Version 189: does not power USB hub

[Barry Landy]

Repeated booting of 189; every time the green "on" lights on my USB hub (from which my keyboard and mouse operate) fail to come on.

If I switch to 189 during a session everything works OK so it seems to be a startup problem.

Everything works fine in version 186.

FWIW the hub is from Stapes with 7 ports.

[Barry Landy]

Grrr... I meant "Staples" of course

[Lars Erdmann]

see doug.zip, Ticket #45.

[Barry Landy]

I know its not your software but just to report that I set the cc field in the header page (as well as having my email in the preferences) and I still did not get an email with your comment.

Will try doug.zip

[Barry Landy]

doug.zip does not help for me.

[Barry Landy]

I have now had time to do more elaborate testing (takes time because of all the reboots!). This is complicated by the fact that even in what appears to be a stable system the USB mouse/Keyboard do not always come up and I have to (usually) power cycle the machine. Sometimes reconnecting the hub helps.hub lights come on but mouse/keyb not working

WinXP does not display this behaviour and interestingly enough neither does BootAble (which is running version 181 of the USB suite)

Experiments (note I have called the "doug.zip" suite 189a)

1) 189 with USBEHCD frrom 186: works 2) 189 with USBEHCD frrom 189a: hub lights come on but mouse/keyb not working 3) all 189a: ditto 4) 189a with USBEHCD frrom 186: hub lights come on; mouse OK BUT keyb not working (I dont really understand this one.....)

5) I then reverted to (1) and just to show how perverse things can be, it did not work, even after reconnecting the hub. I power cycled, booted the maintenance partition (as a test;it was OK) and then rebooted the main ECS21 partition. Mouse and Hub were OK but not keyboard but the keyboard came back after reconnect.

Hope this is helpful!

[Barry Landy]

Having read your item in COMP.OS.OS2.MISC I have attached the output of USBDock (2 zips, both the log and the bin format as I was not sure what is wanted).

For the time being I have reverted to 186 (initialising the USB connected mouse/keyboard too unreliable)

I have also moved the BASEDEV=USB..... lines from near the bottom of the CONFIG (placed by USBCFG) to the start of the BASEDEVs in case it makes a difference.

You ask for traces. With this problem I am (almost always) unable to use the mouse or keyboard so could not issue TRACE commands. Please let me know what data I could usefully get for you.

[Lars Erdmann]

Try doug2.zip from ticket #45

[Barry Landy]

because of the lack of notifications I only just saw this. I am away from that system at present and will try when I return.

[Barry Landy]

With all 4 files from doug.2.zip the system fails to boot; it stops way down the boot sequence. At that point the USB hub has green lights (but only 3 out of 4).

I will try with alt-F2 to see where it is stopping.

[Barry Landy]

Alt-F2 shows that the last command issued from CONFIG.SYS is the call to MPTSTART.

After that there is only the call to DMT and then the initialisation of the screen graphics neither of which happened.

This effect is repeatable

[Barry Landy]

I have replaced usbehci.sys from ehci.zip as in ticket 45 (AB willis) with the remaining usb?hcs.sys and usbd.sys from doug2.zip and mouse and keyboard come up at boot time and work.

(On at least one boot I had to replug the keyboard to get it to work which in the context is a minor inconvenience.....)

[Barry Landy]

Now using Version 190 from Lars (and thank you very much)

Situation is more or less as in the previous post (using AB Willis' fix).

The hub powers up during boot. However I almost always have to replug either the keyboard or the mouse (depending possibly on which port they are plugged to) to get both to work.

This is a minor inconvenience but could be a game breaker if I needed to use the keyboard during the boot process (I have "pauseonerror=no to avoid that"

Is there any tracing I could usefully do?

[Barry Landy]

I did the test of removing pauseonerror=no and the keyboard is not initialised early enough to respond any error message caused by CONFIG.SYS.

[Lars Erdmann]

Updated USBD.SYS: remove delay loops for "HubAckStatusChanges"

[Lars Erdmann]

Try blandy.zip for USBD.SYS and report back if it improves the situation.

[Barry Landy]

yes, that fixes the need to replug the keyboard and/or mouse.

The availability of keyboard during CONFIG processing is presumably a totally different issue.

[Barry Landy]

Sorry: it does not consistently fix the need to replug. So far in 2 boots it has worked once and failed once.

[Lars Erdmann]

Are the mouse and keyboard plugged into the external USB hub ? Or are they directly plugged into the computer ?

[Barry Landy]

External hub. (because of the difficulty of getting them to work I wanted them where I could see them :-) )

I will try with the mouse in the hub and the keyboard directly attached.

[Barry Landy]

WIth the keyboard in directly connected and the mouse in the hub, the keyboard came up every time but the mouse needed "prodding" once. Ecouraged by this I now have both the mouse and keyboard plugged in to the back of the machine and it seems to work. I will only post again if I have problems.

[Lars Erdmann]

Updated USBD.SYS: clear feature change flags on external hub before powering on ports

[Lars Erdmann]

Please try blandy2.zip for USBD.SYS. The problem is that USBD.SYS is not properly managing external USB hubs.

[Barry Landy]

I assume you would like me to test with mouse/keyboard plugged into the hub?

[Barry Landy]

worked fine (one trial) with mouse/keyboard plugged into the hub. I will keep trying...

[Lars Erdmann]

Updated USBD.SYS: doing a TT reset before we power on any hub ports

[Lars Erdmann]

Yes I want you to have mouse/keyboard plugged into the external hub.

If you still run into problems, try blandy3.zip. In fact, try blandy3.zip in any case and tell me for both: blandy2.zip and blandy3.zip if you run into any problems with your external hub.

[Barry Landy]

blandy3 works fine as well as blandy2.

[Lars Erdmann]

If possible, try both versions for the next 2-3 weeks or so. If it still works by then, please close the ticket.

[Barry Landy]

I currently have version 3 in my production system and version 2 in my maintenance system, both working fine. Will keep trying.

[Barry Landy]

Something (good!) that has snuck up on me without my noticing! I have a permanently attached external USB drive. This never used to reliably becaome availableon a cold boot (I needed to do a refresh removable media). After the recent changes it comes up without intervention apparently 100%.

[Barry Landy]

One failure (in many attempts) to initialise the keyboard (version blandy3)

[Barry Landy]

Turned out there were 3 failures in a row and then it became OK again.

PS JUst to note that I am going to change my mobo to one which has a PS2 port. I will keep the mouse on USB and can of course switch the keyboard to USB occasionally to test.

The critical thing (for me, and worth the money to change the mobo) is that I will definitely be able to install ECS.

[Barry Landy]

Well I tried to change the mobo and it didnt work so I am back on the same mobo!

This is the first time I have booted with anchieri.9.zip and while it is wonderful at writing to a JFS memory stick (my other problem) unfortunately it doesnt start the USB attached mouse or keyboard. anchieri.4.zip works fine for mouse and keyboard (but not for writing JFS memory sticks!!)

win some and lose some... !

[Lars Erdmann]

Get 10.191 from Hobbes.

[Barry Landy]

Both problems solved with 10.191: USB hub comes up; mouse&keyboard work; also the repeated frequent write is fine (problem 55)

[Barry Landy]

This ticket can now be closed