On 11/8/2010 11:19 PM, John Rigby wrote:
I pulled this into packaged kernel and built it. With my existing config with is based on the OMAP4 Ubuntu kernel it gets to here:
[ 0.686676] audit: initializing netlink socket (disabled) [ 0.692352] type=2000 audit(1.257:1): initialized [ 0.771179] VFS: Disk quotas dquot_6.5.2 [ 0.775604] Dquot-cache hash table entries: 1024 (order 0, 4096 bytes) [ 0.785278] msgmni has been set to 898 [ 0.791320] alg: No test for stdrng (krng) [ 0.795684] io scheduler noop registered [ 0.799804] io scheduler deadline registered [ 0.804443] io scheduler cfq registered (default) [ 0.812774] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled [ 0.907135] serial8250.0: ttyS0 at MMIO 0x4806a000 (irq = 104) is a ST16654 [ 1.090606] serial8250.1: ttyS1 at MMIO 0x4806c000 (irq = 105) is a ST16654
But the kernel command line as set up by linaro-media-create has console=ttyO2 for panda.
That's correct - you need to use ttyS2 if you do not have CONFIG_SERIAL_OMAP enabled - this option was introduced along with the the new omap-serial driver that was merged in mainline only in the .37 merge window.
I haven't back ported this driver - if someone feels it's useful, it should be trivial to cherry-pick these patches. (ISTR it was a series of around 7 patches, and then there were a couple of fixes on top).
The consequence to end users of enabling this driver is that all OMAP on-SoC serial ports will be owned by this driver and will use ttyOx, and not-on-SoC ports (like on the zooms) will use ttySx as before (although they will be renumbered). Benefits include DMA support
I did not think this qualified as a fix for the .35 kernel, so I did not pick it up. Let me know if it should be.
- Anand