Yes, that's been suggested and MohammadAG already started tests, but no matter if it works (should) or not, it's no viable solution for the problem in the long term

I suspect it is because 4030 or 1707 chip generates an interrupt on VBUS high. I will test it tomorrow.

Yes, that makes sense

This does apply to some hubs, and some devices with external power supplies.
My belkin tetrahub, my Lexmark Z13 printer, for one, will not wake up until there is VBUS.

However - this is a fairly minor point - as there are many external powered hubs that don't bother to check.

is obvious that isp1707 or twl5030 generates interrupt as it starts the program selection of USB connection type. I have no time do dig on code so I bet on 5030.
i can't find anything special about host mode settings in isp1707 datasheet . isp 1707 is a dumb chip, a transceiver, with for convenience, integrated resistors.
ALL HIGH LEVEL of USB logic is at software level ( driver, rtos, ... ).
AFIK omap procesors include MentorGraphics USB Host, Function and OTG protocol stacks, hardware controller drivers and a long list of application-specific class drivers, integrated with the Nucleus Real-Time Operating System.
all we need is enabling host mode stack and the research must be done over this nucleus rtos

do not forget, in host mode across vbus it must be a 120 uF capacitor, and for shure inside n900 is not such a capacitor so i think we need a hacked adapter wich must contain 120uF capacitor but only when vbus is provided by bq24150 .

driving vbus is a secondary problem and host mode can be achieved without this by completely ignored vbus but after removal of bme and usb selection program ( if selection procedure is not included in bme ) and mask interrupt generated by vbus present
n900 as a host can do his job with only dp, dm and ground. vbus can be supplied externally with a permanent 5v power source wired only to peripheral
Vbus on regular PC stay on all the time and usb controller have nothing in common with vbus state.

Can you describe - what is the goal of it?

In my understanding it is just for smoothing VBUS level jitter, isn't it?

this value is specified in the standard.
without this capacitor when a peripherial is inserted may cause a vbus drop and can generate some software interrupts.
there is many voltage comparators with an eye on vbus !

OK, that actually means we may go with 'HOST' mode after insertion.

ACK
Though I'm not even sure this 120uF is mandatory for non-OTG devices. And I always stated we never will see OTG on N900, as it's not possible - if only for the micro-B receptacle.
Anyway VBUS supply for hostmode via bq24150 will work like a charm even without this 120uF - no need to externally retrofit.

Thanks to blueLED for the nice sum-up of all the fuzzy bits scattered all over the place before

from usb 2.0 specs:

"The maximum droop in the hub VBUS is 330 mV, or about 10% of the nominal signal swing from the function.
......................
The hub downstream facing port VBUS power lines must be bypassed (CHPB) with no less than 120 μF of
low-ESR capacitance per hub. Standard bypass methods should be used to minimize inductance and
resistance between the bypass capacitors and the connectors to reduce droop. The bypass capacitors
themselves should have a low dissipation factor to allow decoupling at higher frequencies."

if the vbus are not proper decoupling datalines transitions may have jitter out of specs.
this phenomen is visible on cheap computerboard with integrated graphics, horizontal sync signal may have jitter because insufficient decoupling and filtration of graphic "card" power