Vanilla mainline u-boot with the same tag (v2013.01-rc1) does not have this problem, so something specific to the Linaro u-boot configuration is causing this.
u-boot versions since v2012.07 have removed the clock initialization for lots of clocks that u-boot should not have been initializing and which were causing kernel PM issues due to IP blocks left in improper states.
I don't know where the linaro u-boot repo is for panda, but Is suspect that one of the diffs vs mainline u-boot is that the "enable all clocks" option has been re-enabled in include/configs/omap4_common.h:
#define CONFIG_SYS_CLOCKS_ENABLE_ALL
This option is known to break OMAP4 PM and should not be enabled.
Vanilla mainline u-boot with the same tag (v2013.01-rc1) does not have this problem, so something specific to the Linaro u-boot configuration is causing this.
u-boot versions since v2012.07 have removed the clock initialization for lots of clocks that u-boot should not have been initializing and which were causing kernel PM issues due to IP blocks left in improper states.
I don't know where the linaro u-boot repo is for panda, but Is suspect that one of the diffs vs mainline u-boot is that the "enable all clocks" option has been re-enabled in include/ configs/ omap4_common. h:
#define CONFIG_ SYS_CLOCKS_ ENABLE_ ALL
This option is known to break OMAP4 PM and should not be enabled.