*** Bug 8784 has been marked as a duplicate of this bug. ***

Please attach the output of

WINEDEBUG='+comm' wine application.exe

with wine 0.9.22, and with wine 0.9.23.

Created attachment 7016 [details]
WINEDEBUG + comm output for 0.9.22

output repeats over and over, this is just a snapshot of the output during
running

Created attachment 7017 [details]
WINEDEBUG + comm output for 0.9.23

Created attachment 7018 [details]
WINEDEBUG + comm output for 0.9.24

Created attachment 7032 [details]
ignore some errors

I can't even find where in the kernel source it returns EINVAL for TIOCGICOUNT
:-(.

It could also be that it also failed on wine <= 0.9.22, but wine just ignored
the errors. This patch attempts to do the same, try it on a recent version of
wine and see if it helps.

Ok, I've tried the patch but the program still does not run correctly and I am
unable to program PICs using the device. It seems to be the same behaviour.

I did try to capture the output from the debug but it pours out very rapidly and
quickly over runs the available history in konsole terminal before i can
terminate the program.
I tried various things with | > tee and trying to capture the terminal output to
a file to attach here but couldn't get it to work... any hints welcome :o)

I've reverted to 0.9.22 for the moment.

Thanks for the help so far.

When you get the chance, post the output of
WINEDEBUG='+comm' wine app.exe &> log.txt
with 0.9.22 and with a recent version that doesn't work.

Bug confirmed. I posted a duplicate in bug 10699

*** Bug 10699 has been marked as a duplicate of this bug. ***

This should be fixed for 1.0, especially being a regression.

The write was tested with minicom on the receiving end. It works fine. Reception does not work. 

This confirms first post's findings

Created attachment 10999 [details]
pirelli dp-l10 mobile ft utility with 2.6.24 cp2101 usb serial driver

the cp2101.c driver and most others do not have the TIOCGICOUNT ioctl, too, see att.
my mobile ft app wants it, too, see att.
http://bugs.winehq.org/attachment.cgi?id=10999
some drivers have it, now trying to copy it to cp2101.c.

tom

cp2101.c investigated: still based on the primitive silabs linux driver: extension for counters for TIOCGICOUNT maybe possible but extension for TIOCMIWAIT would need synchronization, adaption to the actual kernel-api and so a complete rewrite from scratch in the style of io_ti.c driver e.g. so NOTOURBUG for wine, driver issue.

If I understand correctly and this is a driver bug, is it possible to detect this case and have Wine print a notice informing the user that the driver is broken, please contact XYZ and let them know you want it fixed?

commit d990c0d8e50c5eaabac8b855b9511cbc42d783bd
Author: Eric Pouech <eric.pouech@wanadoo.fr>
Date:   Sat Sep 30 21:20:41 2006 +0200

    ntdll-kernel32: WaitCommEvent.

    - implemented IOCTL_SERIAL_WAIT_ON_MASK for DeviceIoControl
      on serial lines in ntdll
    - now using thread pool (instead of simple thread) for the
      background operations (this should help some high load
      situations)
    - used this to implement WaitCommEvent on top NtDll functions
    - in kernel32, removed now the no longer used termios/ioctls...
      for comm devices

I have also tried it with my own Serial USB. No luck until now. I'm going to post some more information when I'm able to.

Isn't Bug #9356 a duplicate of this one?

I have this same problem with Huawei E220 Modem's sofware (Tim Web Movel - Brazil) (I use modprobe usbserial in linux, the device is /dev/ttyUSB0).
If there's any patch, I could test it.

This should be fixed now.

Closing bugs fixed in 1.0-rc1.

I have still this bug, tested with wine 1.0rc1 and Prolific Technology, Inc. PL2303 Serial Port. I'm attaching my log (WINEDEBUG +comm).
The application works but does not communicate with the serial device. It's like if there was no device at the other side. (but still tested with cutecom and it is, in fact, receiving data ).
You'll have to reopen the bug! :-(

Created attachment 12957 [details]
Wine USB Prolific serial port debugging (+comm)

Here's the debugging info.

Not fixed. Reopening.

