, share code where we can and make that code obviously marked as such.
that has never been designed to run before relocation.
At least some SPLs do relocation.
Or don't need to exactly.