On Mon 2014-07-14 11:26:32, Tom Rini wrote:
On Wed, Jul 09, 2014 at 10:40:07PM +0200, Pavel Machek wrote:
Fix help text of ext2load and fatload to match code in fs/fs.c
Signed-off-by: Pavel Machek pavel@denx.de
diff --git a/common/cmd_ext2.c b/common/cmd_ext2.c index 5a4bcc1..2675ae8 100644 --- a/common/cmd_ext2.c +++ b/common/cmd_ext2.c @@ -45,7 +45,7 @@ U_BOOT_CMD( U_BOOT_CMD( ext2load, 6, 0, do_ext2load, "load binary file from a Ext2 filesystem",
- "<interface> <dev[:part]> [addr] [filename] [bytes]\n"
- "<interface> [<dev[:part]> [addr [filename [bytes [pos]]]]]\n"
Wait, what? [] is optional, <> is mandatory. With the rest of the related patches I see you fix a problem about assuming addr but we still have to pass along a dev at least, yes?
Do we?
fs/fs.c:
if (fs_set_blk_dev(argv[1], (argc >= 3) ? argv[2] : NULL, fstype)) return 1;
This tries to handle case where interface is specified but dev is not. Aha, and it seems to use bootdevice environment variable if device is not specified (and also "hostfs" does not use bootdevice).
SOCFPGA_CYCLONE5 # setenv bootdevice 0:1 SOCFPGA_CYCLONE5 # fatload mmc ** Bad device mmc 0 **
(In older u-boots, bootdevice is indeed mandatory. And clearly I should get recent u-boot with mmc support; working on that.).
IOW I think the patch is correct as-is.
Best regards, Pavel