Hi all.

Is it really necessary reopening this bug?

I'm using a pl2303 driver (provided with kernel 2.6.21-5) to communicate with my mobile phone using the Windows software "LG Contents Bank" via Wine.

It's working correctly. I'm able to send and receive data using the software.

To make it works, I've linked com1 to /dev/ttyUSB0 in dosdevices dir. I also had to edit and recompile my pl2303 driver, because my cable's VENDOR_ID and PRODUCT_ID didn't match with those registered in the driver module.

I'm using this way since 0.9.49 and it's still working with 1.0rc1

> To make it works, I've linked com1 to /dev/ttyUSB0 in dosdevices dir. I also
> had to edit and recompile my pl2303 driver, because my cable's VENDOR_ID and
> PRODUCT_ID didn't match with those registered in the driver module.


I have linked it too, but what do you mean with that?? What's wrong with VENDOR_ID and PRODUCT_ID?? The driver should recognize those from the cable, isn't it? And anyways, what does it have to do with wine? I haven't yet been able to make this thing work, so I'm quite sure it's correct to reopen the bug to try to solve this strange behaviour. It's not easy to recompile the driver for someone without great knowledge of Linux, and it gets worse when dealing with a precompiled & packaged one. Anyways, it seems an interesting comment, if that's the problem at all.

The devices with their respective kernel modules mentioned in bug 10699 also failed to work with this rc. The problem is still open.

Hi All,

Sorry to add to the list of woes! I am trying to set up BSS Audio's Soundweb designer on a new laptop (acer 5720z) without inbuilt serial port. On my old laptop (thinkpad 570e) the program runs fine under wine 0.9.25 on Debian Etch (current and up to date) using the inbuilt serial port with no more configuration needed than selecting com1 in the soundweb ui. Wine was installed in this case using apt-get.

On the new laptop - running Kubuntu 7.10 (also up to date and otherwise working perfectly) and Wine 0.9.46 the program launches as normal - and unless i symlink /dev/ttyUSB0 > dosdevices/com1 the program quits upon trying to connect to serial.
When i have the symlink in place and have verified the following:

usb 1-1: pl2303 converter now attached to ttyUSB0,

the program will search for devices but return no results. I should probably mention that the software has a serial connection to one of a network of devices and usually finds all the devices.

Before finding this bug - I tried compiling the 0.9.25 version on the new machine with the same results as I get with version 0.9.46

Attached is the result of WINEDEBUG='+comm' wine "C:\Program Files\BSS Audio\Soundweb\SoundWeb.exe"

It would appear that Bug 7526 is also appearing in the output to my terminal.(In reply to comment #3)
> Created an attachment (id=7016) [details]
> WINEDEBUG + comm output for 0.9.22
> 
> output repeats over and over, this is just a snapshot of the output during
> running

Also, I should say that I'm more than happy to try things out (i've already proved my backup systems work once today!) if I can be of any help in fixing this problem - please let me know.

Cheers

Ed

Hi Néstor,

I mention that because my cable wasn't recognized when loading the module. When attached, the device /dev/ttyUSB0 wasn't created, and I had to change the identifiers in source code and recompile it. 

I agree with you, this is not directly related to wine. But I thought this could be your problem too. 

Anyway, sorry for anything. I was just trying to help.

Hello again,

I've compiled the rc1 from source and successfully installed it (Kubuntu Gutsy - up to date) the usb serial adapter now works, very well too for a first test. 

I have a symlink from ~/.wine/dosdevices/com1 > /dev/ttyUSB0

And that's it  - it works! Many thanks to everyone who worked on this, I had a horrible feeling that the next show I was going to put on would be sullied by WinXP.

