Poke5!

[Renate]

That's strange. People always talk about "voodoo" issues with USB2 vs USB3.
I'm not dismissing them, but I've never gotten a handle on any of them.

Also, before you have a loader loaded the interface is very tenuous.
Qualcomm put the cart before the horse and your device is the client.
After the Firehose loader goes in your device is the server.
So after a reset you can not disconnect the cable without doing a full reset again.

[Logseman]

In this case, I can reproduce the result with the ports every time. Here's a video where I swap the cable from a USB3 port (where it says "Could not open device") to a USB2 one (where edl /l works as intended).

Whether I reset the device in between the port swaps does not change the result.

[Renate]

Hmm, some USB2 devices get cranky on a USB3 connection.
But that would mean that the Poke5 is USB2? That would be depressing.
I always use my Poke3 on USB3 and I can verify that it is talking USB3.1 normally and USB3.0 on EDL.
Try this utility while plugged into USB3.
See what the "bcdUSB" says. Also what "Open pipes" says.

[Logseman]

I don’t think any e-reader has a reason to put a higher protocol than USB2, because they’re devices which handle very small files so there’s no advantage to higher transfer speeds. Additionally, more wattage is not going to matter in things that are charged every fortnight.

What’s the name of the Utility I’m supposed to download? The windows WinUSB one?

[Renate]

Sorry, it's in the URL fragment, Microsoft UsbView.exe

[Logseman]

When connected in normal powered-on state, on both USB3 and USB2 ports:

bcdUSB:             0x0210
Open Pipes:              5

When connected in EDL mode, also regardless of the port:

bcdUSB:             0x0210
Open Pipes:              2

[Renate]

Thanks.
I'll admit that the difference between USB3 and USB2 isn't that big even for people like me who use AdbSync.exe to synchronize thousands of documents.

But that may be why it has balkiness on USB3.

[Logseman]

After tests with two additional cables, including the one supplied by Onyx, I'm on the same spot. Either it's "Could not open device" or "Could not write device" whenever I want to flash the QCS_AGATTI loader.

[Renate]

"Could not open device" might mean that you don't have the correct Windows driver loaded or that you have some Qualcomm driver loaded or that something is trying to use your device already.

"Could not write device" depends when and under what circumstances it occurs.

As I don't have a Poke5 or proof that a loader for it exists, I can't say much.

You can try my test utility

C:\>winusb 05c6 9008
Found generic device at 05c6/9008
Path: \\?\usb#vid_05c6&pid_9008#8&1265ce96&0&4#{a5dcbf10-6530-11d2-901f-00c04fb951ed}
In bulk endpoint 81
Out bulk endpoint 01
Everything ok

[Logseman]

In the port where it recognised edl /l and the error is "Could not write device" it says this:

Found generic device at 05c6/9008
Path: \\?\usb#vid_05c6&pid_9008#5&1b87c58d&0&4#{a5dcbf10-6530-11d2-901f-00c04fb951ed}
In bulk endpoint 81
Out bulk endpoint 01
Everything ok

In other ports, where the error is "could not open device" it shows this:

Found generic device at 05c6/9008
Path: \\?\usb#vid_05c6&pid_9008#5&1b87c58d&0&3#{a5dcbf10-6530-11d2-901f-00c04fb951ed}
Could not convert raw USB to WinUsb (00000057)

[Renate]

Originally Posted by Logseman

In other ports...

Where it gives you that 57 error try to locate the device in Device Manager and look for the driver details.
It should say whatever_autogen. Use Zadig again and see if it needs a driver.

Is this Windows 11 or something?

In any case, the port that it works is the port you should be using.
The Sahara protocol goes back and forth and when Sahara doesn't like anything it refuses to receive, hence the "Can not write device".

After every unsuccessful attempt to load a loader you have to do a full device reset.
That's not me, that's the Sahara.

[Logseman]

In this case I'm using Windows 11 indeed, namely the Pro version.

Apparently the drivers that Zadig gives are indeed in a per-port basis in this case. Now that I located the device and installed a new WinUSB driver via Zadig, I can use one of the front ports that didn't work before.

Unfortunately, this doesn't change the known outcome once I attempt to load the Agatti loader, even when I'm resetting the device and sending it back to EDL mode after each failure:

Sending 0014d0e100000000_d40eee56f3194665_FHPRG.bin 100% Ok
Waiting for Firehose... No response, poking...
Could not write device

[Renate]

Home many different Firehose loaders did you try in all?
Did any of them stop loading at 7-8%?

[Logseman]

I tried the three loaders you've linked so far. None of them stop at 8%: the Agatti one gets loaded to 100% with the "could not write" error, and the other ones both stop at 0% with the "error 12" message quoted here.

[Renate]

Well, it seems that you (we) are out of luck.

Here's another test that you can do.
It won't work any better but it will tell us something.
The test is whether this loads 100% or only 8%.
It's "bad" because I put in a spelling mistake, re-hashed it but didn't re-sign it.
Just rename it from .prc to .bin

I went looking for an update.upx for the Poke5 but they haven't released one yet.