Thread: [Ftdi-usb-sio-devel] Interleaved write/read operations on ftdi_sio seem to cause an error

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

In our setup, we have a PC with its USB connected to a FTDI245BL chip
connected to a (hardware) FIFO: we (the PC) first write 128 bytes to the
FTDI, and then we expect to read back the exact same bytes from the
FTDI. That is: the fifo first stores all the 128 bytes we write, then
immediatly (a few nanoseconds later) sends them back to us, all this via
the FTDI of course.

I am running intrepid with kernel 2.6.27-9 on an UP IA32 (Pentium 4)
machine (2.4GHz). The problem I am reporting is also reproducible with
the ftdi_sio module recompiled from a vanilla kernel.org's 2.6.27.9
kernel. My kernel is not tainted.

So, our setup works "most" of the time: we can do thousands of such
write/read sessions of 128 bytes in a row. But at some point the read()
returns with a length of '0', even though we did have everything written
to our FIFO device (we checked with a digital analyzer), and sent back
by our FIFO device to the FTDI (likewise). So I guess the FTDI sent
everything back to the host (at least the debug below seems to confirm).

The problem seems to happen independently on the size of our FIFO
device / the size of the data packets we are using (we tried 32, 128,
256, 2048 bytes); however, the larger they are, the easier we get the
bug. Moreover, it seems related to the load on the PC: the more it is
loaded, the more likely this bug is going to show up. All this tends to
suggest this bug is some kind of driver (synchronization ?) problem.

When I load the ftdi_sio module in debug mode, I get the following
messages when the write/read *works*:
- --------------------------------------------------------------------
[72206.672669] ftdi_sio ttyUSB0: ftdi_write - length = 128, data = be bf
c0 c1 c2 c3 c4 c5 c6 c7 c8 c9 ca cb cc cd ce cf d0 d1 d2 d3 d4 d5 d6 d7
d8 d9 da db dc dd de df e0 e1 e2 e3 e4 e5 e6 e7 e8 e9 ea eb ec ed ee ef
f0 f1 f2 f3 f4 f5 f6 f7 f8 f9 fa fb fc fd fe ff 80 81 82 83 84 85 86 87
88 89 8a 8b 8c 8d 8e 8f 90 91 92 93 94 95 96 97 98 99 9a 9b 9c 9d 9e 9f
a0 a1 a2 a3 a4 a5 a6 a7 a8 a9 aa ab ac ad ae af b0 b1 b2 b3 b4 b5 b6 b7
b8 b9 ba bb bc bd
[72206.672744] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_write write returning: 128
[72206.673520] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_write_bulk_callback - port 0
[72206.688570] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_read_bulk_callback - port 0
[72206.688577] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_process_read - port 0
- --------------------------------------------------------------------

And when it *does not work* (ie. read() returns 0), we get:
- --------------------------------------------------------------------
[72206.720802] ftdi_sio ttyUSB0: ftdi_write - length = 128, data = d9 da
db dc dd de df e0 e1 e2 e3 e4 e5 e6 e7 e8 e9 ea eb ec ed ee ef f0 f1 f2
f3 f4 f5 f6 f7 f8 f9 fa fb fc fd fe ff 80 81 82 83 84 85 86 87 88 89 8a
8b 8c 8d 8e 8f 90 91 92 93 94 95 96 97 98 99 9a 9b 9c 9d 9e 9f a0 a1 a2
a3 a4 a5 a6 a7 a8 a9 aa ab ac ad ae af b0 b1 b2 b3 b4 b5 b6 b7 b8 b9 ba
bb bc bd be bf c0 c1 c2 c3 c4 c5 c6 c7 c8 c9 ca cb cc cd ce cf d0 d1 d2
d3 d4 d5 d6 d7 d8
[72206.720875] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_write write returning: 128
[72206.721639] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_read_bulk_callback - port 0
[72206.721645] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c: (this
is ok on close) nonzero read bulk status received: -75
[72206.721652] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_write_bulk_callback - port 0
[72206.856067] hub 4-0:1.0: port 2 disabled by hub (EMI?), re-enabling...
[72206.856076] usb 4-2: USB disconnect, address 43
[72206.859191] ftdi_sio 4-2:1.0: device disconnected
[72206.859712] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c: ftdi_close
[72206.859722] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_shutdown
[72206.859748] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
ftdi_sio_port_remove
[72206.859752] /tmp/linux-2.6.27.9/drivers/usb/serial/ftdi_sio.c:
remove_sysfs_attrs
[72206.860121] ftdi_sio ttyUSB0: FTDI USB Serial Device converter now
disconnected from ttyUSB0
- --------------------------------------------------------------------

When it fails, I always had the write callback called /after/ the read
callback (as is the case above). But I am not sure it does NEVER happen
when everything is fine too. I can send the full traces to whoever wants
to analyze them.

Code-wise, I do have a problem figuring out what this "-75" return value
stands for. Maybe it could help, maybe it should be ignored in some
cases. Looking at the code, it seems that the snippet involved is:
- --------------------------------------------------------------------
static void ftdi_read_bulk_callback(struct urb *urb)
{
[...]
        int status = urb->status;
[...]
        if (status) {
                /* This will happen at close every time so it is a dbg
not an
                   err */
                dbg("(this is ok on close) nonzero read bulk status
received: %d", status);
                return;
        }
[...]
}
- --------------------------------------------------------------------

So, to summarize my thoughts: could it be that the write/read callbacks
are not allowed to get interleaved ? Or is it something else ? What does
a status of -75 mean ?

I can provide any further information on request.
Thanks a lot for any help !
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkmAE/8ACgkQld7vhusVrCEhkgCfbOO0m4T00Qsc8FGHB31LSqee
Qn0AoIyz7n2WS+XIqKtdWkCM2Qu2P1Wb
=bWPC
-----END PGP SIGNATURE-----

Thread: [Ftdi-usb-sio-devel] Interleaved write/read operations on ftdi_sio seem to cause an error

ftdi-usb-sio-devel