Thanks again,

Ed

Created attachment 13155 [details]
WINEDEBUG="+comm" wine TIM\ Web\ Movel.exe

I did the same steps and it does not work
attached log of
WINEDEBUG="+comm" wine TIM\ Web\ Movel.exe

Device not detected
my kernel is "2.6.17.8-slh-up-1"
and the linux device driver is "usbserial vendor=0x12d1 product=0x1003"
the device is Huawei E220 modem

kernel log (sorry, can't attach more than one file):
-----
May 18 11:08:40 kalango kernel: usbcore: registered new driver usbserial
May 18 11:08:40 kalango kernel: drivers/usb/serial/usb-serial.c: USB Serial support registered for generic
May 18 11:08:40 kalango kernel: usbserial_generic 4-1:1.0: generic converter detected
May 18 11:08:40 kalango kernel: usb 4-1: generic converter now attached to ttyUSB0
May 18 11:08:40 kalango kernel: usbserial_generic 4-1:1.1: generic converter detected
May 18 11:08:40 kalango kernel: usb 4-1: generic converter now attached to ttyUSB1
May 18 11:08:40 kalango kernel: usbserial_generic 4-1:1.2: generic converter detected
May 18 11:08:40 kalango kernel: usb 4-1: generic converter now attached to ttyUSB2
May 18 11:08:40 kalango kernel: usbcore: registered new driver usbserial_generic
-----

The regression is apparently fixed, the other problems will have to wait, deferring.

I'm going to further test it. Now I'm learning some serial kernel stuff things that might be useful for this. (I hope). As far as I see, there are many people who happily work with this driver, and some others that are having problems (like me). There must be some reason for it. Just one little thing: I have not been able to adjust serial port settings for this adapter. I don't know if it's normal, but I'm used to do it on standard ports. Should it be something related  to hardware, or maybe some poor implementation of the driver??

Created attachment 14449 [details]
debug output results of wine-1.1.0

:~> wine --version
wine-1.1.0

output of 

WINEDEBUG="+comm" wine application.exe &> file.op

Comment on attachment 14449 [details]
debug output results of wine-1.1.0


I confirm that the problem still exists with versions of wine => 0.9.23
including the 1.1.0 version.

Created attachment 14766 [details]
+comm log of Dynamic C 9.52

using a "0403:6001 Future Technology Devices International, Ltd FT232 USB-Serial (UART) IC" USB-Serial converter. Kernel modules are ftdi_sio and usbserial.

Without a symlink in dosdevices, I receive an error "could not setup serial port" whereas with the symlink, I receive "No rabbit processor detected". These are the same errors I would get in Windows if I chose a non-existent comm port (ex, Comm 8) and if I chose a valid port but didn't connect the device, respectively.

Attached is the output of "WINEPREFIX=/home/dynamicC/ WINEDEBUG='+comm' wine Dcrab_9.52.exe &> log.txt"

http://bugs.winehq.org/show_bug.cgi?id=8783#c14
still no go here with wine 1.0 and the new silabs driver (GPL'd but not yet in mainstream):
https://www.silabs.com/products/mcu/Pages/USBtoUARTBridgeVCPDrivers.aspx

trace:comm:dump_dcb bytesize=8 baudrate=115200 fParity=0 Parity=0 stopbits=1

trace:comm:wait_for_event device=0xd0 fd=0x00000017 mask=0x00000001 buffer=0x7e1ce9e0 event=0xd4 irq_info=0x16f66c
trace:comm:get_irq_info TIOCGICOUNT err Invalid argument
trace:comm:get_modem_status 0006 -> 
trace:comm:check_events mask 0x00000001
(period)

and strange behaviour:

warn:comm:set_baud_rate You (or a program acting at your behest) have specified
a non-standard baud rate 406250.

and no data transfered:

trace:comm:check_events old->rx          0x00000000 vs. new->rx          0x00000000
trace:comm:check_events old->tx          0x00000000 vs. new->tx          0x00000000
trace:comm:check_events old->frame       0x00000000 vs. new->frame       0x00000000

i will try to implement TIOCGICOUNT next days.

y
tom

(In reply to comment #38)
but read data transfer with minicom works.

