[Nut-upsdev] support for newer mustek powermust model 06da:0003

[Timo Lindfors]

Hi,

Arjen de Korte <nut+devel at de-korte.org> writes:
> I doubt it, but just in case I padded the commands to 8 bytes and
> removed flushing the buffer. Could you try again with version 1157 from
> the trunk?

$ sudo strace -o agiler-svn1157.strace -s4096 -f /lib/nut/megatec_usb -a myups -DDDDD -x vendorid=06da -x productid=0003 -x subdriver=agiler
Network UPS Tools 2.3.0-1157M - Megatec protocol driver 1.5.8 [megatec_usb]
Carlos Rodrigues (c) 2003-2007

Serial-over-USB transport layer for Megatec protocol driver [megatec_usb]
Andrey Lelikov (c) 2006, Alexander Gordeev (c) 2006-2007, Jon Gough (c) 2007

debug level is '5'
Checking device (0000/0000) (005/001)
- VendorID: 0000
- ProductID: 0000
- Manufacturer: unknown
- Product: unknown
- Serial Number: unknown
- Bus: 005
Trying to match device
Device does not match - skipping
Checking device (06DA/0003) (004/002)
- VendorID: 06da
- ProductID: 0003
- Manufacturer: unknown
- Product: 87XXUPS
- Serial Number: unknown
- Bus: 004
Trying to match device
Device matches
failed to claim USB device, trying 2 more time(s)...
detaching kernel driver from USB device...
trying again to claim USB device...
Starting UPS detection process...
Asking for UPS status [Q1]...
Q1 => FAILED [timeout]
Asking for UPS status [Q1]...
Q1 => FAILED [timeout]
Asking for UPS status [Q1]...
Q1 => FAILED [timeout]
Asking for UPS status [Q1]...
Q1 => FAILED [timeout]
Asking for UPS status [Q1]...
Q1 => FAILED [timeout]
0 out of 5 detection attempts succeeded (minimum required 3)
Megatec protocol UPS not detected.

doesn't indeed seem to work so it must be something else then. I
tested again with my modified agiler driver in nut 2.2.0 and the UPS
seems to be still alive. Strace output under svn 1157 is available at

http://iki.fi/lindi/usb/mustek2-snoops/agiler-svn1157.strace

The only obvious between agiler2 (my "fork") and svn head I can see
with

diff -u <(grep ioctl agiler2-mustek2.strace|cut -d' ' -f2-|sed 's at 0x........@@g') <(grep ioctl agiler-svn1157.strace|cut -d' ' -f2-|sed 's at 0x........@@g')

is

-... USBDEVFS_SUBMITURB, {type=1, endpoint=129, status=..., flags=0, buffer_length=8 ...
+... USBDEVFS_SUBMITURB, {type=1, endpoint=129, status=..., flags=0, buffer_length=256 ...

(I replaced unnecessary stuff with "..."s.)

Where does that 8 vs. 256 difference come from? -- It seems that
ser_flush_io(int fd) tries to read 256 bytes. I changed this to 8
bytes but that alone does not seem to help. I now see one 8-byte
SUBMITURB and 5 256-byte SUBMITURBs.

Where do the five other SUBMITURBs come from? -- apparently
ser_get_line() calls agiler_get_data with
buflen=256. get_agiler_data() in nut 2.2.0 had a for-loop to read the
data in 8-byte pieces -- maybe that is still needed?


best regards,
Timo Lindfors