Time taken to change USB state

[Nico3]

I am running USB in Mass Storage host mode. On connecting the device, USB FLASH stick, it takes around 20 seconds to change the Appli_Sate from APPLICATION_START to APPLICATION_READY in usb_host.c file. 

It is normal ? Can the time be reduced ?

Thanks

[FBL]

Hello @Nico3 

Are you sure you are not printing to VCP/ UART or any terminal while debugging? It seems like an amount of time taken for printing debug information.

[Nico3]

Hi

Please note I am not printing anything. I also commented my remaning code and just checked the change in state on inserting the USB stick. It takes ~30 second to go from APPLICATION_START to APPLICATION_READY.

I checked both on STM32F4 discovery as well on board designed by me. It is coming same ~30 seconds.

The USB stick is 32 GB, SCANDISK & FAT formatted.

[Pavel A.]

Then the contrary, add some prints/traces to understand where it spends its time.

Do not set breakpoints because USB requires realtime reaction from the device.

Of course, 30 sec. delay is not normal.

[Nico3]

Please suggest what should I trace. 

[AScha.3]

Hi,

i never tested , but if you want, i could try to catch the time : Appli_Sate from APPLICATION_START to APPLICATION_READY

on my H743 , at 200M core (so not too far away from your F4 ?? cpu ).

[Nico3]

please check if you can.. 

I used 32 GB flashstick and formatted in FAT mode. 

[Pavel A.]

@Nico3 Trace:

- where the host reads various device (the USB stick) descriptors and strings

- where the host decides that the MCS device is fully detected and starts sending class-specific requests

- where the host successfully receives data from MSC device

- where errors and timeouts  are detected in the above steps, and the app retries them

[AScha.3 · Accepted solution]

ok, so via swd + DWT counter ->

about 9,2 ms from inserted to fatfs mounted ok;

is 64 GB (noname chinese..) USB stick, exfat , 50GB free.

btw

How often you call the  MX_USB_HOST_Process(); ?

I call it in main loop about every some 5us ....nothing else to do now, waiting for first command.

So this might be maximum speed, that can be achieved .

ed

+

Just for fun - had the idea, to try an old SanDisk 32GB stick, fat32,  also, got ->

about 15ms insert to mounted. Ok for this piece of plastic shi.t , i think.

[Pavel A.]

> about 9,2 ms

> about 15ms insert to mounted. 

This is what should be.

Time individual MSC class requests to find where it wastes time. The capacity in bytes vs block numbers look suspicious.