(In reply to comment #39)

driver support is in now 

--- cp210x-3.0.0/cp210x/cp210x.c	2008-06-19 23:34:28.000000000 +0200
+++ cp210x-schorpp/cp210x/cp210x.c	2008-08-20 20:15:29.000000000 +0200
@@ -1982,6 +1982,13 @@
 			/* NOTREACHED */
 			break;
 
+                case TIOCGICOUNT:
+                        dbg ("%s (%d) TIOCGICOUNT RX=%d, TX=%d", __FUNCTION__,
+                             port->number, port_priv->icount.rx, port_priv->icount.tx);
+                        if (copy_to_user((void __user *)arg, &port_priv->icount, sizeof(port_priv->icount)))
+                                return -EFAULT;
+                        return 0;
+

but still 

trace:comm:get_irq_info TIOCGICOUNT err Invalid argument

so wine NOIMPL/bug. can one of the wine devs check it, pls? is it in wine devel head?

\WINEDEBUG=+comm wine microbrn.exe  &> file.op 

wine --version
wine-1.1.3

The name of the application I am trying to run has changed - it is just a newer version of the same thing.

The results are the same and since upgrading to a newer OS wine 0.9.22 no longer compiles so I'm really stuck now being unable to run two versions in parallel.

I've seen suggestions that the root cause might be due to a driver / module problem and not directly related to wine - how is that possible when it works with earlier versions?

Is there anything that can be done to resolve this or should the bug be abandoned?

Rich

Created attachment 15735 [details]
debug output from version 1.1.3

Hello Richard

TIOCGICOUNT is not the problem, it didn't work on wine 0.9.22 either. However the error path taken in dlls/ntdll/serial.c from 0.9.23 when TIOCGICOUNT fails is different, and the application takes a different code path and calls IOCTL_SERIAL_CLR_DTR, which it doesn't seem to do in 0.9.22. My patch should have fixed that, but you say it didn't help. Can you attach the output of:

WINEDEBUG='+comm' wine app.exe &> /tmp/log.txt

with my patch?

Also please post the output of:

strace -f -v wine app.exe &> /tmp/strace.txt

preferably both with a broken version, and with a working version of wine.

Damjan,

Your patch (applied manually to wine version 1.1.3 on gentoo AMD64; some of the logic had changed around the spot where your patch went, but I was able to still get it to work) fixed my issue with iGuidance 4 (a GPS navigation program) talking to my USB GPS receiver. It now works, instead of not being able to connect to the GPS receiver. I'm going to test it with wine 1.1.13 as well, since that just went to ~arch recently. Thanks!

-Rich


(In reply to comment #43)
> Hello Richard
> 
> TIOCGICOUNT is not the problem, it didn't work on wine 0.9.22 either. However
> the error path taken in dlls/ntdll/serial.c from 0.9.23 when TIOCGICOUNT fails
> is different, and the application takes a different code path and calls
> IOCTL_SERIAL_CLR_DTR, which it doesn't seem to do in 0.9.22. My patch should
> have fixed that, but you say it didn't help. Can you attach the output of:
> 
> WINEDEBUG='+comm' wine app.exe &> /tmp/log.txt
> 
> with my patch?
> 
> Also please post the output of:
> 
> strace -f -v wine app.exe &> /tmp/strace.txt
> 
> preferably both with a broken version, and with a working version of wine.
> 

Are you saying my patch didn't work before, but works with 1.1.3 (or is it 1.1.13)?

Please attach the changed patch for "1.1.3" just in case.

Created attachment 19188 [details]
Damjan Jovanovic's patch rolled against 1.1.14

Damjan Jovanovic's patch, originally from 7/6/2007, modified to reflect updates to serial.c, rolled against 1.1.14. Fixed my issues with USB/Serial for GPS. Also applies to wine 1.1.3, but the location marker is as follows for 1.1.3:

@@ -1047,13 +1047,8 @@ static NTSTATUS wait_on(HANDLE hDevice,

I never tried your patch from 7/6/2007 until yesterday. It wouldn't apply anymore (because the logic in the function had changed, along with the line numbers), so I applied the changes by hand, using my best judgment. However, the patch I just submitted, rolled against 1.1.14, fixed my problem (see comments on the submitted patch for details on 1.1.3, which works). Thanks again!

