Bug 146663

Created attachment 112066 [details]
Candidate #1

This is a backport from 2.6.11 which ought to address both TX not working
and the break control.

Hi Pete,

I have tried this patch on standard 2.4.21-27.0.2.EL sources, but the
compilation process ended up with:

--------------------------------------------------------------------------------
pl2303.c: In function `pl2303_set_termios':
pl2303.c:234: warning: `priv' might be used uninitialized in this function
pl2303.c: In function `pl2303_open':
pl2303.c:426: `priv' undeclared (first use in this function)
pl2303.c:426: (Each undeclared identifier is reported only once
pl2303.c:426: for each function it appears in.)
make[3]: *** [pl2303.o] Error 1
make[3]: Leaving directory
`/usr/src/linux-2.4.21-27.0.2.EL.pl2303/drivers/usb/serial'
make[2]: *** [_modsubdir_serial] Error 2
make[2]: Leaving directory `/usr/src/linux-2.4.21-27.0.2.EL.pl2303/drivers/usb'
make[1]: *** [_modsubdir_usb] Error 2
make[1]: Leaving directory `/usr/src/linux-2.4.21-27.0.2.EL.pl2303/drivers'
make: *** [_mod_drivers] Error 2
--------------------------------------------------------------------------------

Regards,
            Juanjo

Created attachment 112083 [details]
Additional patch to Candidate #1

Hi Pete,

I have solved the compilation errors with this small patch. With the resulting
module TX and break seems to work fine.

Please tell me if I have to open a new bug, but I have observed that
disconnecting the USB-Serial converter while it is in use (by minicom, for
example) leads to a kernel panic when new data is transferred (pressing a key
in minicom):

--------------------------------------------------------------------------------

devel.si.uji.es login:
Red Hat Enterprise Linux AS release 3 (Taroon Update 4)
Kernel 2.4.21-27.0.2.ELsmp on an i686

devel.si.uji.es login: Unable to handle kernel NULL pointer dereference at
virtual address 00000060
 printing eip:
f897aa17
*pde = 36fee001
*pte = 3b1da067
Oops: 0002
lp parport ide-cd cdrom autofs4 smbus audit acenic e100 ipt_REJECT ipt_LOG
ipt_state ip_conntrack iptable_filter ip_tables floppy sg microcode pl2303
usbseria
CPU:	0
EIP:	0060:[<f897aa17>]    Tainted: P
EFLAGS: 00010286

EIP is at serial_write [usbserial] 0xa7 (2.4.21-27.0.2.ELsmp/i686)
eax: 00000000	ebx: 00000060	ecx: 00000060	edx: 00000000
esi: 00000000	edi: 00000001	ebp: 00000001	esp: f732beb0
ds: 0068   es: 0068   ss: 0068
Process minicom (pid: 6194, stackpage=f732b000)
Stack: 00000000 c0383180 00000292 f732a000 c043b280 00000001 bfff9e40 f732a000
       f6fa5000 c01b6206 f6fa5000 00000001 bfff9e40 00000001 f6fa5974 00000000
       00000000 f732a000 00000000 00000000 00000000 c3573bfc f732a000 c3573980
Call Trace:   [<c01b6206>] write_chan [kernel] 0x1e6 (0xf732bed4)
[<c01b44ad>] do_tty_write [kernel] 0x14d (0xf732bf40)
[<c01b186f>] tty_write [kernel] 0x14f (0xf732bf6c)
[<c01b6020>] write_chan [kernel] 0x0 (0xf732bf70)
[<c0164257>] sys_write [kernel] 0x97 (0xf732bf94)

Code: f0 ff 4e 60 0f 88 dd 25 00 00 80 7e 35 00 74 29 f0 ff 46 60

Kernel panic: Fatal exception
--------------------------------------------------------------------------------


Note that this problem was also observed in setup (3). Also note that the
kernel is Tainted because we use Fujitsu-siemens serverview agents, I can
repeat the test if needed.

Regards,
	   Juanjo

The oops on unplug is a manifestation of bug 132994. Let's not consider
it in the context of this bug.

Created attachment 112106 [details]
Candidate #2 - corrected with customer feedback

Thanks a lot for testing. Most people just wait until I push a test
through the build system, which takes time. So, this is welcome.

A test kernel 2.4.21-31.EL.usbserial.2 is available at
 ftp://people.redhat.com/zaitcev/rhel3usb/

I also have problem with this device at RHES4. The first time I try to send 
data with it, it is no problem. But the second time I send, the port opens and 
no more happens. I have to unplug it and start from the beginning.

Magnus, please file a bug. Don't pollute unrelated bugs.

Juanjo, did the usbserial.X work for you?

Pete, I have tested the tx and BREAK issue on a server running
kernel-smp-2.4.21-31.EL.usbserial.5 and it works fine.

Note that we have upgraded to RHEL4 the server where the Aten UC-232A USB-serial
converter is attached, so we aren't using this kernel (2.4.21-31.EL.usbserial.5)
in our production server.

A fix for this problem has just been committed to the RHEL3 U7
patch pool this evening (in kernel version 2.4.21-39.EL).

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on the solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHSA-2006-0144.html