-Rich

(In reply to comment #45)
> Are you saying my patch didn't work before, but works with 1.1.3 (or is it
> 1.1.13)?
> 
> Please attach the changed patch for "1.1.3" just in case.
> 

Damjan Jonaovic's patch applied to 1.1.33 fixed the bug for the k128 kitsrus pic programmer with microbrn v25ep. Although, it seems all the patch did was ignore the error, attached is the log for +comm.

Created attachment 24950 [details]
+comm trace for microbrn with DJ's patch applied to 1.1.33

(In reply to comment #48)
> Damjan Jonaovic's patch applied to 1.1.33 fixed the bug for the k128 kitsrus
> pic programmer with microbrn v25ep. Although, it seems all the patch did was
> ignore the error, attached is the log for +comm.

Spoke to soon, the patch allows the device to be recognized by the program but it can't actually communicate in any meaningful way.

Hi,

I have tried the following program in wine 1.1.37 (which works in Windows Vista natively) and it is not showing the list of com ports as expected:
http://www.4dsystems.com.au/downloads/Software/4DGL-Workshop2-IDE/4DGL_Workshop2_IDE.zip

I tried making the symbolic link in ~/.wine/dosdevices/com1 -> /dev/ttyUSB0

I also tried changing ~/.wine/system.reg to have the following lines (as per http://wiki.jswindle.com/index.php/Wine_Registry#The_Registry_Settings):
[Hardware\\Devicemap\\Serialcomm] 1231984861 @=""
"Serial0"="COM1"

The last thing allowed me to see the COM port, but when I try to download from 4DGL workshop to my display module, it gives the following error:
Port: COM1
No hardware detected.

What am I doing wrong, or is this a wine bug?

Thanks,
Anna.

Windows has some more registry key for com ports. Maybe the program look for them too e.g.:
HKLM\Hardware\Description\System\*\*\SerialController\*
HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Port
HKLM\SYSTEM\CurrentControlSet\Enum\ACPI\PNP0501\1\Device Parameters

(In reply to comment #52)
> Windows has some more registry key for com ports. Maybe the program look for
> them too e.g.:
> HKLM\Hardware\Description\System\*\*\SerialController\*
> HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Port
> HKLM\SYSTEM\CurrentControlSet\Enum\ACPI\PNP0501\1\Device Parameters

Hi,
Thanks, I tried those but they didn't work. FYI after I ran the program, it changed the entries to the following:

[HKLMHardwareDescriptionSystem**SerialController*] 1240428288

[HKLMSOFTWAREMicrosoftWindows NTCurrentVersionPort] 1240428288

[HKLMSYSTEMCurrentControlSetEnumACPIPNP0501\1Device Parameters] 1240428288

What other lines do I need to try?

Regards,
Anna.

Hi,

I use a program (LG Contents Bank) to get pics from my Cell Phone. It uses a virtual serial port (those USB -> Serial converters). So, I create a link named "com1" pointing to ttyUSB0 in dosdevices.

Just like you said, I'm not able to list the serial ports on the app running over wine (An old issue, I guess). But, if edit the app's .ini file, and set manually the desired COM port, it works properly.

This program you're trying to use (4DGL-WorkShop2) has also a config file (4DGL.ini, in Dep directory) and a parameter with the COM port to use ( Section [plataform], parameter port ). Maybe if you set it manually, maybe it works.

Other thing: I know this may look obvious, but your create device is really ttyUSB0 right? You can confirm this pluging it and typing "dmesg" in terminal

Best regards,
pstglia

(In reply to comment #51)
> Hi,
> 
> I have tried the following program in wine 1.1.37 (which works in Windows Vista
> natively) and it is not showing the list of com ports as expected:
> http://www.4dsystems.com.au/downloads/Software/4DGL-Workshop2-IDE/4DGL_Workshop2_IDE.zip
> 
> I tried making the symbolic link in ~/.wine/dosdevices/com1 -> /dev/ttyUSB0
> 
> I also tried changing ~/.wine/system.reg to have the following lines (as per
> http://wiki.jswindle.com/index.php/Wine_Registry#The_Registry_Settings):
> [Hardware\\Devicemap\\Serialcomm] 1231984861 @=""
> "Serial0"="COM1"
> 
> The last thing allowed me to see the COM port, but when I try to download from
> 4DGL workshop to my display module, it gives the following error:
> Port: COM1
> No hardware detected.
> 
> What am I doing wrong, or is this a wine bug?
> 
> Thanks,
> Anna.

Hi,

Thanks for your response. I checked the 4DGL.ini file and it already contains COM1 as the port.

Also yes, I have confirmed that /dev/ttyUSB0 is the correct name for the device. I have tested it using minicom and I can see messages being received from the device.

Regards,
Anna.

(In reply to comment #54)
> Hi,
> 
> I use a program (LG Contents Bank) to get pics from my Cell Phone. It uses a
> virtual serial port (those USB -> Serial converters). So, I create a link named
> "com1" pointing to ttyUSB0 in dosdevices.
> 
> Just like you said, I'm not able to list the serial ports on the app running
> over wine (An old issue, I guess). But, if edit the app's .ini file, and set
> manually the desired COM port, it works properly.
> 
> This program you're trying to use (4DGL-WorkShop2) has also a config file
> (4DGL.ini, in Dep directory) and a parameter with the COM port to use ( Section
> [plataform], parameter port ). Maybe if you set it manually, maybe it works.
> 
> Other thing: I know this may look obvious, but your create device is really
> ttyUSB0 right? You can confirm this pluging it and typing "dmesg" in terminal
> 
> Best regards,
> pstglia
> 
> (In reply to comment #51)
> > Hi,
> > 
> > I have tried the following program in wine 1.1.37 (which works in Windows Vista
> > natively) and it is not showing the list of com ports as expected:
> > http://www.4dsystems.com.au/downloads/Software/4DGL-Workshop2-IDE/4DGL_Workshop2_IDE.zip
> > 
> > I tried making the symbolic link in ~/.wine/dosdevices/com1 -> /dev/ttyUSB0
> > 
> > I also tried changing ~/.wine/system.reg to have the following lines (as per
> > http://wiki.jswindle.com/index.php/Wine_Registry#The_Registry_Settings):
> > [Hardware\\Devicemap\\Serialcomm] 1231984861 @=""
> > "Serial0"="COM1"
> > 
> > The last thing allowed me to see the COM port, but when I try to download from
> > 4DGL workshop to my display module, it gives the following error:
> > Port: COM1
> > No hardware detected.
> > 
> > What am I doing wrong, or is this a wine bug?
> > 
> > Thanks,
> > Anna.

Created attachment 32821 [details]
WINEDEBUG="+comm" output

Run as follows: WINEDEBUG="+comm" wine GpsDump458.exe &> GpsDump458.debug.txt

I am trying to use GpsDump 4.58 to obtain flights from a Flymaster B1Nav GPS device. I have linked /dev/ttyUSB0 to com1 in dosdevices, and the application appears to be able to communicate to some extent with the device. I can see the serial number and firmware version. However, when it tries to download the flight list - which involves more intensive communication, wine gives a communication error (E503 COM port read error). Attached is the debug output.

Running wine 1.2-0ubuntu6~lucid5.

I also tried the latest version of wine (1.3.11) to no avail.

More info on the Flymaster B1Nav:

Manufacturer: FTDI, Description: FT232R USB UART

Hi, all.

Serial Port related problem is hard to debug since it requires extra hardware that wine developers may not have.
I just find a much simpler way of testing Serial Port stuff; connecting the RXD and TXD pins together to build a loopback Serial Port.
Check the following document for details:
http://www.quatech.com/pdf/hyperterminal.pdf

For the software part, I'd recommend open source Tera Term instead of proprietary Hyper Terminal.
http://ttssh2.sourceforge.jp/

I've successfully seen character echoing back on Tera Term, Windows 7 with loopback USB Serial Port.

I stumbled across this bug myself, though the solution turned out to be pretty simple for me:

ln -s /dev/ttyUSB0 ~/.wine/dosdevices/com1

sudo adduser {username} dialout

logout, login and now I can talk to my 2-way radio USB serial port.

Hope this helps someone.

Paul

This bug most likely is already fixed.

Closing bugs fixed in 1.7.12.